Jakob Foerster an accredited Machine Learning Research Scientist who has been at the forefront of research on Multi-Agent Learning speaks with interviewer Kegan Strawn.

Dr. Foerster explains why incorporating uncertainty into multi-agent interactions is essential to creating robust algorithms that can operate not only in games but in real-world applications.

Jakob Foerster
Jakob Foerster is an Associate Professor at the University of Oxford. His papers have gained prestigious awards at top machine learning conferences (ICML, AAAI) and have helped push deep multi-agent reinforcement learning to the forefront of AI research.

Jakob previously worked at Facebook AI Research and received his Ph.D. from the University of Oxford under the supervision of Shimon Whiteson. During his Ph.D., Jakob also interned at Google Brain, OpenAI, and DeepMind.

Jakob’s research interests span Deep Multi-Agent Reinforcement Learning, Human-AI Coordination, Emergent Communication, Search, Planning, and Game Theory.

Links

• Download mp3 (36.4 MB)
• Subscribe to Robohub using iTunes, RSS, or Spotify
• Support us on Patreon

As AI surpasses human abilities in Go and poker – two decades after Deep Blue trounced chess grandmaster Garry Kasparov – it is seeping into our lives in ever more profound ways. It affects the way we search the web, receive medical advice and whether we receive finance from our banks.

Researchers have devised a new way to monitor sleep stages without sensors attached to the body. Their device uses an advanced artificial intelligence algorithm to analyze the radio signals around the person and translate those measurements into sleep stages: light, deep, or rapid eye movement (REM). Image: Christine Daniloff/MIT

More than 50 million Americans suffer from sleep disorders, and diseases including Parkinson’s and Alzheimer’s can also disrupt sleep. Diagnosing and monitoring these conditions usually requires attaching electrodes and a variety of other sensors to patients, which can further disrupt their sleep.

To make it easier to diagnose and study sleep problems, researchers at MIT and Massachusetts General Hospital have devised a new way to monitor sleep stages without sensors attached to the body. Their device uses an advanced artificial intelligence algorithm to analyze the radio signals around the person and translate those measurements into sleep stages: light, deep, or rapid eye movement (REM).

“Imagine if your Wi-Fi router knows when you are dreaming, and can monitor whether you are having enough deep sleep, which is necessary for memory consolidation,” says Dina Katabi, the Andrew and Erna Viterbi Professor of Electrical Engineering and Computer Science, who led the study. “Our vision is developing health sensors that will disappear into the background and capture physiological signals and important health metrics, without asking the user to change her behavior in any way.”

Katabi worked on the study with Matt Bianchi, chief of the division of sleep medicine at MGH, and Tommi Jaakkola, the Thomas Siebel Professor of Electrical Engineering and Computer Science and a member of the Institute for Data, Systems, and Society at MIT. Mingmin Zhao, an MIT graduate student, is the paper’s first author, and Shichao Yue, another MIT graduate student, is also a co-author.

The researchers will present their new sensor at the International Conference on Machine Learning on Aug. 9.

Remote sensing

Katabi and members of her group in MIT’s Computer Science and Artificial Intelligence Laboratory have previously developed radio-based sensors that enable them to remotely measure vital signs and behaviors that can be indicators of health. These sensors consist of a wireless device, about the size of a laptop computer, that emits low-power radio frequency (RF) signals. As the radio waves reflect off of the body, any slight movement of the body alters the frequency of the reflected waves. Analyzing those waves can reveal vital signs such as pulse and breathing rate.

“It’s a smart Wi-Fi-like box that sits in the home and analyzes these reflections and discovers all of these changes in the body, through a signature that the body leaves on the RF signal,” Katabi says.

Katabi and her students have also used this approach to create a sensor called WiGait that can measure walking speed using wireless signals, which could help doctors predict cognitive decline, falls, certain cardiac or pulmonary diseases, or other health problems.

After developing those sensors, Katabi thought that a similar approach could also be useful for monitoring sleep, which is currently done while patients spend the night in a sleep lab hooked up to monitors such as electroencephalography (EEG) machines.

“The opportunity is very big because we don’t understand sleep well, and a high fraction of the population has sleep problems,” says Zhao. “We have this technology that, if we can make it work, can move us from a world where we do sleep studies once every few months in the sleep lab to continuous sleep studies in the home.”

To achieve that, the researchers had to come up with a way to translate their measurements of pulse, breathing rate, and movement into sleep stages. Recent advances in artificial intelligence have made it possible to train computer algorithms known as deep neural networks to extract and analyze information from complex datasets, such as the radio signals obtained from the researchers’ sensor. However, these signals have a great deal of information that is irrelevant to sleep and can be confusing to existing algorithms. The MIT researchers had to come up with a new AI algorithm based on deep neural networks, which eliminates the irrelevant information.

“The surrounding conditions introduce a lot of unwanted variation in what you measure. The novelty lies in preserving the sleep signal while removing the rest,” says Jaakkola. Their algorithm can be used in different locations and with different people, without any calibration.

Using this approach in tests of 25 healthy volunteers, the researchers found that their technique was about 80 percent accurate, which is comparable to the accuracy of ratings determined by sleep specialists based on EEG measurements.

“Our device allows you not only to remove all of these sensors that you put on the person, and make it a much better experience that can be done at home, it also makes the job of the doctor and the sleep technologist much easier,” Katabi says. “They don’t have to go through the data and manually label it.”

Sleep deficiencies

Other researchers have tried to use radio signals to monitor sleep, but these systems are accurate only 65 percent of the time and mainly determine whether a person is awake or asleep, not what sleep stage they are in. Katabi and her colleagues were able to improve on that by training their algorithm to ignore wireless signals that bounce off of other objects in the room and include only data reflected from the sleeping person.

The researchers now plan to use this technology to study how Parkinson’s disease affects sleep.

“When you think about Parkinson’s, you think about it as a movement disorder, but the disease is also associated with very complex sleep deficiencies, which are not very well understood,” Katabi says.

The sensor could also be used to learn more about sleep changes produced by Alzheimer’s disease, as well as sleep disorders such as insomnia and sleep apnea. It may also be useful for studying epileptic seizures that happen during sleep, which are usually difficult to detect.

From the Australian government’s new “data-driven profiling” trial for drug testing welfare recipients, to US law enforcement’s use of facial recognition technology and the deployment of proprietary software in sentencing in many US courts … almost by stealth and with remarkably little outcry, technology is transforming the way we are policed, categorized as citizens and, perhaps one day soon, governed.

We are only in the earliest stages of so-called algorithmic regulation – intelligent machines deploying big data, machine learning and artificial intelligence (AI) to regulate human behaviour and enforce laws – but it already has profound implications for the relationship between private citizens and the state.

Furthermore, the rise of such technologies is occurring at precisely the moment when faith in governments across much of the Western world has plummeted to an all-time low. Voters across much of the developed world increasingly perceive establishment politicians and those who surround them to be out-of touch bubble-dwellers and are registering their discontent at the ballot box.

---

Setting some ground rules

Pic2Recipe, an artificial intelligence system developed at MIT, can take a photo of an entree and suggest a similar recipe to it. Photo: Jason Dorfman/MIT CSAIL

There are few things social media users love more than flooding their feeds with photos of food. Yet we seldom use these images for much more than a quick scroll on our cellphones. Researchers from MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) believe that analyzing photos like these could help us learn recipes and better understand people’s eating habits.

In a new paper with the Qatar Computing Research Institute (QCRI), the team trained an artificial intelligence system called Pic2Recipe to look at a photo of food and be able to predict the ingredients and suggest similar recipes.

“In computer vision, food is mostly neglected because we don’t have the large-scale datasets needed to make predictions,” says Yusuf Aytar, an MIT postdoc who co-wrote a paper about the system with MIT Professor Antonio Torralba. “But seemingly useless photos on social media can actually provide valuable insight into health habits and dietary preferences.”

How it works

The web has spurred a huge growth of research in the area of classifying food data, but the majority of it has used much smaller datasets, which often leads to major gaps in labeling foods.

In 2014 Swiss researchers created the “Food-101” dataset and used it to develop an algorithm that could recognize images of food with 50 percent accuracy. Future iterations only improved accuracy to about 80 percent, suggesting that the size of the dataset may be a limiting factor.

Even the larger datasets have often been somewhat limited in how well they generalize across populations. A database from the City University in Hong Kong has over 110,000 images and 65,000 recipes, each with ingredient lists and instructions, but only contains Chinese cuisine.

The CSAIL team’s project aims to build off of this work but dramatically expand in scope. Researchers combed websites like All Recipes and Food.com to develop “Recipe1M,” a database of over 1 million recipes that were annotated with information about the ingredients in a wide range of dishes. They then used that data to train a neural network to find patterns and make connections between the food images and the corresponding ingredients and recipes.

Given a photo of a food item, Pic2Recipe could identify ingredients like flour, eggs, and butter, and then suggest several recipes that it determined to be similar to images from the database. (The team has an online demo where people can upload their own food photos to test it out.)

“You can imagine people using this to track their daily nutrition, or to photograph their meal at a restaurant and know what’s needed to cook it at home later,” says Christoph Trattner, an assistant professor at MODUL University Vienna in the New Media Technology Department who was not involved in the paper. “The team’s approach works at a similar level to human judgement, which is remarkable.”

The system did particularly well with desserts like cookies or muffins, since that was a main theme in the database. However, it had difficulty determining ingredients for more ambiguous foods, like sushi rolls and smoothies.

It was also often stumped when there were similar recipes for the same dishes. For example, there are dozens of ways to make lasagna, so the team needed to make sure that system wouldn’t “penalize” recipes that are similar when trying to separate those that are different. (One way to solve this was by seeing if the ingredients in each are generally similar before comparing the recipes themselves).

In the future, the team hopes to be able to improve the system so that it can understand food in even more detail. This could mean being able to infer how a food is prepared (i.e. stewed versus diced) or distinguish different variations of foods, like mushrooms or onions.

The researchers are also interested in potentially developing the system into a “dinner aide” that could figure out what to cook given a dietary preference and a list of items in the fridge.

“This could potentially help people figure out what’s in their food when they don’t have explicit nutritional information,” says Hynes. “For example, if you know what ingredients went into a dish but not the amount, you can take a photo, enter the ingredients, and run the model to find a similar recipe with known quantities, and then use that information to approximate your own meal.”

The project was funded, in part, by QCRI, as well as the European Regional Development Fund (ERDF) and the Spanish Ministry of Economy, Industry, and Competitiveness.

Image: MIT News

Laparoscopy is a surgical technique in which a fiber-optic camera is inserted into a patient’s abdominal cavity to provide a video feed that guides the surgeon through a minimally invasive procedure. Laparoscopic surgeries can take hours, and the video generated by the camera — the laparoscope — is often recorded. Those recordings contain a wealth of information that could be useful for training both medical providers and computer systems that would aid with surgery, but because reviewing them is so time consuming, they mostly sit idle.

Researchers at MIT and Massachusetts General Hospital hope to change that, with a new system that can efficiently search through hundreds of hours of video for events and visual features that correspond to a few training examples.

In work they presented at the International Conference on Robotics and Automation this month, the researchers trained their system to recognize different stages of an operation, such as biopsy, tissue removal, stapling, and wound cleansing.

But the system could be applied to any analytical question that doctors deem worthwhile. It could, for instance, be trained to predict when particular medical instruments — such as additional staple cartridges — should be prepared for the surgeon’s use, or it could sound an alert if a surgeon encounters rare, aberrant anatomy.

“Surgeons are thrilled by all the features that our work enables,” says Daniela Rus, an Andrew and Erna Viterbi Professor of Electrical Engineering and Computer Science and senior author on the paper. “They are thrilled to have the surgical tapes automatically segmented and indexed, because now those tapes can be used for training. If we want to learn about phase two of a surgery, we know exactly where to go to look for that segment. We don’t have to watch every minute before that. The other thing that is extraordinarily exciting to the surgeons is that in the future, we should be able to monitor the progression of the operation in real-time.”

Joining Rus on the paper are first author Mikhail Volkov, who was a postdoc in Rus’ group when the work was done and is now a quantitative analyst at SMBC Nikko Securities in Tokyo; Guy Rosman, another postdoc in Rus’ group; and Daniel Hashimoto and Ozanan Meireles of Massachusetts General Hospital (MGH).

Representative frames

The new paper builds on previous work from Rus’ group on “coresets,” or subsets of much larger data sets that preserve their salient statistical characteristics. In the past, Rus’ group has used coresets to perform tasks such as deducing the topics of Wikipedia articles or recording the routes traversed by GPS-connected cars.

In this case, the coreset consists of a couple hundred or so short segments of video — just a few frames each. Each segment is selected because it offers a good approximation of the dozens or even hundreds of frames surrounding it. The coreset thus winnows a video file down to only about one-tenth its initial size, while still preserving most of its vital information.

For this research, MGH surgeons identified seven distinct stages in a procedure for removing part of the stomach, and the researchers tagged the beginnings of each stage in eight laparoscopic videos. Those videos were used to train a machine-learning system, which was in turn applied to the coresets of four laparoscopic videos it hadn’t previously seen. For each short video snippet in the coresets, the system was able to assign it to the correct stage of surgery with 93 percent accuracy.

“We wanted to see how this system works for relatively small training sets,” Rosman explains. “If you’re in a specific hospital, and you’re interested in a specific surgery type, or even more important, a specific variant of a surgery — all the surgeries where this or that happened — you may not have a lot of examples.”

Selection criteria

The general procedure that the researchers used to extract the coresets is one they’ve previously described, but coreset selection always hinges on specific properties of the data it’s being applied to. The data included in the coreset — here, frames of video — must approximate the data being left out, and the degree of approximation is measured differently for different types of data.

Machine learning can be thought of as a problem of approximation, however. In this case, the system had to learn to identify similarities between frames of video in separate laparoscopic feeds that denoted the same phases of a surgical procedure. The metric of similarity that it arrived at also served to assess the similarity of video frames that were included in the coreset, to those that were omitted.

“Interventional medicine — surgery in particular — really comes down to human performance in many ways,” says Gregory Hager, a professor of computer science at Johns Hopkins University who investigates medical applications of computer and robotic technologies. “As in many other areas of human endeavor, like sports, the quality of the human performance determines the quality of the outcome that you achieve, but we don’t know a lot about, if you will, the analytics of what creates a good surgeon. Work like what Daniela is doing and our work really goes to the question of: Can we start to quantify what the process in surgery is, and then within that process, can we develop measures where we can relate human performance to the quality of care that a patient receives?”

“Right now, efficiency” — of the kind provided by coresets — “is probably not that important, because we’re dealing with small numbers of these things,” Hager adds. “But you could imagine that, if you started to record every surgery that’s performed — we’re talking tens of millions of procedures in the U.S. alone — now it starts to be interesting to think about efficiency.”

The world’s brightest minds in Artificial Intelligence (AI) and humanitarian action will meet with industry leaders and academia at the AI for Good Global Summit, 7-9 June 2017, to discuss how AI will assist global efforts to address poverty, hunger, education, healthcare and the protection of our environment. The event will in parallel explore means to ensure the safe, ethical development of AI, protecting against unintended consequences of advances in AI.

View the live webcast at: http://bit.ly/AI-for-Good-Webcast.

The event is co-organized by ITU and the XPRIZE Foundation, in partnership with 20 other United Nations (UN) agencies, and with the participation of more than 70 leading companies and academic and research institutes.

“Artificial Intelligence has the potential to accelerate progress towards a dignified life, in peace and prosperity, for all people,” said UN Secretary-General António Guterres. “The time has arrived for all of us – governments, industry and civil society – to consider how AI will affect our future. The AI for Good Global Summit represents the beginnings of our efforts to ensure that AI charts a course that will benefit all of humanity.”

The AI for Good Global Summit will emphasize AI’s potential to contribute to the pursuit of the UN Sustainable Development Goals.

Opening sessions will share expert insight into the state of play in AI, with leading minds in AI giving voice to their greatest ambitions in driving AI towards social good. ‘Breakthrough’ sessions will propose strategies for the development of AI applications and systems able to promote sustainable living, reduce poverty and deliver citizen-centric public services.

“Today, we’ve gathered here to discuss how far AI can go, how much it will improve our lives, and how we can all work together to make it a force for good,” said ITU Secretary-General Houlin Zhao. “This event will assist us in determining how the UN, ITU and other UN Agencies can work together with industry and the academic community to promote AI innovation and create a good environment for the development of artificial intelligence.”

“The AI for Good Global Summit has assembled an impressive, diverse ecosystem of thought leaders who recognize the opportunity to use AI to solve some of the world’s grandest challenges,” said Marcus Shingles, CEO of the XPRIZE Foundation. “We look forward to this Summit providing a unique opportunity for international dialogue and collaboration that will ideally start to pave the path forward for a new future of problem solvers working with XPRIZE and beyond.”

The AI for Good Global Summit will be broadcast globally as well as captioned to ensure accessibility.

You can view the live webcast at: http://bit.ly/AI-for-Good-Webcast.

For more information about this event, please visit: AI for Good Global Summit web page.

In episode two of season three Neil takes us through the basics on dropout, we chat about the definition of inference (It’s more about context than you think!) and hear an interview with Jennifer Chayes of Microsoft.

If you liked this article, you may also want to read:

• Talking Machines: Restricted Boltzmann Machines, with Eric Lander
• Talking Machines: Automatic Translation and t-SNE, with Hal Daume
• Talking Machines: Generative art and Hamiltonian Monte Carlo, with Doug Eck
• Talking Machines: Machine learning and the Flint water crisis, with Jake Abernethy
• Talking Machines: Gaussian processes and OpenAI, with IIya Sutskever

See all the latest robotics news on Robohub, or sign up for our weekly newsletter.

In 1985, a twenty-two year old Garry Kasparov became the youngest World Chess Champion. Twelve years later, he was defeated by the only player capable of challenging the grandmaster, IBM’s Deep Blue. That same year (1997), RoboCup was formed to take on the world’s most popular game, soccer, with robots. Twenty years later, we are on the threshold of the accomplishing the biggest feat in machine intelligence, a team of fully autonomous humanoids beating human players at FIFA World Cup soccer.

Many of the advances that have led to the influx of modern autonomous vehicles and machine intelligence are the result of decades of competitions. While Deep Blue and AlphaGo have beat the world’s best players at board games, soccer requires real-world complexities (see chart) in order to best humans on the field. This requires RoboCup teams to combine a number of mechatronic technologies within a humanoid device, such as real-time sensor fusion, reactive behavior, strategy acquisition, deep learning, real-time planning, multi-agent systems, context recognition, vision, strategic decision-making, motor control, and intelligent robot control.

Professor Daniel Lee of University of Pennsylvania’s GRASP Lab described the RoboCup challenges best, “Why is it that we have machines that can beat us in chess or Jeopardy but we can beat them in soccer? What makes it so difficult to embody intelligence into the physical world?” Lee explains, “It’s not just the soccer domain. It’s really thinking about artificial intelligence, robotics, and what they can do in a more general context.”

RoboCup has become so important that the challenge of soccer has now expanded into new leagues that focus on many commercial endeavors from social robotics, to search & rescue, to industrial applications. These leagues have a number of subcategories of competition with varying degrees of difficulty. In less than two months, international teams will convene in Nagoya, Japan for the twenty-first games. As a preview of what to expect, let’s review some of last year’s winners. And just maybe it could give us a peek at the future of automation.

RoboCup Soccer

While Iran’s human soccer team is 28th in the world, their robot counterparts (Baset Pazhuh Tehran) won 1st place in the AdultSize Humanoid competition. Baset’s secret sauce is its proprietary algorithms for motion control, perception, and path planning. According to Baset’s team description paper, the key was building a “fast and stable walk engine” based upon the success of past competitions. This walk engine was able to combine “all actuators’ data in each joint, and changing the inverse and forward kinematics” to “avoid external forces affecting robot’s stability, this feature plays an important role to keep the robot standing when colliding to the other robots or obstacles.” Another big factor was their goalkeeper that used a stereo vision sensor to detect incoming plays and win the competition by having “better percepts of goal poles, obstacles, and the opponent’s goalkeeper. To locate each object in real self-coordinating system.” The team is part of a larger Iranian corporation, Baset, that could deploy this perception in the field. Baset’s oil and gas clients could benefit from better localization techniques and object recognition for pipeline inspections and autonomous work vehicles. If by 2050 RoboCup’s humanoids will be capable of playing humans in soccer, one has to wonder if Baset’s mechanical players will spend their offseason working in the Arabian peninsula?

RoboCup Rescue 

In 2001 the RoboCup organization added simulated rescue to the course challenge, paving the way for many life-saving innovations already being embraced by first responders. The course starts with a simulated earthquake environment whereby the robot performs a search and rescue mission lasting 20 minutes. The skills are graded by overcoming a number of obstacles that are designed to assess the robot’s autonomous operation, mobility, and object manipulation. Points are given by the number of victims found by the robot, details about the victims, and the quality of the area mapped. In 2016, students from the King Mongkut’s University of Technology North Bangkok won first place with their Invigorating Robot Activity Project (or iRAP).

Similar to Baset, iRap’s success is largely based upon solving problems from previous contests where they placed consistently in the top tier. The team had a total of four robots: one autonomous robot, two tele-operative robots, and one aerial drone. Each of the robots had multiple sensors related to providing critical data, such as CO2 levels, temperature, positioning, 2D mapping, images, and two-way communications. iRap’s devices managed to navigate with remarkable ease the test environment’s rough surfaces, hard terrains, rolling floor, stairs, and inclined floor. The most impressive performer was the caged quadcopter using enhanced sensors to localize itself within an outdoor search perimeter. According to the team’s description paper, “we have developed the autonomously outdoor robot that is the aerial robot. It can fly and localize itself by GPS sensor. Besides, the essential sensors for searching the victim.” It is interesting to note that the Thai team’s design was remarkably similar to Flyability’s Gimball that won first place in the UAE’s 2015 Drones for Good Competition. Like the RoboCup winner, the Gimball was designed specifically for search & rescue missions using a lightweight carbon fiber cage.

As RoboCup contestants push the envelope of navigation mapping technologies, it is quite possible that the 2017 fleet could develop subterranean devices that could actively find victims within minutes of being buried by the earth.

RoboCup @Home

The home, like soccer, is one of the most chaotic conditions for robots to operate successfully. It is also one of the biggest areas of interest for consumers. Last year, RoboCup @Home celebrated its 10th anniversary by bestowing the top accolade to Team-Bielefeld (ToBI) of Germany. ToBi built a humanoid-like robot that was capable of learning new skills through natural language within unknown environments. According to the team’s paper, “the challenge is two-fold. On the one hand, we need to understand the communicative cues of humans and how they interpret robotic behavior. On the other hand, we need to provide technology that is able to perceive the environment, detect and recognize humans, navigate in changing environments, localize and manipulate objects, initiate and understand a spoken dialogue and analyse the different scenes to gain a better understanding of the surrounding.” In order to achieve these ambitious objectives the team created a Cognitive Interaction Toolkit (CITK) to support an “aggregation of required system artifacts, an automated software build and deployment, as well as an automated testing environment.”

Infused with its proprietary software the team’s primary robot, the Meka M1 Mobile Manipulator (left) demonstrated the latest developments in human-robot-interactions within the domestic setting. The team showcased how the Meka was able to open previously shut doors, navigate safely around a person blocking its way, and recognize and grasp many household objects. According to the team, “the robot skills proved to be very effective for designing determined tasks, including more script-like tasks, e.g. ’Follow-Me’ or ’Who-is-Who’, as well as more flexible tasks including planning and dialogue aspects, e.g. ’General-PurposeService-Robot’ or ’Open-Challenge’.”

RoboCup @Work

The @Work category debuted in 2016 with the German team from the Leibniz Universität Hannover (LUHbots) winning first place. While LUHbots’ hardware was mostly off the shelf parts (a mobile robot KUKA youBot), the software utilized a number of proprietary algorithms. According to the paper, “in the RoboCup we use this software e.g. to grab objects using inverse kinematics, to optimize trajectories and to create fast and smooth movements with the manipulator. Besides the usability the main improvements are the graph based planning approach and the higher control frequency of the base and the manipulator.” The key to using this approach within a factory setting is its robust object recognition. The paper explains, “the robot measures the speed and position of the object. It calculates the point and time where the object reaches the task place. The arm moves above the calculated point. Waits for the object and accelerates until the arm is directly above the moving-object with the same speed. Overlapping the down movement with the current speed until gripping the object. The advantage of this approach is that while the calculated position and speed are correct every orientation and much higher objects can be gripped.”

Similar to other finalists, LUHbots’ object recognition software became the determining factor to its success. RoboCup’s goal of playing WorldCup soccer with robots may seem trivial, but its practice is anything but meaningless. In each category, the advances developed on the field of competitive science are paying real dividends on a global scale across many industries.

In the words of the RoboCup mission statement: “The ultimate goal is to ‘develop a robot soccer team which beats the human world champion team.’ (A more modest goal is ‘to develop a robot soccer team which plays like human players.’) Needless to say, the accomplishment of the ultimate goal will take decades of effort. It is not feasible with current technologies to accomplish this in the near future. However, this goal can easily lead to a series of well-directed subgoals. Such an approach is common in any ambitious or overly ambitious project. In the case of the American space program, the Mercury project and the Gemini project, which manned an orbital mission, were two precursors to the Apollo mission. The first subgoal to be accomplished in RoboCup is ‘to build real and software robot soccer teams which play reasonably well with modified rules.’ Even to accomplish this goal will undoubtedly generate technologies, which will impact a broad range of industries.”

* Editor’s note: thank you to Robohub for providing a twenty-year history of RoboCup videos. 

“The new frontier is learning how to design the relationships between people, robots, and infrastructure,” says David Mindell, the Dibner Professor of the History of Engineering and Manufacturing, and a professor of aeronautics and astronautics. “We need new sensors, new software, new ways of architecting systems.” Photo: Len Rubenstein

Science and technology are essential tools for innovation, and to reap their full potential, we also need to articulate and solve the many aspects of today’s global issues that are rooted in the political, cultural, and economic realities of the human world. With that mission in mind, MIT’s School of Humanities, Arts, and Social Sciences has launched The Human Factor — an ongoing series of stories and interviews that highlight research on the human dimensions of global challenges. Contributors to this series also share ideas for cultivating the multidisciplinary collaborations needed to solve the major civilizational issues of our time.

David Mindell, the Frances and David Dibner Professor of the History of Engineering and Manufacturing and Professor of Aeronautics and Astronautics at MIT, researches the intersections of human behavior, technological innovation, and automation. Mindell is the author of five acclaimed books, most recently “Our Robots, Ourselves: Robotics and the Myths of Autonomy” (Viking, 2015). He is also the co-founder of Humatics Corporation, which develops technologies for human-centered automation. SHASS Communications recently asked him to share his thoughts on the relationship of robotics to human activities, and the role of multidisciplinary research in solving complex global issues.

Q: A major theme in recent political discourse has been the perceived impact of robots and automation on the United States labor economy. In your research into the relationship between human activity and robotics, what insights have you gained that inform the future of human jobs, and the direction of technological innovation?

A: In looking at how people have designed, used, and adopted robotics in extreme environments like the deep ocean, aviation, or space, my most recent work shows how robotics and automation carry with them human assumptions about how work gets done, and how technology alters those assumptions. For example, the U.S. Air Force’s Predator drones were originally envisioned as fully autonomous — able to fly without any human assistance. In the end, these drones require hundreds of people to operate.

The new success of robots will depend on how well they situate into human environments. As in chess, the strongest players are often the combinations of human and machine. I increasingly see that the three critical elements are people, robots, and infrastructure — all interdependent.

Q: In your recent book “Our Robots, Ourselves,” you describe the success of a human-centered robotics, and explain why it is the more promising research direction — rather than research that aims for total robotic autonomy. How is your perspective being received by robotic engineers and other technologists, and do you see examples of research projects that are aiming at human-centered robotics?

A: One still hears researchers describe full autonom as the only way to go; often they overlook the multitude of human intentions built into even the most autonomous systems, and the infrastructure that surrounds them. My work describes situated autonomy, where autonomous systems can be highly functional within human environments such as factories or cities. Autonomy as a means of moving through physical environments has made enormous strides in the past ten years. As a means of moving through human environments, we are only just beginning. The new frontier is learning how to design the relationships between people, robots, and infrastructure. We need new sensors, new software, new ways of architecting systems.

Q: What can the study of the history of technology teach us about the future of robotics?

A: The history of technology does not predict the future, but it does offer rich examples of how people build and interact with technology, and how it evolves over time. Some problems just keep coming up over and over again, in new forms in each generation. When the historian notices such patterns, he can begin to ask: Is there some fundamental phenomenon here? If it is fundamental, how is it likely to appear in the next generation? Might the dynamics be altered in unexpected ways by human or technical innovations?

One such pattern is how autonomous systems have been rendered less autonomous when they make their way into real world human environments. Like the Predator drone, future military robots will likely be linked to human commanders and analysts in some ways as well. Rather than eliding those links, designing them to be as robust and effective as possible is a worthy focus for researchers’ attention.

Q: MIT President L. Rafael Reif has said that the solutions to today’s challenges depend on marrying advanced technical and scientific capabilities with a deep understanding of the world’s political, cultural, and economic realities. What barriers do you see to multidisciplinary, sociotechnical collaborations, and how can we overcome them?

A: I fear that as our technical education and research continues to excel, we are building human perspectives into technologies in ways not visible to our students. All data, for example, is socially inflected, and we are building systems that learn from those data and act in the world. As a colleague from Stanford recently observed, go to Google image search and type in “Grandma” and you’ll see the social bias that can leak into data sets — the top results all appear white and middle class.

Now think of those data sets as bases of decision making for vehicles like cars or trucks, and we become aware of the social and political dimensions that we need to build into systems to serve human needs. For example, should driverless cars adjust their expectations for pedestrian behavior according to the neighborhoods they’re in?

Meanwhile, too much of the humanities has developed islands of specialized discourse that is inaccessible to outsiders. I used to be more optimistic about multidisciplinary collaborations to address these problems. Departments and schools are great for organizing undergraduate majors and graduate education, but the old two-cultures divides remain deeply embedded in the daily practices of how we do our work. I’ve long believed MIT needs a new school to address these synthetic, far-reaching questions and train students to think in entirely new ways.

Interview prepared by MIT SHASS Communications
Editorial team: Emily Hiestand (series editor), Daniel Evans Pritchard

Ex Machina. Source: Youtube/Universal Pictures

This week I attended an “Artificial Intelligence (AI) Roundtable” of leading scientists, entrepreneurs and venture investors. As the discussion focused mainly on basic statistical techniques, I left feeling unfulfilled. My friend, Matt Turck, recently wrote that “just about every major tech company is working very actively on AI,” which also means that every startup hungry for capital is purchasing a dot ‘ai’ domain name. As the lines blur between what is and what really isn’t, I feel it necessary to provide readers with a quick lens of how to view intelligent agents for mechatronics.

For 65 years, The Turing Test remained unsolvable until a computer program called “Eugene Goostman” conquered it in 2014. The chatbot, which simulates a 13-year-old Ukrainian boy, did the unthinkable, fooling a group of human judges into thinking it was more real than a live person on the other side of the screen. Alan Turing’s original thesis in developing the schema was testing the premise – “Can Machines Think?”

According to the competition’s organizer, Kevin Warwick of Coventry University,
“The words Turing test have been applied to similar competitions around the world. However, this event involved the most simultaneous comparison tests than ever before, was independently verified and, crucially, the conversations were unrestricted. We are therefore proud to declare that Alan Turing’s test was passed for the first time.” Since then, many have debated whether the threshold was indeed crossed then, or before, or ever…

Last December, the Turing Test’s sound barrier (of sorts) was broken by a group of Researchers from MIT’s Computer Science and Artificial Intelligence Lab (CSAIL). Using a deep-learning algorithm, the AI fooled human viewers into thinking the images and corresponding sounds were indistinguishable. As a silent video clip played (above), the computer program produced a sound that was realistic enough to fool even the most hardened audiophiles. According to the research paper, the authors envision their algorithms in the future be used to automatically produce movies, TV shows, as well as helping robots better understand their environments.

Lead author Andrew Owens said: “being able to predict sound is an important first step toward being able to predict the consequences of physical interactions with the world. A robot could look at a sidewalk and instinctively know that the cement is hard and the grass is soft, and therefore know what would happen if they stepped on either of them.”

Unpacking Owens’ statement, one uncovers a lot of theory of artificial intelligence. The algorithm is using a deep learning approach to automatically train the computer to match sounds to pictures through experience. According to the paper, the researchers spent several months recording close to 1,000 videos of approximately 46,000 sounds, each representing unique objects being “hit, scraped and prodded with a drumstick.”

Typically artificial intelligence starts with creating an agent to solve a specific program, often agents are symbolic or logical, but deep learning approaches like the MIT example uses sub-symbolic neural networks that attempt to emulate how the human brain learns. Autonomous devices, like robots, use machine learning approaches to combine algorithms with experiences.

AI is very complex and relies on theories derived from a multitude of disciplines, including computer science, mathematics, psychology, linguistics, philosophy, neuroscience and statistics. For the purposes of this article, it may be best to group modern approaches into two categories: “supervised” and “unsupervised” learning. Supervised uses a method of “continuous target values” to mine data to predict outcomes, similar to how nominal variables are used to label values in statistics. In unsupervised learning there is no notion of a target value. Rather, algorithms perform operations based upon clustering data into classifications and then determine relationships between the inputs and outputs with numerical regression or other filtering methods. The primary difference between the two approaches is that in unsupervised learning the computer program is able to automatically connect and label patterns from streams of inputs. In the above example, the algorithm connects the silent video images to the library of drumstick sounds.

Now that the connections are made between patterns and inputs, the next step is to control behavior. The most rudimentary AI applications are divided between classifiers (“if shiny then silver”) and controllers (“if shiny then pick up”). It is important to note that controllers also classify conditions before performing actions. Classifiers use pattern matching to determine the closest match. In supervised learning, each pattern belongs to a certain predefined class. A data set is the labeling of classes and the observations received through experience. The more experience the increased number of classes and likewise data to input.

In robotics, unsupervised (machine) learning is required for object manipulation, navigation, localization, mapping and motion planning, which becomes increasingly more challenging in unstructured environments such as, enhanced manufacturing and autonomous driving. As a result, deep learning has given birth to many sub-specialities of AI, such as computer vision, speech recognition and natural language processing. Rodney Brooks (above) professed in 1986 a new theory of AI that led to the greatest advancement in machine intelligence for robotics. Brook’s “Subsumption Architecture” created a paradigm for real-time learning via live interaction through sensor inputs.

In his own words, Brooks said in a 2015 interview, “The work I did in the 80s on what I called subsumption architecture then led directly to the iRobot Roomba. And there are 14 million of them deployed worldwide. And it was also used in the iRobot PackBot, and there were 4,500 PackBots in Iraq and Afghanistan remediating roadside bombs – and by the way, there are some PackBots and Warriors from iRobot inside Fukushima now, using the subsumption architecture. There’s a variation of subsumption inside Baxter – we call it behaviour-based now, but it’s descendent of that subsumption architecture. And that’s what lets Baxter be aware of different things in parallel, for instance, it’s picked something up, it’s put it in a box, something goes wrong, and it drops the object, sadly. A traditional robot would just continue and sort of mime putting the thing in the box, but Baxter is aware of that changes behaviour – that’s using the behaviour-based approach, which is a variation on subsumption. So it is part of Baxter’s intelligence.”

Many next generation computer scientists are focusing on designing not just the software that acts like the brain, but hardware that is designed like a human cranium. Today, infamous deep learning algorithms, like Siri and Google Translate, run on traditional computing processing platforms that consume a lot of energy as the logic and memory boards are separated. Last year, Google’s AlphaGo, the most successful deep learning program to date, was only able to beat the world-champion human player in Go after being trained on a database of thirty million moves running on approximately one million watts of power.

When asking Dr. Devanand Shenoy, formerly with the U.S. Department of Energy about Google’s project, he said: “AlphaGo had to be retrained for every new game (a feature of narrow AI where the machine does one thing very well, even better than humans). Once learning and training algorithms are implemented on neuromorphic hardware that is distributed, asynchronous, perhaps event driven, and fault-tolerant, the ability to process data with many orders of magnitude improvements in energy-efficiency as well as superhuman speed across multiple applications could be possible in the future. Recent trends in transfer learning with reservoir computing as an example, suggest that artificial network configurations may be trained to learn more than one application.”

Superhuman intelligence could sound super scary, however, I am sobered by the predictions in the 1960s by the founder of artificial intelligence, Professor Herbert Simon, who stated then: “machines will be capable, within twenty years, of doing any work a man can do.”

As this article only touches the surface of artificial intelligence, this subject matter will be further explored on June 13th in New York City at RobotLabNYC’s next event with Dr. Howard Morgan (FirstRound Capital) and Tom Ryden (MassRobotics). There are limited seats available, so be sure to RSVP today.

In this article, we explain in plain language machine learning.

Learning to learn

In 1959, Arthur Samuel, a pioneer in the field of machine learning (ML) defined it as the “field of study that gives computers the ability to learn without being explicitly programmed”.

ML can be understood as computational methods that use experience to improve performance or to make accurate predictions. In this case, experience refers to past information or data that is available to us, which has been labelled and categorized. As with any computational exercise, the quality and amount of the data will be crucial to the accuracy of the predictions that will be made.

Looking through this lens, ML seems to be a lot like statistical modelling. In statistical modelling, we collect data, verify that it is clean — in other words, that we have completed, corrected, or deleted any incomplete, incorrect, or irrelevant parts of the data — and then use this clean dataset to test hypotheses and make predictions and forecasts. The idea behind statistical modelling is the attempt to represent complex issues in relatively generalizable terms, which is to say, terms that explain most events studied. Effectively, we programme the algorithm to perform certain functions based on the data we submit. Put differently, the algorithm is static. It needs a programmer to tell it what to do when it is fed with data. So far, it makes indeed sense to use this approach when activated by the programmer.

But with ML, the procedure is flipped. Rather than preselecting a model and feeding data into it, in ML it is the data that determines which analytic technique should be selected to best perform the task at hand. In other words, the computer uses the data that it has to select and train the algorithm. Hence the algorithm is no longer static. It analyses the data to which it is exposed, makes a determination on the best course of action, and then acts. In essence, it “learns” from the data and in doing so, knowledge can be extracted from the data.

This method of learning is based on repetition. Remember that an algorithm is nothing more than a set of instructions that a computer uses to transform an input into a particular output. Thus in ML, the learning aspect is just an algorithm repeating its execution operation over and over again and making slight adjustments until a certain set of conditions is met. The litmus test of a learning algorithm is when it is able to predict when new data is given to it on which it had not previously been trained.

Evolution of ML

Data obviously plays a primary role in this methodological process. More importantly, it is the structure of the data that determines how the learning process will occur. It is here that we see the three-levels of ML:

Supervised ML

Here the computer is trained on data that is well labelled. This means that the data is already tagged with the correct label or the correct outcome. For example, if we were to teach a computer to distinguish between the picture of a cat and a dog, then we would tag the data in the following way:

This labelling process should be done by the programmer and having learned the difference, the ML algorithm can now classify new information that is given to it and determine if the new image is that of a dog or a cat.

Based on this simplistic method, supervised ML can be used to perform much more complicated operations. One use is understanding how to read figures and alphabets. The way one person writes the number “1” or the letter “A” will not be the same way as another person does.

By feeding the computer with vast amounts of labelled examples of the number “1” or “A”, we can train the algorithm to see the various versions these figures. The computer thus begins to learn the variations and becomes increasingly competent at understanding these patterns.

Today, computers are better than humans in recognizing such patterns of handwriting. The larger the dataset, the better trained the algorithm. Once trained, the algorithm is given new data and uses its past experience to predict an outcome.

Unsupervised ML

This is where the algorithm is trained using a dataset that does not have any labels. The algorithm is not told what the data represents. In this case, the learning process is dependent on the identification of patterns that are repeatedly created in the data. Using the cat and dog example, the algorithm begins to separate the images it receives based on the inherent characteristics of dogs.

In unsupervised learning, the algorithms must use methods of estimation based on inferential statistics to discover patterns, relationships and correlations with the raw, unlabelled dataset. As patterns are identified, the algorithm uses statistics to identify boundaries within the dataset. Data with similar patterns are grouped together, creating subsets of data. As the classification process continues, the algorith begins to understand the dataset it is analysing, allowing it to predict the categorization of future data.

This clustering of data can automate decision making, adding a layer of sophistication to unsupervised learning. More importantly, it allows us to leverage data in a new way. What we lack in knowledge we make up for in data.

Reinforcement ML

Reinforcement learning is like unsupervised ML in that the training data is also unlabelled. However, when asked a question about the data, the outcome is graded – so there is still a level of supervision. The algorithm is presented with data that lacks labels, but is given an example with a positive or negative result. This positive or negative grade provides a feedback loop to the algorithm allowing it to determine if the solution it is providing is solving a problem or not. Effectively, it is the computerised version of human trial and error learning.

Reinforcement ML is often used to make strategies. As decisions lead to consequences, the output action is prescriptive, and not just descriptive, as in unsupervised learning. This kind of learning has been used to train computers how to play games. This is the idea behind the company DeepMind, acquired by Google in 2014, which trained which they trained their algorithm to learn how to play Atari.

It went on to create AlphaGo which defeated the best human professional Go player 4-1 in the game of Go, one of the most complex games in the world.

Implications for businesses

Today machine learning is being used in a number of areas. Google’s self-driving car was developed using machine learning and today machines can lip read faster than humans. ML has been infiltrating almost every sector of finance in recent years. For instance, ML is being used for algorithmic trading, analysing time series data, portfolio management, fraud detection, customer service, news analysis, investment strategy construction, etc.

But the real power of machine learning is unleashed with neural networks.

This article was originally on The Conversation. Click here to access the original.

Last week I had the pleasure of debating the question “does AI pose a threat to society?” with friends and colleagues Christian List, Maja Pantic and Samantha Payne. The event was organised by the British Academy and brilliantly chaired by the Royal Society’s director of science policy Claire Craig. Here follows my opening statement:

My opening statement

One Friday afternoon in 2009 I was called by a science journalist at, I recall, the Sunday Times. He asked me if I knew that there was to be a meeting of the AAAI to discuss robot ethics. I said no I don’t know of this meeting. He then asked “are you surprised they are meeting to discuss robot ethics” and my answer was no. We talked some more and agreed it was actually a rather dull story: a case of scientists behaving responsibly. I really didn’t expect the story to appear but checked the Sunday paper anyway, and there in the science section was the headline Scientists fear revolt of killer robots. (I then spent the next couple of days on the radio explaining that no, scientists do not fear a revolt of killer robots.)

So, fears of future super intelligence – robots taking over the world – are greatly exaggerated: the threat of an out-of-control super intelligence is a fantasy – interesting for a pub conversation perhaps. It’s true we should be careful and innovate responsibly, but that’s equally true for any new area of science and technology. The benefits of robotics and AI are so significant, the potential so great, that we should be optimistic rather than fearful. Of course robots and intelligent systems must be engineered to very high standards of safety for exactly the same reasons that we need our washing machines, cars and airplanes to be safe. If robots are not safe people will not trust them. To reach it’s full potential what robotics and AI needs is a dose of good old fashioned (and rather dull) safety engineering.

In 2011 I was invited to join a British Standards Institute working group on robot ethics, which drafted a new standard BS 8611 Guide to the ethical design of robots and robotic systems, published in April 2016. I believe this to be the world’s first standard on ethical robots.

Also in 2016 the very well regarded IEEE standards association – the same organization that gave us WiFi – launched a Global initiative on Ethical Considerations in AI and Autonomous Systems. The purpose of this Initiative is to ensure every technologist is educated and empowered to prioritize ethical considerations in the design and development of autonomous and intelligent systems; in a nutshell, to ensure ethics are baked in. In December we published Ethically Aligned Design: A Vision for Prioritizing Human Well Being with AI and Autonomous Systems. Within that initiative I’m also leading a new standard on transparency in autonomous systems, based on the simple principle that it should always be possible to find out why an AI or robot made a particular decision.

We need to agree ethical principles, because they are needed to underpin standards – ways of assessing and mitigating the ethical risks of robotics and AI. But standards needs teeth and in turn underpin regulation. Why do we need regulation? Think of passenger airplanes; the reason we trust them is because it’s a highly regulated industry with an amazing safety record, and robust, transparent processes of air accident investigation when things do go wrong. Take one example of a robot that we read a lot about in the news – the Driverless Car. I think there’s a strong case for a driverless car equivalent of the CAA, with a driverless car accident investigation branch. Without this it’s hard to see how driverless car technology will win public trust.

Does AI pose a threat to society? No. But we do need to worry about the down to earth questions of present day rather unintelligent AIs; the ones that are deciding our loan applications, piloting our driverless cars or controlling our central heating. Are those AIs respecting our rights, freedoms and privacy? Are they safe? When AIs make bad decisions, can we find out why? And I worry too about the wider societal and economic impacts of AI. I worry about jobs of course, but actually I think there is a bigger question: how can we ensure that the wealth created by robotics and AI is shared by all in society?

Thank you.

The featured image of this article was used to advertise the BA’s series of events on the theme Robotics, AI and Society. The reason I draw your attention to it here is that one of the many interesting questions to the panel was about the way that AI tends to be visualised in the media. This kind of human face coalescing (or perhaps emerging) from the atomic parts of the AI seems to have become a trope for AI. Is it a helpful visualisation of the human face of AI, or does it mislead to an impression that AI has human characteristics?

Talking Machines is entering its third season and going through some changes. Our founding host Ryan is moving on and in his place, Neil Lawrence of Amazon is taking over as co-host. We say thank you and goodbye to Ryan with an interview about his work.

If you liked this article, you may also want to read:

• Talking Machines: Restricted Boltzmann Machines, with Eric Lander
• Talking Machines: Automatic Translation and t-SNE, with Hal Daume
• Talking Machines: Generative art and Hamiltonian Monte Carlo, with Doug Eck
• Talking Machines: Machine learning and the Flint water crisis, with Jake Abernethy
• Talking Machines: Gaussian processes and OpenAI, with IIya Sutskever

See all the latest robotics news on Robohub, or sign up for our weekly newsletter.

The many potential social and economic benefits from advances in AI-based technologies depend entirely on the environment in which these technologies evolve, says the Royal Society. According to a new report from the UK’s science academy, urgent consideration needs to be given to the “careful stewardship” needed over the next ten years to ensure that the dividends from machine learning – the form of artificial intelligence that allows machines to learn from data – benefit all in UK society.

Machine Learning: the power and promise of computers that learn by example, published today (25 April 2017), comes at a critical time in the rapid development and use of this technology, and the growing debate about how it will reshape the UK economy and people’s lives. Crucially the report calls for research funding bodies to support a new wave of machine learning research that goes beyond technical challenges, and into areas aimed at addressing public confidence in machine learning – vital to the UK maintaining its internationally competitive edge at the forefront of this area.

The report also offers the first evidence about the UK public’s views on machine learning, including the application areas about which they are particularly positive, and the need for the real-world data feeding the growth of this technology to be dealt with fairly and securely.

Professor Peter Donnelly FRS, chair of the report’s working group and Director of the Wellcome Trust Centre for Human Genetics and Professor of Statistical Science at the University of Oxford, says:

“Machine learning is already used in many apps and services that we encounter every day. It is used to tag people in our photos, by our phones to interpret voice commands, by internet retailers to make recommendations, and by banks to spot unusual activity on a credit or debit card. However, these current applications only scratch the surface of understanding just how powerful a technology this could be.

“Machine learning will have an increasing impact on our lives and lifestyles over the next five to ten years. There is much work to be done so that we take advantage of machine learning’s potential and ensure that the benefits are shared, especially as this could be a key area of opportunity for the UK in the coming years.”

An action plan so no one is left behind and the benefits are shared

The report calls for action in a number of key areas over the next five to ten years to create an environment of “careful stewardship” that can help ensure that the benefits of this technology are felt broadly. Understanding who will be most affected, how the benefits are likely to be distributed, and where the opportunities for growth lie, will be key to designing the most effective interventions to enable people and businesses to adapt to, and take advantage of, the machine learning driven changes to their lives and livelihoods.

Supporting the development of skills at every level

• 23% of the UK population lack basic digital skills
• Build digital skills and understanding at every level from schools to universities, and into the workplace, and ensure that opportunities are not limited by gender, ethnicity or socio-economic background
• Government should consider introducing funded Masters courses in Machine Learning to develop a pool of informed users of machine learning across business, industry, and research sectors
• There is a critical need for increased training at PhD level and beyond to invest in the next generation of research leaders in machine learning.

Creating opportunities to use machine learning

• Integrate machine learning into the UK’s industrial strategy, to help businesses make the most of its potential
• Support a new wave of research in machine learning, including in areas that can address social or ethical concerns.

Creating a data environment that supports machine learning

• 90% of the world’s data has been created within the last five years
• Continue to build on the UK’s track record of open data and safe sharing of data
• Where there is value in accessing data that cannot be open – for example medical data or commercially sensitive industry data – ensuring there are frameworks and agreements in place which facilitate appropriate data sharing in these circumstances.

An enabling governance environment

• Support an informed public debate about what we want machine learning to do, and how the benefits are distributed
• Develop a new framework for data governance – one that can keep pace with the challenge of the governance of data and its uses in the 21st century. While there may be specific questions about the use of machine learning in particular contexts, these should be handled in a sector-specific way.

Results from the UK’s first in-depth assessment of public views on machine learning – carried out by the Royal Society and Ipsos MORI – demonstrate that while most people have not heard the term ‘machine learning’ (only 9% have), the vast majority have heard about or used at least one of its applications. The public do not have a single view on machine learning – attitudes, positive or negative, vary depending on the circumstances in which machine learning is being used. For example, computers that could help with medical diagnoses, or could help relieve pressure on professionals in core services such as health and social care, education, and policing were viewed more favourably than computers that could make investments in the stock market.

As with most new technologies, some of those interviewed expressed concerns. These included: depersonalisation, or machine learning systems replacing valued human experiences; the potential impact of machine learning on employment; the potential for machine learning systems to cause harm, for example accidents in autonomous vehicles; and machine learning systems restricting choice, such as when directing consumers to specific products and services.

Ongoing public confidence will be central to realising the benefits that machine learning promises, and continued engagement between machine learning researchers and practitioners and the public will be important as the field develops. The report also calls on researchers to consider the wider impact of their work and to receive training in recognising the ethical implications.

Most applications of deep learning use “convolutional” neural networks, in which the nodes of each layer are clustered, the clusters overlap, and each cluster feeds data to multiple nodes (orange and green) of the next layer. Image: Jose-Luis Olivares/MIT

In the past 10 years, the best-performing artificial-intelligence systems — such as the speech recognizers on smartphones or Google’s latest automatic translator — have resulted from a technique called “deep learning.”

Deep learning is, in fact, a new name for an approach to artificial intelligence called neural networks, which have been going in and out of fashion for more than 70 years. Neural networks were first proposed in 1944 by Warren McCullough and Walter Pitts, two University of Chicago researchers who moved to MIT in 1952 as founding members of what’s sometimes called the first cognitive science department.

Neural nets were a major area of research in both neuroscience and computer science until 1969, when, according to computer science lore, they were killed off by the MIT mathematicians Marvin Minsky and Seymour Papert, who a year later would become co-directors of the new MIT Artificial Intelligence Laboratory.

The technique then enjoyed a resurgence in the 1980s, fell into eclipse again in the first decade of the new century, and has returned like gangbusters in the second, fueled largely by the increased processing power of graphics chips.

“There’s this idea that ideas in science are a bit like epidemics of viruses,” says Tomas Poggio, the Eugene McDermott Professor of Brain and Cognitive Sciences at MIT, an investigator at MIT’s McGovern Institute for Brain Research, and director of MIT’s Center for Brains, Minds, and Machines. “There are apparently five or six basic strains of flu viruses, and apparently each one comes back with a period of around 25 years. People get infected, and they develop an immune response, and so they don’t get infected for the next 25 years. And then there is a new generation that is ready to be infected by the same strain of virus. In science, people fall in love with an idea, get excited about it, hammer it to death, and then get immunized — they get tired of it. So ideas should have the same kind of periodicity!”

Weighty matters

Neural nets are a means of doing machine learning, in which a computer learns to perform some task by analyzing training examples. Usually, the examples have been hand-labeled in advance. An object recognition system, for instance, might be fed thousands of labeled images of cars, houses, coffee cups, and so on, and it would find visual patterns in the images that consistently correlate with particular labels.

Modeled loosely on the human brain, a neural net consists of thousands or even millions of simple processing nodes that are densely interconnected. Most of today’s neural nets are organized into layers of nodes, and they’re “feed-forward,” meaning that data moves through them in only one direction. An individual node might be connected to several nodes in the layer beneath it, from which it receives data, and several nodes in the layer above it, to which it sends data.

To each of its incoming connections, a node will assign a number known as a “weight.” When the network is active, the node receives a different data item — a different number — over each of its connections and multiplies it by the associated weight. It then adds the resulting products together, yielding a single number. If that number is below a threshold value, the node passes no data to the next layer. If the number exceeds the threshold value, the node “fires,” which in today’s neural nets generally means sending the number — the sum of the weighted inputs — along all its outgoing connections.

When a neural net is being trained, all of its weights and thresholds are initially set to random values. Training data is fed to the bottom layer — the input layer — and it passes through the succeeding layers, getting multiplied and added together in complex ways, until it finally arrives, radically transformed, at the output layer. During training, the weights and thresholds are continually adjusted until training data with the same labels consistently yield similar outputs.

Minds and machines

The neural nets described by McCullough and Pitts in 1944 had thresholds and weights, but they weren’t arranged into layers, and the researchers didn’t specify any training mechanism. What McCullough and Pitts showed was that a neural net could, in principle, compute any function that a digital computer could. The result was more neuroscience than computer science: The point was to suggest that the human brain could be thought of as a computing device.

Neural nets continue to be a valuable tool for neuroscientific research. For instance, particular network layouts or rules for adjusting weights and thresholds have reproduced observed features of human neuroanatomy and cognition, an indication that they capture something about how the brain processes information.

The first trainable neural network, the Perceptron, was demonstrated by the Cornell University psychologist Frank Rosenblatt in 1957. The Perceptron’s design was much like that of the modern neural net, except that it had only one layer with adjustable weights and thresholds, sandwiched between input and output layers.

Perceptrons were an active area of research in both psychology and the fledgling discipline of computer science until 1959, when Minsky and Papert published a book titled “Perceptrons,” which demonstrated that executing certain fairly common computations on Perceptrons would be impractically time-consuming.

“Of course, all of these limitations kind of disappear if you take machinery that is a little more complicated — like, two layers,” Poggio says. But at the time, the book had a chilling effect on neural-net research.

“You have to put these things in historical context,” Poggio says. “They were arguing for programming — for languages like Lisp. Not many years before, people were still using analog computers. It was not clear at all at the time that programming was the way to go. I think they went a little bit overboard, but as usual, it’s not black and white. If you think of this as this competition between analog computing and digital computing, they fought for what at the time was the right thing.”

Periodicity

By the 1980s, however, researchers had developed algorithms for modifying neural nets’ weights and thresholds that were efficient enough for networks with more than one layer, removing many of the limitations identified by Minsky and Papert. The field enjoyed a renaissance.

But intellectually, there’s something unsatisfying about neural nets. Enough training may revise a network’s settings to the point that it can usefully classify data, but what do those settings mean? What image features is an object recognizer looking at, and how does it piece them together into the distinctive visual signatures of cars, houses, and coffee cups? Looking at the weights of individual connections won’t answer that question.

In recent years, computer scientists have begun to come up with ingenious methods for deducing the analytic strategies adopted by neural nets. But in the 1980s, the networks’ strategies were indecipherable. So around the turn of the century, neural networks were supplanted by support vector machines, an alternative approach to machine learning that’s based on some very clean and elegant mathematics.

The recent resurgence in neural networks — the deep-learning revolution — comes courtesy of the computer-game industry. The complex imagery and rapid pace of today’s video games require hardware that can keep up, and the result has been the graphics processing unit (GPU), which packs thousands of relatively simple processing cores on a single chip. It didn’t take long for researchers to realize that the architecture of a GPU is remarkably like that of a neural net.

Modern GPUs enabled the one-layer networks of the 1960s and the two- to three-layer networks of the 1980s to blossom into the 10-, 15-, even 50-layer networks of today. That’s what the “deep” in “deep learning” refers to — the depth of the network’s layers. And currently, deep learning is responsible for the best-performing systems in almost every area of artificial-intelligence research.

Under the hood

The networks’ opacity is still unsettling to theorists, but there’s headway on that front, too. In addition to directing the Center for Brains, Minds, and Machines (CBMM), Poggio leads the center’s research program in Theoretical Frameworks for Intelligence. Recently, Poggio and his CBMM colleagues have released a three-part theoretical study of neural networks.

The first part, which was published last month in the International Journal of Automation and Computing, addresses the range of computations that deep-learning networks can execute and when deep networks offer advantages over shallower ones. Parts two and three, which have been released as CBMM technical reports, address the problems of global optimization, or guaranteeing that a network has found the settings that best accord with its training data, and overfitting, or cases in which the network becomes so attuned to the specifics of its training data that it fails to generalize to other instances of the same categories.

There are still plenty of theoretical questions to be answered, but CBMM researchers’ work could help ensure that neural networks finally break the generational cycle that has brought them in and out of favor for seven decades.

MIT Professor Daniela Rus, director of CSAIL, said the goal of a new SystemsThatLearn@CSAIL initiative is “to create a new generation of AI tools that are deeply rooted in systems.” Photo: Jason Dorfman/MIT CSAIL

From self-driving cars to the internet of things, artificial intelligence (AI) has reached new levels of sophistication in recent years. With that in mind, this week MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) launched an industry collaboration focused on using machine learning to create functional human-like systems.

Nearly 40 senior researchers will participate in the new “SystemsThatLearn@CSAIL” (STL) initiative alongside a range of organizations that include founding members BT, Microsoft, Nokia Bell Labs, Salesforce, and Schlumberger. Member companies will work with CSAIL scientists to suggest new lines of research and develop real-world applications.

“Developing capabilities in AI and machine learning are key to the future of fields like finance, energy, manufacturing, and health care,” says STL Executive Director Lori Glover. “While the demand for expertise is great, the supply of talent remains small and unevenly distributed. By democratizing the field of AI, SystemsThatLearn@CSAIL is an effort to address that skills gap.”

STL builds on CSAIL’s Big Data Initiative, which developed tools for handling complex datasets. While many machine-learning solutions are trained and deployed in separate phases, STL aims to integrate these processes, focusing on a range of resources to handle distributed data and computing power.

Another goal is to make key aspects of data science less laborious. A 2016 report found that data scientists spend 80 percent of their time collecting and organizing data, and only 20 percent analyzing it.

“By promoting industry interaction with academia, we’re hoping to create new tools and systems that can increase productivity by automating much of the tedious work of data science,” says MIT Professor Samuel Madden, one of STL’s two faculty leads alongside Professor Tommi Jaakkola. “We are already seeing that areas like autonomous vehicles and personalized health care have the potential to transform entire industries.”

In her opening remarks, CSAIL Director Daniela Rus described AI and systems researchers as two communities that would benefit from stronger collaboration.

“Our grand vision is to create a new generation of AI tools that are deeply rooted in systems and that can make those systems better,” Rus said. “My aspiration is to get to a place where machine learning becomes a normal part of what an operating system does.”

One STL project is the data discovery tool “Data Civilizer,” which allows organizations to discover related datasets from thousands of distinct business databases and files. Another is “Model DB,” a machine-learning management system that saves time for data scientists and lets them easily correlate performance on particular training examples with specific model features.

According to STL Technical Director Stephen Buckley, many of the software tools they develop will be released under MIT’s open source license.

“The ultimate aim is to democratize access and use of machine learning tools, without requiring advanced knowledge of the underlying technologies,” says Buckley.

YouTube: Dylan F Glas

“Robot’s Delight – A Lyrical Exposition on Learning by Imitation from Human-Human Interaction” is a video submission that won Best Video at the 2017 ACM/IEEE International Conference on Human-Robot Interaction (HRI 2017). The team also provides an in-depth explanation of the techniques and robotics in the video.

Watch it here:

Select video clips (c) 2016, IEEE. Reused, with permission.

Although social robots are growing in popularity and technical feasibility, it is still unclear how we can effectively program social behaviors. There are many difficulties in programming social robots — we need to design hundreds or thousands of dialogue rules, anticipate situations the robot will face, handle common recognition errors, and program the robot to respond to many variations of human speech and behavior. Perhaps most challenging is that we often do not understand the reasoning behind our own behavior and so it is hard to program such implicit knowledge into robots.

In this video, we present two studies exploring learning-by-imitation from human-human interaction. In these studies, we developed techniques for learning typical actions and execution logic directly from people interacting naturally with each other. We propose that this is more scalable and robust than developing interaction logic by hand, and it requires much less effort.

In the first study, we asked participants to role-play a shopkeeper and a customer in a camera shop scenario, and we recorded their motion and speech in 178 interactions. By extracting typical motion and speech actions using unsupervised clustering algorithms, we created a set of robot behaviors and trained a machine learning classifier to predict which of those actions a human shopkeeper would have performed in any given situation. Using this classifier, we programmed a Robovie robot to imitate the movement and speech behavior of the shopkeeper, e.g. greeting the customer, answering questions about camera features, and introducing different cameras.

Experiment results showed that our techniques enabled the robot to perform correct behaviors 84.8% of the time, which was particularly interesting since speech recognition was only 76.8% accurate. This illustrates the robustness of the system to sensor noise, which is one advantage of using noisy, real-world data for training.

In the second study, we used a similar technique to train the android ERICA to imitate people’s behavior in a travel agent scenario. In this case, the challenge was to model the topic of the interaction so that ERICA could learn to answer ambiguous questions like “how much does this package cost?”.

We did this by observing that utterances in the same topic tend to occur together in interactions, so we calculated co-occurrence metrics similar to those used in product recommendation systems for online shopping sites. Using these metrics, we were able to cluster the customer and shopkeeper actions into topics, and these topics were used to improve ERICA’s predictor in order to answer ambiguous questions.

In both of these studies, we illustrated a completely hands-off approach to developing robot interaction logic – the robots learned only from example data of people interacting with each other, and no designer or programmer was needed! We think scalable, data-driven techniques like these promise to be powerful tools for developing even richer, more humanlike interaction logic for robots in the future.

The extended abstract for this video can be found here.

For the full details of the Robovie study, please see our IEEE Transactions on Robotics paper.

Phoebe Liu, Dylan F. Glas, Takayuki Kanda, and Hiroshi Ishiguro, Data-Driven HRI: Learning Social Behaviors by Example from Human-Human Interaction, in IEEE Transactions on Robotics, Vol. 32, No. 4, pp. 988-1008, 2016.

Author’s preprint is here.

In a surprise move which shook Baidu’s stock, Andrew Ng, Baidu’s Chief Scientist and partner with Baidu’s founder Robin Li Yanhong in expanding Baidu’s AI activities into machine learning, driverless vehicles, and to power AI usage in advertising, maps, takeout delivery, and voice-activated search, resigned.

Baidu’s AI group has expanded to over 1,300 people in China and Silicon Valley. Ng, who has been with Baidu since 2014 oversaw most of that change.

Some see Ng’s departure as a setback to their AI efforts, hence the shakeup in the stock price. Others comment that the timing was right both for him and for Baidu.

Ng blogged about his resignation that:

I will be resigning from Baidu, where I have been leading the company’s AI Group. Baidu’s AI is incredibly strong, and the team is stacked up and down with talent; I am confident AI at Baidu will continue to flourish. After Baidu, I am excited to continue working toward the AI transformation of our society and the use of AI to make life better for everyone.

In January, Lu Qi, an AI exec from Microsoft, was hired as COO which included oversight of Baidu’s AI research and business. Today he announced Ng’s replacement as head of AI Group to be Wang Haifeng, also an AI expert, and previous president of the Association for Computational Linguistics, an international scientific and professional society for people working on computational problems involving human language.

One analyst, Kitty Fok, from IDC, a provider of research, consulting and events for the data and telecom industries, said of Ng’s choice to leave Baidu:

He wants to start his own company. Ng saw a golden opportunity to become an entrepreneur. It’s perfect timing for him because he has the expertise, he’s well known in his field, and interest is high for innovative tech start-ups. I wouldn’t be surprised if Baidu, Tencent Holdings and Alibaba Group would be interested in investing in Ng’s potential new AI enterprise.

Engineers have designed a microfluidic device they call a “tree-on-a-chip,” which mimics the pumping mechanism of trees and other plants.

Trees and other plants, from towering redwoods to diminutive daisies, are nature’s hydraulic pumps. They are constantly pulling water up from their roots to the topmost leaves, and pumping sugars produced by their leaves back down to the roots. This constant stream of nutrients is shuttled through a system of tissues called xylem and phloem, which are packed together in woody, parallel conduits.

Now engineers at MIT and their collaborators have designed a microfluidic device they call a “tree-on-a-chip,” which mimics the pumping mechanism of trees and plants. Like its natural counterparts, the chip operates passively, requiring no moving parts or external pumps. It is able to pump water and sugars through the chip at a steady flow rate for several days. The results are published this week in Nature Plants.

Anette “Peko” Hosoi, professor and associate department head for operations in MIT’s Department of Mechanical Engineering, says the chip’s passive pumping may be leveraged as a simple hydraulic actuator for small robots. Engineers have found it difficult and expensive to make tiny, movable parts and pumps to power complex movements in small robots. The team’s new pumping mechanism may enable robots whose motions are propelled by inexpensive, sugar-powered pumps.

“The goal of this work is cheap complexity, like one sees in nature,” Hosoi says. “It’s easy to add another leaf or xylem channel in a tree. In small robotics, everything is hard, from manufacturing, to integration, to actuation. If we could make the building blocks that enable cheap complexity, that would be super exciting. I think these [microfluidic pumps] are a step in that direction.”

Hosoi’s co-authors on the paper are lead author Jean Comtet, a former graduate student in MIT’s Department of Mechanical Engineering; Kaare Jensen of the Technical University of Denmark; and Robert Turgeon and Abraham Stroock, both of Cornell University.

A hydraulic lift

The group’s tree-inspired work grew out of a project on hydraulic robots powered by pumping fluids. Hosoi was interested in designing hydraulic robots at the small scale, that could perform actions similar to much bigger robots like Boston Dynamic’s Big Dog, a four-legged, Saint Bernard-sized robot that runs and jumps over rough terrain, powered by hydraulic actuators.

“For small systems, it’s often expensive to manufacture tiny moving pieces,” Hosoi says. “So we thought, ‘What if we could make a small-scale hydraulic system that could generate large pressures, with no moving parts?’ And then we asked, ‘Does anything do this in nature?’ It turns out that trees do.”

The general understanding among biologists has been that water, propelled by surface tension, travels up a tree’s channels of xylem, then diffuses through a semipermeable membrane and down into channels of phloem that contain sugar and other nutrients.

The more sugar there is in the phloem, the more water flows from xylem to phloem to balance out the sugar-to-water gradient, in a passive process known as osmosis. The resulting water flow flushes nutrients down to the roots. Trees and plants are thought to maintain this pumping process as more water is drawn up from their roots.

“This simple model of xylem and phloem has been well-known for decades,” Hosoi says. “From a qualitative point of view, this makes sense. But when you actually run the numbers, you realize this simple model does not allow for steady flow.”

In fact, engineers have previously attempted to design tree-inspired microfluidic pumps, fabricating parts that mimic xylem and phloem. But they found that these designs quickly stopped pumping within minutes.

It was Hosoi’s student Comtet who identified a third essential part to a tree’s pumping system: its leaves, which produce sugars through photosynthesis. Comtet’s model includes this additional source of sugars that diffuse from the leaves into a plant’s phloem, increasing the sugar-to-water gradient, which in turn maintains a constant osmotic pressure, circulating water and nutrients continuously throughout a tree.

Running on sugar

With Comtet’s hypothesis in mind, Hosoi and her team designed their tree-on-a-chip, a microfluidic pump that mimics a tree’s xylem, phloem, and most importantly, its sugar-producing leaves.

To make the chip, the researchers sandwiched together two plastic slides, through which they drilled small channels to represent xylem and phloem. They filled the xylem channel with water, and the phloem channel with water and sugar, then separated the two slides with a semipermeable material to mimic the membrane between xylem and phloem. They placed another membrane over the slide containing the phloem channel, and set a sugar cube on top to represent the additional source of sugar diffusing from a tree’s leaves into the phloem. They hooked the chip up to a tube, which fed water from a tank into the chip.

With this simple setup, the chip was able to passively pump water from the tank through the chip and out into a beaker, at a constant flow rate for several days, as opposed to previous designs that only pumped for several minutes.

“As soon as we put this sugar source in, we had it running for days at a steady state,” Hosoi says. “That’s exactly what we need. We want a device we can actually put in a robot.”

Hosoi envisions that the tree-on-a-chip pump may be built into a small robot to produce hydraulically powered motions, without requiring active pumps or parts.

“If you design your robot in a smart way, you could absolutely stick a sugar cube on it and let it go,” Hosoi says.

This research was supported, in part, by the Defense Advance Research Projects Agency.

In a poignant play traveling throughout the UK, a robot is co-star and companion to the wife of the (now deceased) robot builder, with the wife developing early Alzheimer’s. The play explores very human themes about love, death, and disease, all handled extremely sensitively with RoboThespian playing a large role.

Jon Welch, the writer and director, said of the play:

“It’s a story about a robot maker. All of his life he builds robots, but he develops degenerative illness in mid-life and realizes he’s not going to live to remain a companion to his wife. His wife, by now, is developing early Alzheimer’s, so he builds his final creation, his final robot to be a companion to his wife.”

The robot is from Engineered Arts, a 12-year-old UK company that develops an ever expanding range of humanoid and semi-humanoid robots featuring natural human-like movement and advanced social behaviours. RoboThespian, Socibot and Byrun are their most prominent robot creations.

“We have pre-programmed every single thing the robot says and every single thing the robot does — all the moves. There’s about nearly 400 separate queues but they are made up of other files, all stuck together so there’s probably a couple of thousand cues in reality. So the robot will always say the same thing and move the same way, depending on what queue is been triggered at what particular time.”

This promotional video for the play is well worth watching:

Amongst all the activity in autonomously driven vehicle joint ventures, new R&D facilities, strategic acquisitions (such as Mobileye being acquired by Intel) and booming startup fundings, two big players in the industry, NVIDIA and Bosch, are partnering to develop an AI self-driving car supercomputer.

Bosch CEO Dr Volkmar Denner announced the partnership during his keynote address at Bosch Connected World, in Berlin.

“Automated driving makes roads safer, and artificial intelligence is the key to making that happen,” said Denner. “We are making the car smart. We are teaching the car how to maneuver through road traffic by itself.”

The Bosch AI car computer will use NVIDIA AI PX technology, the upcoming AI car superchip, advertised as the world’s first single-chip processor designed to achieve Level-4 autonomous driving (see ADAS chart). The unprecedented level of performance is necessary to handle the massive amount of computation required for the various tasks self-driving vehicles must perform which include running deep neural nets to sense surroundings, understanding the 3D environment, localizing themselves on an HD map, predicting the behavior and position of other objects, as well as computing car dynamics and a safe path forward.

Source: Frost & Sullivan;VDS Automotive SYS Konferenz 2014/

Essentially, the NVIDIA platform enables vehicles to be trained on the complexities of driving, operated autonomously and updated over the air with new features and capabilities. And Bosch, which is the one of the world’s largest auto parts makers, has the Tier 1 credentials to mass-produce this AI-enabled supercomputer for a good portion of the auto industry.

“Self-driving cars is a challenge that can finally be solved with recent breakthroughs in deep learning and artificial intelligence,” said Jen-Hsun Huang, founder and CEO, NVIDIA. “Using DRIVE PX AI car computer, Bosch will build automotive-grade systems for the mass production of autonomous cars. Together we will realize a future where autonomous vehicles make mobility safe and accessible to all.”

Nvidia is also partnering with automakers Audi and Mercedes-Benz.

Bottom line:

“This is the kind of strategic tie-up that lets both partners do what they do best – Nvidia can focus on developing the core AI supercomputing tech, and Bosch can provide relationships and sales operations that offer true scale and reach,” says Darrell Etherington for TechCrunch.

Whether an A.I. ought to be granted patent rights is a timely question given the increasing proliferation of A.I. in the workplace. Examples: Daimler-Benz has tested self-driving trucks on public roads^[1], A.I. technology has been applied effectively in medical advancements, psycholinguistics, tourism and food preparation,^[2] a film written by an A.I. recently debuted online^[3] and A.I. has even found its way into the legal profession,^[4] and current interest in the question of whether an A.I. can enjoy copyright rights with several articles having already being published on the subject of A.I. and copyright rights.^[5]

In 2014 the U.S. Copyright Office updated its Compendium of U.S. Copyright Office Practices with, inter alia, a declaration that  the Office will not register works produced by a machine or mere mechanical process that operates randomly or automatically without any creative input or intervention from a human author.”^[6]

To grant or not to grant: A human prerequisite?

One might argue that Intellectual Property (IP) laws and IP Rights were designed to exclusively benefit human creators and inventors^[7] and thus would exclude non-humans from holding IP rights. The U.S. Copyright Office’s December 2014 update to the Compendium of U.S. Copyright Office Practices that added requirements for human authorship^[8] certainly adds weight to this view.

However, many IP laws were drafted well before the emergence of A.I. and in any case, do not explicitly require that a creator or inventor be ‘human.’ The World Intellectual Property Organization’s (WIPOs) definition of Intellectual Property talks about creations of the mind^[9] but does not specify whether it must be a human mind. Similarly, provisions in laws promoting innovation and IP rights, such as the so-called Intellectual Property Clause of the U.S. Constitution^[10], also do not explicitly mention a ‘human’ requirement. Finally, it ought to be noted that while the U.S. Copyright Office declared it would not register works produced by a machine or mere mechanical process without human creative input, it did not explicitly state that an A.I. could not have copyright rights.^[11]

Legal personhood

One might also argue that an A.I. is not human, and is therefore not a legal person and, thus, is not entitled to apply for much less be granted a patent. New Zealand’s Patents Act, for example, refers to a patent ‘applicant’ as a ‘person’.^[12]

Yet this line of argument could be countered by an assertion that a legal ‘person’ need not be ‘human’ as is the case of a corporation and there are many examples of patents assigned to corporations.^[13]

The underlying science

To answer the question of patent rights for an A.I. we need to examine how modern A.I. systems work and, as an example, consider how machine translation applications such as Google Translate function.

While such systems are marketed as if they’re “magic brains that just understand language”,^[14]the problem is that there is currently no definitive scientific description for language^[15] or language processing^[16]. Thus, such language translation systems cannot function by mimicking the processes of the brain.

Rather, they employ a scheme known as Statistical Machine Translation (SMT) whereby online systems search the Internet identifying documents that have already been translated by human translators– for example books, and organizations like the United Nations, or websites. The system scans these texts for statistically significant patterns and once the computer finds a pattern it uses the pattern to translate similar text in the future.^[17] This, as Jaron Lanier and others note, means that the people who created the translations and make translation systems possible are not paid to for their contributions^[18].

Many modern A.I. systems are largely big data models that operate by defining a real world problem that needs to be solved, then conceiving a conceptual model to solve this problem which is typically a statistical analysis that falls into one of three categories: regression, classification or missing data. Data is then fed into the model and used to refine and calibrate the model. As the model is increasingly refined it is used to guide the location of data and, after a number of rounds of refinement finally results in a model capable of some predictive functionality.^[19]

Big data models can be used to discover patterns in large data sets^[20] but also can, as in the case of translation systems, exploit statistically significant correlations in data.

None of this, however, suggests that current A.I. systems are capable of inventive or creative capacity.

Patentability?

So to get a patent, an invention must:

• Be novel in that it does not form part of the prior art^[21]
• Have an inventive step in that it not obvious to a person skilled in the art^[22]
• Be useful^[23]
• it must not fall into an excluded category that can include discoveries, presentations of information and mental processes or rules or methods for performing a mental act.^[24]

Why discoveries are not inventions is tied with the issue of obviousness and as noted by Buckley J. in Reynolds v. Herbert Smith & Co., Ltd^[25] who stated:

“Discovery adds to the amount of human knowledge, but it does so only by lifting the veil and disclosing something which before had been unseen or dimly seen. Invention also adds to human knowledge, but not merely by disclosing something. Invention necessarily involves also the suggestion of an act to be done, and it must be an act which results in a new product, or a new result, or a new process, or a new combination for producing an old product or an old result.”

Therefore in order to get a patent, an A.I. must first be capable of producing a patentable invention but, given current technology, is this even possible?

A thought exercise

Consider the following:

You believe that as a person exercises more, he/she consumes more oxygen and have tasked your A.I. with analyzing the relationship between oxygen consumption and exercise.

You provide the A.I. with a model suggesting that oxygen consumption increases with physical exertion and data that shows oxygen consumption among people performing little, moderate (e.g. walking briskly) and heavy exercise (e.g. running).

The A.I. reviews the data, refines the model, collects more data and comes up with a predictive model (e.g. when a person exercises X amount, he/she consumes Y amount of oxygen and when the person doubles his/her exertion, his oxygen consumption rate triples).

As this is essentially a statistical regression, the model will not always completely accurate in its predictions due to differences between individuals (i.e. for some persons the model will predict oxygen consumption fairly accurately, for others its results will be far off).

However, this particular model has another, more fundamental limitation – it fails to consider that a human cannot exercise beyond a certain point because his/her heart would be incapable of sustaining such levels of exertion^[26] or because over-exercise may trigger an unexpected reaction (e.g. death).^[27]

If one were to feed this model data of persons who have collapsed or died during exercise (and thus, in the latter case, not consume any oxygen), would the A.I. be able to ‘think outside its box’ and:

• Question the cause of these data discrepancies and have the initiative to conduct further investigation?
• Note and correct the limitation in the original model (which would require a significant amendment)?

Or would it simply alter the existing model by changing the slope of the regression line?

SMT and other A.I. have similar limitations, in the case of SMT, once the system is built, linguistic knowledge becomes necessary to achieve perfect translation at all grammatical levels^[28] and SMT systems presently cannot translate cultural components of the source text into the target language, provide very literal, word for word translations that do not recognize idioms, slang, and terms that are not in the machine’s memory and lack human creativity^[29] To do so would require a change to the underlying machine translation model, and the question arises whether this would have to be done by the human creators of the SMT or whether the SMT itself would be able to make the necessary corrections and adjustments to the model.

Should the SMT or, in the earlier example the A.I., be unable to improve and, in this case, innovate on the existing model does it have the creative or inventive capacity to conceive an invention is truly inventive? And if either the SMT or the AI can produce something that appears novel and inventive, given the nature of how A.I. presently operates (i.e. as big data models), would such a product be the result of an analysis of existing data to uncover hitherto unseen relationships – in other words, a discovery?

Returning to the original question about patent rights for an A.I., perhaps the question we should ask is not whether an A.I. should be able to get a patent, but whether an A.I., given current technology, can create a patentable invention in the first place and if the answer to that question is ‘no’, then the question of granting patent rights to an A.I. is moot.

References:

[1]  Jon Fingas, ‘Daimler tests a self-driving, mass-produced truck on real roads’, Engadget, Oct. 4, 2016.

[2] Sophie Curtis, ‘Cognitive Cooking: How Is A.I. Changing Foodtech?’ RE•WORK, April 19 2016.

[3] Analee Newitz, ‘Movie written by algorithm turns out to be hilarious and intense’ ArsTechnica, June 9, 2016.

[4]  Susan Beck, ‘AI Pioneer ROSS Intelligence Lands Its First Big Law Clients, The American Lawyer, May 6, 2016.

[5] See for example:

• Madeleine de Cock Buning ‘Is the EU exposed on the copyright of robot creations?‘ www.roboticslawjournal.com,Nov. 2015.
• Daniela Hernandez, ‘Could you take a robot to court if it steals your work? Fusion, Nov. 11, 2015.

[6] Copyright Office, Compendium of U.S. Copyright Office Practices (3d ed. 2014). § 313.2

[7] Hettinger argued that ‘the most powerful intuition supporting property rights is that people are entitled to the fruits of their labor’

• See: Edwin Hettinger, “Justifying Intellectual Property’, Philosophy & Public Affairs, Vol. 18, No. 1, Winder 1989, p. 31-52

[8] Copyright Office, Compendium of U.S. Copyright Office Practices (3d ed. 2014). § 306.

[9] According to WIPO, ‘Intellectual property (IP) refers to creations of the mind, such as inventions; literary and artistic works; designs; and symbols, names and images used in commerce.’

[10] Article I, Section 8, Clause 8 of the United States Constitution states: The Congress shall have Power To…. promote the Progress of Science and useful Arts, by securing for limited Times to Authors and Inventors the exclusive Right to their respective Writings and Discoveries.

[11] According to the U.S. Copyright Office, ‘Copyright exists from the moment the work is created. You will have to register, however, if you wish to bring a lawsuit for infringement of a U.S. work’

See ‘Copyright in General’,

[12] s.5(1) Patents Act 2013

[13] See for example, U.S. Patent 5,953,441 ‘Fingerprint sensor having spoof reduction features and related methods’ which was assigned to Harris Corporation.

[14] Nilagia McCoy  ‘Jaron Lanier: The Digital Economy Since Who Owns The Future?’ October 8, 2015.

[15] This was noted by Jaron Lanier who delivered the keynote address at the opening of the Conference on the Global Digital Content Market taking place from April 20-22, 2016 at WIPO Headquarters in Geneva, Switzerland.

• Catherine Saez, ‘IP Should Serve More Actors In New Ways, Keynote Speaker At WIPO Says’, Intellectual Property Watch, April 21, 2016.

[16] See for example:

• Erin Harte, ‘Language Processing In The Human Brain’, Brainworld, July 22, 2011.
• AD Friederici, The brain basis of language processing: from structure to function, Physiol Rev. 2011 Oct;91(4):1357-92.
• Jerry Ball, Mary Freiman, Stuart Rodgers & Christopher Myers, ‘Toward a Functional Model of Human Language Processing.’

[17] Inside Google Translate, July 9, 2010.

[18] Catherine Jewell, ‘Digital pioneer, Jaron Lanier, on the dangers of “free” online culture,’ WIPO Magazine, April 2016.

[19] Noah Silverman.

[20] i.e. ‘data mining’.

[21] s 6, New Zealand Patents Act 2013

[22] s 7, New Zealand Patents Act 2013

[23] s 10, New Zealand Patents Act 2013

[24] See for example:. 93(2) Hong Kong Patents Ordinance, Cap 514 or the Canadian Patent Act R.S.C., 1985, c. P-4

[25] (1903), 20 R.P.C. 123

[26] Wikipedia, ‘Heart Rate’.

[27] This may be caused by a congenital condition but some also believe that prolonged excessive exercise may exacerbate matters. See:

• Lou Schuler, ‘Death by Exercise’ Men’s Health June 8, 2003.
• Timothy D Noakes, ‘Sudden Death and Exercise’.

[28] Mireia Farrus, Marta R. Costa-Jussa, Jose B. Marino, Marc Poch, Adolfo Hernandez, Carlos Henrıque,  Jose A. R. Fonollosa, ‘Overcoming statistical machine translation limitations: error analysis and proposed solutions for the Catalan–Spanish language pair’ Language Resources & Evaluation, DOI 10.1007/s10579-011-9137-0.

[29] ‘Machine Translation vs. Human Translators’ 

See also:

• Wikipedia, ‘Statistical Machine Translation’.

A quick, hassle-free way to stay on top of robotics news, our robotics digest is released on the first Monday of every month. Sign up to get it in your inbox.

Big Ideas

February is only just gone, and already 2017 is shaping up to be a year full of big ideas and ambitions. The Future of Life Institute, for example, just published the Asilomar AI principles: 23 guidelines to ensure AI developments are beneficial to humanity. They are calling for shared responsibility and caution against an AI arms race. The principles have already been supported by more than 700 researchers, including the likes of Elon Musk, Demis Hassabis and Stephen Hawking.

But February wasn’t just about making new rules, it was also about breaking the mould. In Switzerland, Kickstarter Accelerator launched its second programme, which will help 30 selected international start-ups develop their ideas and products with the ultimate goal of disrupting one of the programme’s chosen verticals to benefit wider society.

Image: Phillipe Rossier

And in Dubai, big plans for passenger drones to carry people through the city’s skies, as early as July this year, were announced. Meanwhile in the UK, the public is invited to take a look back at the history of humanoid robots from the very (very) early days to the present day, and to ask the critical questions, such as ‘will robots bring us closer together or drive us apart’ and ‘how will humanoids impact on our lives’?

Robots – paying their fair share?

Bill Gates suggested robots that are being used to replace human workers should ‘pay taxes’ to ensure there isn’t a gap in tax income due to increased automation. While that might sound like a good idea, the European Parliament this month already rejected a robot tax, instead, calling for new legislation to regulate the rise of robots.

Going mobile

One of the areas in which robots are most definitely on the rise, and could soon be competing with their human counterparts, is in delivery services. Delivery robots are becoming more reliable and more mobile, and many big firms have announced plans to employ some form of a delivery robot in the near future. No wonder then that the states of Georgia and Virginia have brought in legislation for delivery robots and driverless vehicles.

On the innovation front, Agility Robotics showed off their ostrich-inspired robot, which has the potential to be deployed over uneven terrain in disaster relief, and even package delivery (including up steps!). Similarly, Boston Dynamics’ latest addition to its family, 6.5ft tall ‘Handle’, has wheels and legs/arms in a clever combo that enables the robot to be incredibly agile, lift and carry boxes, and even tackle stairs. And Italian company Vespa, most famous for its iconic scooters, announced plans for a small robot called Gita that will carry your groceries for you.

But robots that carry and deliver things aren’t the only mobile robots on the block. The TERESA project is developing semi-autonomous, mobile telepresence robots to enhance social interactions. The robots can autonomously perform low-level behaviours, such as navigating their way through groups of people and maintaining a socially appropriate distance, so that the remote user can focus on high-level decisions such as who to talk to and what to say.

Going nuclear

In Japan, small mobile robots are being used to look inside the stricken Fukushima Daiichi nuclear power plant following the 2011 tsunami disaster. The latest to take on this task was a scorpion-shaped robot developed by Toshiba Corp.

But while the robot probe did manage to transmit some data, it failed to provide vital information about the location of suspected melted fuel, which is urgently needed to inform clean-up activities at the site. The robot’s left roller-belt failed before it could reach a grate that might have allowed a view of the leaked fuel. Structural damage and higher than expected radiation mean that researchers now need to go back to the drawing board in order to design robots better suited for the harsh environment of the reactors.

The Zwentendorf Nuclear Power Plant in Austria was completely built but never put into operation. Reactor supervision and decommissioning is one important operational area for robots in the RN domain.

With the lessons learnt from Fukushima and the earlier nuclear disaster in Chernobyl, the European Reference Network for Critical Infrastructure Protection (ERNCIP) has established a group to specifically focus on radiological and nuclear threats, developing protocols and response procedures, including some on the use of robots in various scenarios involving radiation.

Decorative drones and the Super Bowl

On a lighter note, viewers of the Super Bowl on 5 February watched a fabulous light show, thanks to Lady Gaga and a swarm of 300 ‘Shooting Star’ Intel drones that created an airborne image of Old Glory. What most of us probably didn’t realise at the time was that a 34.5 mile radius drone ban during the Super Bowl, put in place by the FAA, meant that the light show had actually been filmed earlier, and was sync’d into Lady Gaga’s performance later.

Lady Gaga performing at Super bowl LI Source: NFL/YouTube

And if you missed that, don’t despair. There’s plenty more drone-related fun to be had. Drone racing, for example? Australian company Freedom Drone Sports is working on a bigger, better drone at the moment, which could soon be seen at ‘giant drone racing’ events. That sounds pretty fun.

And how about drone art? There are plenty of weird and wonderful ways in which drones are being used in the arts, including theatre and musical shows. And at the Imperial War museum in London, visitors can now admire the work of Mahwish Chishty, setting Pakistani folk art against America’s drone warfare.

Military drones

Drone warfare is of course not primarily a matter of art at all. In fact, it’s a very real threat for millions of people. The US has been using drones in conflicts in the Middle East for years. And in Mosul,  the so-called Islamic State recently deployed drones against the Iraqi army and Kurdish forces.

But drone threats are not just relevant in the Middle East. France reported that it’s  monitoring its airspace for any suspicious activity from foreign aircraft and drones, and has been running simulations of potential drone threats since the 2015 Paris attacks.

In the US, the army is keen to integrate an electronic warfare system into its Gray Eagle drones, while the Navy scrapped its unmanned warfare office. And China is planning to use a giant solar-powered drone in the mesosphere as an early warning system, as well as for weather and disaster surveillance, and communication.

Regulating robotics

When it comes to domestic drone use, by companies such as Amazon or individual drone enthusiasts, February was buzzing with regulation chatter, but with few resulting decisions. The long-awaited framework for drone flight over people was delayed and is now caught up in President Trump’s temporary stop on new regulations, meaning it may be some time before there’s more clarity on how drones can and can’t be used near people.

In San Diego, regulations to allow local agencies to enforce safe drone operations were proposed by a working group made up of representatives from emergency services and other agencies. The regulations were passed by a City Council sub-committee, but needs to be voted on by the full Council before becoming law. And a bill barring the use of drones near state prisons and county jails was passed in Senate, but still needs to go to the S.C. House of Representatives before taking effect.

Drones are, of course, not the only type of robot that requires new regulations and laws. In a popular video (above) Polo Sant’Anna Valdera explored legal and ethical challenges of robot regulation, and posed the question: ‘What legal rights should robots have’?

Robocars

It seems every month for the last few years there’s at least one big story on autonomous cars, and February 2017 was no exception. This month, California published a summary of robocar testing in the state, showing that Google/Waymo left its competition far (far) behind by doing around 30 times as much test driving as everyone else put together!

But while Google is clearly in the lead when it comes to distance, Drive.ai released a video showing its self-driving car navigating the rainy streets of the Bay Area, in the middle of night, with ease. This is not a small feat considering that the car’s sensors had to deal with reflected lights from the wet road surface, congestion, and a broken traffic light.

Away from the sunshine state, the self-driving car industry is still booming. Car giant Ford, for example, just invested $1 billion in Pittsburgh based start-up Argo AI, becoming the majority shareholder.

But it’s not just good news and easy rides for the sector. Some have expressed concerns over the risk of hacking and cyber security for autonomous cars. This is a problem that hasn’t been fully resolved by manufacturers yet, and will need more research.

And in a worrying yet slightly amusing twist, companies are finding that their test engineers are falling asleep during test drives. Turns out rides are just too smooth. And with nothing to do, it engineers can’t help but doze off. This could spell trouble for semi-autonomous cars that require human monitoring and intervention, such as those planned by BMW, Mercedes-Benz and Audi.

Artificial intelligence

February 2017 was a pretty exciting month for those interested in AI. At MIT’s CSAIL lab, they’re working on a  wearable AI that can detect the tone of a conversation and could help people with social anxiety or Asperger’s syndrome understand social interactions better. And Kickstarter campaign  Bots_alive smashed its crowdfunding target, and will offer an app that’ll give Hexbug spider robots a bit of character via machine learning.

Samsung Simband. Image: Jason Dorfman, MIT CSAIL

But it’s not all about developing nifty consumer products: Discussions this month highlighted the potential for  AI to manage the UK’s national power grid at a time when increasing supply by renewables has to be supplemented by occasional boosts from other sources due to variable demand.

And a team at MIT  improved automated planners, with potential applications in flight scheduling and optimisation, by giving them some human intuition, winning the International Conference on Automated Planning and Scheduling’s International Planning Competition.

Finally, Facebook is planning to use  AI to monitor its content for terrorist propaganda materials in a bid to be less dependent on users reporting offensive or worrying content.

Money

Talking of Facebook, the social media giant was also made the news this February when it was told to  pay game development firm ZeniMax $500 million in a lawsuit pertaining to Oculus Rift, the virtual reality headset.

In other business news, the International Federation of Robotics forecast  massive growth for the robotic cleaning market, with vacuums clearly in the lead for most sought-after product. But other must-have appliances, such as automated mopping devices, are also increasingly becoming available.

iRobot Braava jet 240. Source: irobot.com

And finally, mobile phone and network provider  Verizon bought drone operations and management company Skyward.

Automation – the end of the human workforce?

The debate about robots replacing workers is almost as old as the field of robotics itself. With new automation technologies coming on the market almost daily, the debate is perhaps even more pressing than ever.

For those who are concerned about increasing automation, the news that internet giant Amazon plans to construct massive two-story supermarkets that could, allegedly, be run by just three human staff, won’t be well received. Similarly, with increasing automation in the farming sector, some may worry that the role of the farmer will never be the same. And with AI technologies becoming ever more intelligent, some have now started talking about  AI-based legal systems.

But despite all these developments, the contribution of human workers will be vital for many years to come, and although robots might replace human workers in some positions, they’re also likely to create new jobs.

Bio-bots and new tech in action

And to wrap up February’s digest, there are also quite a few cool videos that we just didn’t want to keep from you. DARPA showed off their  robot arm as it catches a UAV mid-flight, engineers at MIT developed a  gel robot that can catch (and not hurt) live fish, researchers at CalTech and the University of Illinois Urbana-Champaign  made a flying bat bot, a team of students  designed the NAST OpenHand M2S, and  computer simulations of the running gaits of the fruit fly helped us to understand how robots might walk and climb better.

And if you have a dog, you’ll probably quite like 2015’s Robot Launch start-up winner CleverPet, a robot that’ll  keep your dog entertained while you’re away (or sat on the couch watching TV).

Teach your bot

Last but not least, we’ve published our  educational tutorials on Python programming, which will help you begin programming your NAO. There’s videos, step-by-step guides and lots of helpful hints and tips – so why not have a go?

Upcoming events for March – April 2017

Conf on Human-Robot Interaction: March 6-9, Vienna, Austria.

Age of Drones Expo: March 14-15, Hamburg, Germany.

European Robotics Forum: March 22-24, Edinburgh, Scotland.

Automate: April 3-6, Chicago, IL.

Unmanned Security Expo: April 4-7, Las Vegas, NV.

Robobusiness Europe: April 19-21, The Hague, Netherlands.

ICCAR: April 22-24, Nagoya, Japan.

Artificial intelligence (AI) already plays a major role in human economies and societies, and it will play an even bigger role in the coming years. To ponder the future of AI is thus to acknowledge that the future is AI.

This will be partly owing to advances in “deep learning,” which uses multi-layer neural networks that were first theorized in the 1980s. With today’s greater computing power and storage, deep learning is now a practical possibility, and a deep-learning application gained worldwide attention in 2016 by beating the world champion in Go. Commercial enterprises and governments alike hope to adapt the technology to find useful patterns in “Big Data” of all kinds.

In 2011, IBM’s Watson marked another AI watershed, by beating two previous champions in Jeopardy!, a game that combines general knowledge with lateral thinking. And yet another significant development is the emerging “Internet of Things,” which will continue to grow as more gadgets, home appliances, wearable devices, and publicly-sited sensors become connected and begin to broadcast messages around the clock. Big Brother won’t be watching you, but a trillion little brothers might be.

Image: CB Insights

From Jeopardy winners and Go masters to infamous advertising-related racial profiling, it would seem we have entered an era in which artificial intelligence developments are rapidly accelerating. But a fully sentient being whose electronic “brain” can fully engage in complex cognitive tasks using fair moral judgement remains, for now, beyond our capabilities.

Unfortunately, current developments are generating a general fear of what artificial intelligence could become in the future. Its representation in recent pop culture shows how cautious – and pessimistic – we are about the technology. The problem with fear is that it can be crippling and, at times, promote ignorance.

Learning the inner workings of artificial intelligence is an antidote to these worries. And this knowledge can facilitate both responsible and carefree engagement.

The core foundation of artificial intelligence is rooted in machine learning, which is an elegant and widely accessible tool. But to understand what machine learning means, we first need to examine how the pros of its potential absolutely outweigh its cons.

Data is the key

Simply put, machine learning refers to teaching computers how to analyse data for solving particular tasks through algorithms. For handwriting recognition, for example, classification algorithms are used to differentiate letters based on someone’s handwriting. Housing data sets, on the other hand, use regression algorithms to estimate in a quantifiable way the selling price of a given property.Machine learning, then, comes down to data. Almost every enterprise generates data in one way or another: think market research, social media, school surveys, automated systems. Machine learning applications try to find hidden patterns and correlations in the chaos of large data sets to develop models that can predict behaviour.

Credit: Jonathan Khoo/Flickr, CC

Machine learning, then, comes down to data. Almost every enterprise generates data in one way or another: think market research, social media, school surveys, automated systems. Machine learning applications try to find hidden patterns and correlations in the chaos of large data sets to develop models that can predict behaviour.

Data have two key elements – samples and features. The former represents individual elements in a group; the latter amounts to characteristics shared by them.

Look at social media as an example: users are samples and their usage can be translated as features. Facebook, for instance, employs different aspects of “liking” activity, which change from user to user, as important features for user-targeted advertising.

Facebook friends can also be used as samples, while their connections to other people act as features, establishing a network where information propagation can be studied.

My Facebook friends network: each node is a friend who might or might not be connected to other friends. The larger the node, the more connections one has. Similar colours indicate similar social circles. https://lostcircles.com/

Outside of social media, automated systems used in industrial processes as monitoring tools use time snapshots of the entire process as samples, and sensor measurements at a particular time as features. This allows the system to detect anomalies in the process in real time.

All these different solutions rely on feeding data to machines and teaching them to reach their own predictions once they have strategically assessed the given information. And this is machine learning.

Human intelligence as a starting point

Any data can be translated into these simple concepts and any machine-learning application, including artificial intelligence, uses these concepts as its building blocks.

Once data are understood, it’s time to decide what do to with this information. One of the most common and intuitive applications of machine learning is classification. The system learns how to put data into different groups based on a reference data set.

This is directly associated with the kinds of decisions we make every day, whether it’s grouping similar products (kitchen goods against beauty products, for instance), or choosing good films to watch based on previous experiences. While these two examples might seem completely disconnected, they rely on an essential assumption of classification: predictions defined as well-established categories.

When picking up a bottle of moisturiser, for example, we use a particular list of features (the shape of the container, for instance, or the smell of the product) to predict – accurately – that it’s a beauty product. A similar strategy is used for picking films by assessing a list of features (the director, for instance, or the actor) to predict whether a film is in one of two categories: good or bad.

By grasping the different relationships between features associated with a group of samples, we can predict whether a film may be worth watching or, better yet, we can create a program to do this for us.

But to be able to manipulate this information, we need to be a data science expert, a master of maths and statistics, with enough programming skills to make Alan Turing and Margaret Hamilton proud, right? Not quite.

You don’t have to be Alan Turing to have a go at machine learning. Credit: CyberHades/Flickr, CC

We all know enough of our native language to get by in our daily lives, even if only a few of us can venture into linguistics and literature. Maths is similar; it’s around us all the time, so calculating change from buying something or measuring ingredients to follow a recipe is not a burden. In the same way, machine-learning mastery is not a requirement for its conscious and effective use.

Yes, there are extremely well-qualified and expert data scientists out there but, with little effort, anyone can learn its basics and improve the way they see and take advantage of information.

Algorithm your way through it

Going back to our classification algorithm, let’s think of one that mimics the way we make decisions. We are social beings, so how about social interactions? First impressions are important and we all have an internal model that evaluates in the first few minutes of meeting someone whether we like them or not.

Two outcomes are possible: a good or a bad impression. For every person, different characteristics (features) are taken into account (even if unconsciously) based on several encounters in the past (samples). These could be anything from tone of voice to extroversion and overall attitude to politeness.

For every new person we encounter, a model in our heads registers these inputs and establishes a prediction. We can break this modelling down to a set of inputs, weighted by their relevance to the final outcome.

For some people, attractiveness might be very important, whereas for others a good sense of humour or being a dog person says way more. Each person will develop her own model, which depends entirely on her experiences, or her data.

Different data result in different models being trained, with different outcomes. Our brain develops mechanisms that, while not entirely clear to us, establish how these factors will weigh out.

What machine learning does is develop rigorous, mathematical ways for machines to calculate those outcomes, particularly in cases where we cannot easily handle the volume of data. Now more than ever, data are vast and everlasting. Having access to a tool that actively uses this data for practical problem solving, such as artificial intelligence, means everyone should and can explore and exploit this. We should do this not only so we can create useful applications, but also to put machine learning and artificial intelligence in a brighter and not so worrisome perspective.

There are several resources out there for machine learning although they do require some programming ability. Many popular languages tailored for machine learning are available, from basic tutorials to full courses. It takes nothing more than an afternoon to be able to start venturing into it with palpable results.

All this is not to say that the concept of machines with human-like minds should not concern us. But knowing more about how these minds might work will gives us the power to be agents of positive change in a way that can allow us to maintain control over artificial intelligence and not the other way around.

This article was originally published on The Conversation. Read the original article.

If you liked this article, you may also want to read:

• Building character AI through machine learning
• What is machine learning?
• Brian Cox presents Science Matters: Machine learning and artificial intelligence
• Talking Machines: Machine learning and the Flint water crisis, with Jake Abernethy
• Talking Machines: Computational learning theory and machine learning for understanding cells, with Aviv Regev

See all the latest robotics news on Robohub, or sign up for our weekly newsletter.

Science fiction has already explored the theme of robot rights, such as the film Bicentennial Man. Image: Columbia Pictures

Science fiction likes to depict robots as autonomous machines, capable of making their own decisions and often expressing their own personalities. Yet we also tend to think of robots as property, and as lacking the kind of rights that we reserve for people.

But if a machine can think, decide and act on its own volition, if it can be harmed or held responsible for its actions, should we stop treating it like property and start treating it more like a person with rights?

What if a robot achieves true self-awareness? Should it have equal rights with us and the same protection under the law, or at least something similar?

These are some of the issues being discussed by the European Parliament’s Committee on Legal Affairs. Last year it released a draft report and motion calling for a set of civil law rules on robotics regulating their manufacture, use, autonomy and impact upon society.

Of the legal solutions proposed, perhaps most interesting was the suggestion of creating a legal status of “electronic persons” for the most sophisticated robots.

Approaching personhood

The report acknowledged that improvements in the autonomous and cognitive abilities of robots makes them more than simple tools, and makes ordinary rules on liability, such as contractual and tort liability, insufficient for handling them.

For example, the current EU directive on liability for harm by robots only covers foreseeable damage caused by manufacturing defects. In these cases, the manufacturer is responsible. However, when robots are able to learn and adapt to their environment in unpredictable ways, it’s harder for a manufacturer to foresee problems that could cause harm.

The report also questions about whether or not sufficiently sophisticated robots should be regarded as natural persons, legal persons (like corporations), animals or objects. Rather than lumping them into an existing category, it proposes that a new category of “electronic person” is more appropriate.

The report does not advocate immediate legislative action, though. Instead it proposes that legislation be updated if robots become more complex; if and when they develop more behavioural sophistication. If this occurs, one recommendation is to reduce the liability of “creators” proportional to the autonomy of the robot, and that a compulsory “no-fault” liability insurance could cover the shortfall.

But why go so far as to create a new category of “electronic persons”? After all, computers still have a long way to go before they match human intelligence, if they ever do.

But it can be agreed that robots – or more precisely the software that controls them – is becoming increasingly complex. Autonomous (or “emergent”) machines are becoming more common. There are ongoing discussions about the legal liability for autonomous vehicles, or whether we might be able to sue robotic surgeons.

These are not complicated problems as long as liability rests with the manufacturers. But what if manufacturers cannot be easily identified, such as if open source software is used by autonomous vehicles? Whom do you sue when there are millions of “creators” all over the world?

Artificial intelligence is also starting to live up to its moniker. Alan Turing, the father of modern computing, proposed a test in which a computer is considered “intelligent” if it fools humans into believing that the computer is human by its responses to questions. Already there are machines that are getting close to passing this test.

There are also other incredible successes, such as the computer that creates soundtracks to videos that are indistinguishable from natural sounds, the robot that can beat CAPTCHA, one that can create handwriting indistinguishable from human handwriting and the AI that recently beat some of the world’s best poker players.

Robots may eventually match human cognitive abilities and they are becoming increasingly human-like, including the ability to “feel” pain.

If this progress continues, it may not be long before self-aware robots are not just a product of fantastic speculation.

The EU report is among the first to formally consider these issues, but other countries are also engaging. Peking University’s Yueh-Hsuan Weng writes that Japan and South Korea expect us to live in a human-robot coexistence by 2030. Japan’s Ministry of Economy, Trade and Industry has created a series of robot guidelines addressing business and safety issues for next generation robots.

Electronic persons

If we did give robots some kind of legal status, what would it be? If they behaved like humans we could treat them like legal subjects rather than legal objects, or at least something in between. Legal subjects have rights and duties, and this gives them legal “personhood”. They do not have to be physical persons; a corporation is not a physical person but is recognised as a legal subject. Legal objects, on the other hand, do not have rights or duties although they may have economic value.

Assigning rights and duties to an inanimate object or software program independent of their creators may seem strange. However, with corporations we already see extensive rights and obligations given to fictitious legal entities.

Perhaps the approach to robots could be similar to that of corporations? The robot (or software program), if sufficiently sophisticated or if satisfying certain requirements, could be given similar rights to a corporation. This would allow it to earn money, pay taxes, own assets and sue or be sued independently of its creators. Its creators could, like directors of corporations, have rights or duties to the robot and to others with whom the robot interacts.

Robots would still have to be partly treated as legal objects since, unlike corporations, they may have physical bodies. The “electronic person” could thus be a combination of both a legal subject and a legal object.

The European Parliament will vote on the resolution this month. Regardless of the result, reconsidering robots and the law is inevitable and will require complex legal, computer science and insurance research.

This article was originally published on The Conversation. Read the original article.

Enter artificial intelligence (AI)

There has been much hoopla about AI-based legal systems that, some might have you believe, may soon replace lawyers (no doubt causing a degree of anxiety among some legal professionals). There is some misunderstanding among many lawyers, and much of the public, about what AI systems are presently capable of. Can a legal AI, based on current technology, actually “think” like a lawyer? No. At best, today’s AI is an incomplete substitute for a human lawyer, although it could reduce the need for some lawyers (I’ll get to all that later).

However, something we should think seriously about right now is the long-term implication of the introduction of AI into the legal environment—notably the potential loss of legal wisdom.

Why doesn’t AI think like a human?

Let’s explore why AI doesn’t actually mimic the human brain. As an example, let’s look at automated translation systems such as those available from Google, Facebook or Microsoft. Such systems might appear to work the way human translators do, but what they actually do is match patterns derived from analyses of thousands, if not millions, of pages of text found on the web, employing a technology known as statistical machine translation. For instance, if such a system wants to know how to translate the English greeting “hello” into French, it scans English and French translations on the web, statistically analyses the correlations between “hello” and various French greetings, then comes to the conclusion that the French equivalent of “hello” is “bonjour”.

Current AI is good at this kind of pattern matching, but less so at cognition and deductive reasoning. Consider the human brain: not only does it store a large number of associations, and accesses useful memories (sometimes quickly, sometimes not), it also transforms sensory and other information into generalisable representations invariant to unimportant changes, stores episodic memories and generalises learned examples into understanding. These are key cognitive capabilities yet to be matched by current AI technology.

Thus, while present AI-based legal systems might analyse judicial decisions—for example, to help litigators gain insights to a judge’s behaviour or a barrister’s track record—they do so by scrutinising existing data to reveal patterns, and not by extrapolating from the content of those decisions the way an experienced human legal professional might.

The temptation to make redundant

As AI systems become more capable, the temptation grows to use such systems not only to supplement but also to eliminate the need for some personnel. An AI system weak in cognition but strong in pattern matching probably could not replace an experienced professional in terms of drawing inferences, deductive reasoning or combining different practice areas to arrive at more comprehensive solutions. However, it could perform certain tasks such as searching for patterns of words in documents for evidence gathering that have hitherto been delegated to lower level staff—such as paralegals, trainees, and junior associates—and do so better than any human could.

While one might argue that the introduction of AI systems will lighten the workload of legal professionals and thereby improve their quality of life, it also potentially diminishes the need for junior legal staff, which would only exacerbate the oversupply problem in the legal profession.

Shrink now, suffer later?

If fewer junior legal professionals are hired, this implies a smaller population of lower level staff, thus a smaller feeder pool for more senior positions. And, as more tasks are automated, this could deprive junior legal professionals of opportunities to gain important experience—ie, get their 10,000 hours. Will this result in fewer quality, experienced legal professionals in the future?

And the future of legal AI?

There are yet two more (albeit related) things to think about.

First: the development and maintenance of a good AI system requires both technical and legal competency. Put another way, a legal AI system programmed by systems experts ignorant in the law will be seriously, if not fatally, flawed. Thus, if we want to continue to develop more capable legal AI systems, good content providers—ie, good lawyers—will be needed.

Second: as laws, the legal business and social environments in their respective jurisdictions evolve, developments that might not have been anticipated just a few years earlier will emerge. Only the very best legal and other minds will be able to cope with some of these developments—and update the relevant legal AI systems accordingly. For example, when the US passed the Leahy-Smith America Invents Act (AIA) in 2011, it introduced new review procedures for existing patents with the intent of improving patent quality. It also had the effect of introducing several unintended consequences, including the use of such procedures by hedge funds to invalidate patents for the purpose of affecting the stock price of the companies holding the patents and the negative impact the AIA has had on inventors. Updating an AI system to properly incorporate these developments requires not only a deep understanding of US patent law but also a perspective on patents, finance and the impact of patent policy and procedures on innovation—something that can only really be appreciated after years of experience. Moreover, this is something that could not have been programmed into an AI system half a decade ago, and such content could probably not have been provided by a less capable, less experienced legal professional to an AI developer.

So what, if anything, can be done?

Sadly, there are no easy answers. Graduating fewer lawyers might alleviate the problem of oversupply, but would also result in unemployment at educational institutions.

While the government (or government-backed NGO) could establish some sort of training centre for under-employed junior lawyers, where these professionals could offer services pro bono to build their experience, this also smacks of government interference in the private practice market. But we need to start thinking of solutions now.  The introduction of AI into the legal profession and the potential prospect of putting more lawyers out of work could have profound implications for legal AI systems and the profession as a whole.

MIT Professor Regina Barzilay has struck up new research collaborations, drawn in MIT students, launched projects with local doctors, and begun empowering cancer treatment with the machine-learning insight that has already transformed many areas of modern life.
Photo: Lillie Paquette/School of Engineering

Computer scientist Regina Barzilay is working with MIT students and medical doctors in an ambitious bid to revolutionize cancer care. She is relying on a tool largely unrecognized in the oncology world but deeply familiar to hers: machine learning.

Barzilay, the Delta Electronics Professor of Electrical Engineering and Computer Science, was diagnosed with breast cancer in 2014. She soon learned that good data about the disease is hard to find. “You are desperate for information — for data,” she says now. “Should I use this drug or that? Is that treatment best? What are the odds of recurrence? Without reliable empirical evidence, your treatment choices become your own best guesses.”

Across different areas of cancer care — be it diagnosis, treatment, or prevention — the data protocol is similar. Doctors start the process by mapping patient information into structured data by hand, and then run basic statistical analyses to identify correlations. The approach is primitive compared with what is possible in computer science today, Barzilay says.

These kinds of delays and lapses (which are not limited to cancer treatment), can really hamper scientific advances, Barzilay says. For example, 1.7 million people are diagnosed with cancer in the U.S. every year, but only about 3 percent enroll in clinical trials, according to the American Society of Clinical Oncology. Current research practice relies exclusively on data drawn from this tiny fraction of patients. “We need treatment insights from the other 97 percent receiving cancer care,” she says.

To be clear: Barzilay isn’t looking to up-end the way current clinical research is conducted. She just believes that doctors and biologists — and patients — could benefit if she and other data scientists lent them a helping hand. Innovation is needed and the tools are there to be used.

Barzilay has struck up new research collaborations, drawn in MIT students, launched projects with doctors at Massachusetts General Hospital, and begun empowering cancer treatment with the machine learning insight that has already transformed so many areas of modern life.

Machine learning, real people

At the MIT Stata Center, Barzilay, a lively presence, interrupts herself mid-sentence, leaps up from her office couch, and runs off to check on her students.

She returns with a laugh. An undergraduate group is assisting Barzilay with a federal grant application, and they’re down to the wire on the submission deadline. The funds, she says, would enable her to pay the students for their time. Like Barzilay, they are doing much of this research for free, because they believe in its power to do good. “In all my years at MIT I have never seen students get so excited about the research and volunteer so much of their time,” Barzilay says.

At the center of Barzilay’s project is machine learning, or algorithms that learn from data and find insights without being explicitly programmed where to look for them. This tool, just like the ones Amazon, Netflix, and other sites use to track and predict your preferences as a consumer, can make short work of gaining insight into massive quantities of data.

Applying it to patient data can offer tremendous assistance to people who, as Barzilay knows well, really need the help. Today, she says, a woman cannot retrieve answers to simple questions such as: What was the disease progression for women in my age range with the same tumor characteristics?

What a machine can see

Working closely with collaborators Taghian Alphonse, chief of breast radiation oncology at Massachusetts General Hospital (MGH); Kevin Hughes, co-director of the Avon Comprehensive Breast Evaluation Center at MGH; and Constance Lehman, the chief of the breast imaging division at MGH, Barzilay intends to bring data science into clinical research nationwide. But first, she’s content with connecting her world with theirs.

Barzilay’s work in natural language processing (NLP) enables machines to search, summarize, and interpret textual documents, such as those about cancer patients in pathology reports. Using NLP tools, she and her students extracted clinical information from 108,000 reports provided by area hospitals. The database they’ve created has an accuracy rate of 98 percent. Next she wants to incorporate treatment outcomes into it.

For another study, Barzilay has developed a database that Hughes and his team can use to monitor the development of atypias, which help identify which patients are at risk of developing cancer later in life.

Machines are good at making predictions — “Why not throw all the information you have about a breast cancer patient into a model?” she says — but Barzilay is wary of having the recommendations arrive as highly complex, computational recommendations without explanation. Jointly with Tommi Jaakkola, a professor of electrical engineering and computer science at MIT, and graduate student Tao Lei, she is also developing interpretable neural models that can justify and explain the machine-based predictive reasoning.

Barzilay is also looking at how new tools can help do preventive work. Mammograms contain lots of information that may be hard for a human eye to decipher. Machines can detect subtle changes and are more capable of detecting low-level patterns. Jointly with Lehman and graduate student Nicolas Locascio, Barzilay is applying deep learning for automating analysis of mammogram data. As the first step, they are aiming to compute density and other scores currently derived by radiologists who manually analyze these images. Their ultimate goal is to identify patients who are likely to develop a tumor before it’s even visible on a mammogram, and also to predict which patients are heading toward recurrence after their initial treatment.

Ultimate success, Barzilay says, will involve drawing on computer science in unexpected ways, and pushing it in a variety of new health-related directions.

Outside her door, several of Barzilay’s students are talking ideas, hunching over laptops, and drinking coffee. An object set against the back wall resembles an odd coatrack. Guided by an idea from Taghian, six undergraduate students, led by graduate student Julian Straub, built a device that uses machine-learning to detect lymphedema, a swelling of the extremities that can be caused by the removal of or damage to lymph nodes as part of cancer treatment. It can be disabling and incurable unless detected early. Because of their high cost, these machines — lymphometers — are rare in the U.S.; very few hospitals have them.

Students have created an affordable version. And they hope to start testing this device at MGH in a couple of months. “These students are doing amazing work,” says Barzilay. “These innovations will make a really big difference. It is an entry point. There is so much to do. We are just getting started.”

If you liked this article, you may also be interested in:

• What is machine learning?
• Brian Cox presents Science Matters: Machine learning and artificial intelligence
• 5 global problems that AI could help us solve
• Open Bionics partners with NHS for feasibility study to develop hands for amputees

See all the latest robotics news on Robohub, or sign up for our weekly newsletter.

There’s a great deal of concern over artificial intelligence; what it means for our jobs, whether robots will one day replace us in the workplace, whether it will one day lead to robot wars. But current research projects show that artificial intelligence (AI) can also be used for the greater good. Here are five global problems that machine learning could help us solve.

1. Healthcare

One of the biggest benefits of AI is its ability to trawl through massive amounts of data in record time. This helps researchers pinpoint areas of focus for their own research.

For example, a recent ground-breaking discovery on the disease Amyotrophic Lateral Sclerosis (ALS), was made through a partnership between Barrow Neurological Institute and the artificial intelligence company IBM Watson Health.

IBM Watson, the artificial intelligence computer, reviewed thousands of pieces of research and was able to identify new genes linked to ALS.

“Traditional research tools are fast becoming inadequate to help data scientists and researchers keep pace with any global problems that AI could help us solve and find relevant insights among the now billions of documents which are spread all over the world,” said the company in a press release.

“The discovery gives ALS researchers new insights that will pave the way for the development of new drug targets and therapies to combat one of the world’s most devastating and deadly diseases.”

https://youtu.be/F-qBLH6EfR8

Another promising use for AI within healthcare is its ability to predict the outcome of drug treatments. For instance, cancer patients are often given the same drug, and then monitored to see the effectiveness of that drug. AI could use data to predict which patients could benefit from using a particular drug, providing a highly personalized approach, and saving valuable time and money.

Image: Bloomberg

2. Making driving safer

Despite the crashes involving self-driving cars that have hit the headlines this year, this area of AI could dramatically reduce deaths and injuries on our roads.

According to a report by Stanford University, not only will self-driving cars reduce traffic related deaths and injuries, but they could bring about changes in our lifestyles as well. We will have more time to work or entertain ourselves during commutes, and we may have more choice over where we base ourselves:

“The increased comfort and decreased cognitive load with self-driving cars and shared transportation may affect where people choose to live,” the report says.

3. Transforming how we learn

Earlier this year, students at Georgia Tech university in the US were startled to discover that their helpful teaching assistant had in fact been a robot all along. After initial teething problems, the robot started answering the students’ questions with 97% certainty.

The university designed the robot after their research showed that one of the main factors behind students dropping out is a lack of support.

People learn differently, at different speeds and with different starting points. Artificial intelligence could usher in a future where we all learn in a much more personalised way. But no education system in the world can afford a tutor for every child, so this is where AI might be able to step in. Artificial tutors, made to look and sound as much like humans as possible, could take the lead in delivering personalised education.

4. Help us be smarter about energy

Artificial intelligence could help us be smarter about our energy consumption. In fact, this is already happening.

Image: Google DeepMind

Google and other tech giants have enormous data centres that require a massive amount of energy to run the servers and keep them cool. Google has used its artificial intelligence platform Deep Mind to predict when its data centres will get too hot. Cooling systems are only activated when required. AI has saved Google around 40% in energy costs at its server farms.

5. Helping wildlife

As in the case of healthcare, being able to analyse massive amounts of data can transform wildlife conservation.

For instance, by tracking animal movements, we can see where they go, and what habitats we need to protect. This study uses computing power to figure out the best places to create wildlife corridors for wolverines and grizzly bears in Montana. Wildlife corridors are continuous areas of protected land that link zones of biological significance that the animals can use to move safely through the wilderness.

Challenges

Using artificial intelligence is not without its challenges, however. One of the biggest of these is – how do we keep the systems safe? Algorithms are based on data, so any change to that data will change the behaviour and outcomes.

“Almost anything bad you can think of doing to a machine-learning model can be done right now,” said one expert at a recent AI conference in Spain. “And defending it is really, really hard.”

This post was originally published on WEForum.org.

You might also enjoy the following articles:

• Why artificial intelligence could be key to future-proofing the grid
• Wearable AI that can detect the tone of a conversation
• Artificial intelligence and ethics: Who does the thinking?
• Soft exosuit economies: Understanding the costs of lightening the load
• MIT Media Lab to participate in $27 million initiative on AI ethics and governance

See all the latest robotics news on Robohub, or sign up for our weekly newsletter.

Researchers from MIT’s Computer Science and Artificial Intelligence Laboratory are trying to improve automated planners by giving them the benefit of human intuition. By encoding the strategies of high-performing human planners in a machine-readable form, they were able to improve the performance of competition-winning planning algorithms by between 10 and 15 percent on a challenging set of problems. Image: Jose-Luis Olivares/MIT

Every other year, the International Conference on Automated Planning and Scheduling hosts a competition in which computer systems designed by conference participants try to find the best solution to a planning problem, such as scheduling flights or coordinating tasks for teams of autonomous satellites.

On all but the most straightforward problems, however, even the best planning algorithms still aren’t as effective as human beings with a particular aptitude for problem-solving — such as MIT students.

Researchers from MIT’s Computer Science and Artificial Intelligence Laboratory are trying to improve automated planners by giving them the benefit of human intuition. By encoding the strategies of high-performing human planners in a machine-readable form, they were able to improve the performance of competition-winning planning algorithms by 10 to 15 percent on a challenging set of problems.

The researchers are presenting their results this week at the Association for the Advancement of Artificial Intelligence’s annual conference.

“In the lab, in other investigations, we’ve seen that for things like planning and scheduling and optimization, there’s usually a small set of people who are truly outstanding at it,” says Julie Shah, an assistant professor of aeronautics and astronautics at MIT. “Can we take the insights and the high-level strategies from the few people who are truly excellent at it and allow a machine to make use of that to be better at problem-solving than the vast majority of the population?”

The first author on the conference paper is Joseph Kim, a graduate student in aeronautics and astronautics. He’s joined by Shah and Christopher Banks, an undergraduate at Norfolk State University who was a research intern in Shah’s lab in the summer of 2016.

The human factor

Algorithms entered in the automated-planning competition — called the International Planning Competition, or IPC — are given related problems with different degrees of difficulty. The easiest problems require satisfaction of a few rigid constraints: For instance, given a certain number of airports, a certain number of planes, and a certain number of people at each airport with particular destinations, is it possible to plan planes’ flight routes such that all passengers reach their destinations but no plane ever flies empty?

A more complex class of problems — numerical problems — adds some flexible numerical parameters: Can you find a set of flight plans that meets the constraints of the original problem but also minimizes planes’ flight time and fuel consumption?

Finally, the most complex problems — temporal problems — add temporal constraints to the numerical problems: Can you minimize flight time and fuel consumption while also ensuring that planes arrive and depart at specific times?

For each problem, an algorithm has a half-hour to generate a plan. The quality of the plans is measured according to some “cost function,” such as an equation that combines total flight time and total fuel consumption.

Shah, Kim, and Banks recruited 36 MIT undergraduate and graduate students and posed each of them the planning problems from two different competitions, one that focused on plane routing and one that focused on satellite positioning. Like the automatic planners, the students had a half-hour to solve each problem.

“By choosing MIT students, we’re basically choosing the world experts in problem solving,” Shah says. “Likely, they’re going to be better at it than most of the population.”

Encoding strategies

Certainly, they were better than the automatic planners. After the students had submitted their solutions, Kim interviewed them about the general strategies they had used to solve the problems. Their answers included things like “Planes should visit each city at most once,” and “For each satellite, find routes in three turns or less.”

The researchers discovered that the large majority of the students’ strategies could be described using a formal language called linear temporal logic, which in turn could be used to add constraints to the problem specifications. Because different strategies could cancel each other out, the researchers tested each student’s strategies separately, using the planning algorithms that had won their respective competitions. The results varied, but only slightly. On the numerical problems, the average improvement was 13 percent and 16 percent, respectively, on the flight-planning and satellite-positioning problems; and on the temporal problems, the improvement was 12 percent and 10 percent.

“The plan that the planner came up with looked more like the human-generated plan when it used these high-level strategies from the person,” Shah says. “There is maybe this bridge to taking a user’s high-level strategy and making that useful for the machine, and by making it useful for the machine, maybe it makes it more interpretable to the person.”

In ongoing work, Kim and Shah are using natural-language-processing techniques to make the system fully automatic, so that it will convert users’ free-form descriptions of their high-level strategies into linear temporal logic without human intervention.

Read the preprint accepted at AAAI here. 

An Air Force airman at the control module of an MQ-9 Reaper. Credit: John Bainter/USAF

Machine ethics offers an alternative solution for artificial intelligence (AI) safety governance. In order to mitigate risks in human-robot interactions, robots will have to comply with humanity’s ethical and legal norms, once they’ve merged into our daily life with highly autonomous capability. In terms of technical challenges, there are still many open questions in machine ethics. For example, what is deontic logic and how can it be used for improving AI safety? How do we fashion the knowledge representation for ethical robots? What about the adaptiveness of the ethical governance? These are all significant questions for us to investigate.

In this interview, we invite Prof. Ronald C. Arkin to share his insights on robot ethics, with a focus on its technical aspects.

Date: February 1st 2017

Interviewee: Prof. Ronald C. Arkin, Regents’ Professor and Associate Dean in the College of Computing at the Georgia Institute of Technology and Executive Committee Member of IEEE Global Initiative for Ethical Considerations in Artificial Intelligence and Autonomous Systems

Interviewer: Dr. Yueh-Hsuan Weng, Co-founder of ROBOLAW.ASIA and Fellow at CLAST AI & Law Research Committee (Beijing) and Tech and Law Center (Milan).

Ron Arkin

Q1 WENG: Thank you for agreeing to an interview with us. Can you tell us a little about your background?

ARKIN: I have been a roboticist for 35 years, so it’s hard to tell [only] a little, but I will do my best. Currently I am a Regents’ Professor and Associate Dean in the College of Computing at the Georgia Institute of Technology. I got my Ph.D. from the University of Massachusetts in 1987, and have been working on robotics since that time. I actually have degrees in chemistry as well—a bachelor’s and a master’s in organic chemistry. So, I have a rather diverse background and followed a non-traditional path compared to most computer scientists.

Q2 WENG: What is your definition of “Ethical Robots”?

ARKIN: That’s an interesting question! I have been asked that one before. I would say an ethical robot is a robotic system that is able to comply with the ethical and social norms of human beings—in some cases in accordance with international law, in other cases in accordance with national or local law. In addition, also being able to comport with humans in a natural manner, obey, and adhere to the same principles as an ethical person.

Q3 WENG: Could you please tell us why you are particularly interested in “Embedded Ethics”? Why is this concept so important?

ARKIN: That’s why you need to read the preface of my book [Governing Lethal Behaviour in Autonomous Robots]. It’s a rather long story. But let me try to give you a short version. Again, it dates back to a meeting—the first international symposium in Roboethics in 2004—where I was exposed to international conventions, e.g., the Geneva convention. It also led me to understand the Vatican’s position on how they viewed appropriate ethical interactions. Perhaps even more important than that one experience was the fact that we roboticists that have been in the field for some time finally started to have success, and we were seeing our technology move out of the laboratory community and into the real world. As such, it wasn’t hard to see how this could go in a variety of different directions.

Some robotics technology could benefit humanity, some not so much. At that point, one has to ask, “who is responsible for this?” So I assumed that, in part, it was myself. I believe that I have to take the responsibility for the technology that I helped to create. The issue is that we, as roboticists, had actually succeeded in many ways in terms of creating technology that we only envisioned when we first started that is going to have potentially profound impact on our society and the world as a whole. As such, we should be proactive in the ways in which we move forward [in order] to get a better understanding of the consequences of the work, and the social implications of the technology we are designing.

https://www.youtube.com/watch?v=jOJwkvihYtc

Q4 WENG: The IEEE Global Initiative for Ethical Considerations in Artificial Intelligence and Autonomous Systems has published their report “Ethically Aligned Design” in December 2016. As a member of the IEEE Global Initiative, could you please tell us about your role?

ARKIN: I serve a variety of roles. First, I am on the executive committee of that particular initiative, so I work at a high-level with about 30 other people or so. We have committees that large. I also co-chair the committee in Affective Computing. Joanna Bryson is my co-chair on that.

Affective Computing is concerned with the issues surrounding the illusion of emotions we can potentially create in these artifacts, and the ways human beings can bond to them. We are exploring a variety of different things, including robot intimacy, or the notion of applying robotic nudges and moving people in certain directions subtly, and in so doing to change the way people act and behave using these technologies, while trying to understand what may or may not be appropriate guidelines based on ethical considerations.

It’s about 15-20 people in that committee. We are one of the newer committees and we only have a couple of pages in the initial design report. But actually, in the June meeting in Austin, for the next release of the document 2.0, we will have as much as the other committees do. Finally, I am also a member of the humanitarian committee. I think it is called economic and humanitarian. I don’t know the exact title, but I am interested in the humanitarian side, and potential uses of robotics for humanitarian applications.

Q5 WENG: What is “Deontic Logic”? Why does it matter to machine ethics?

ARKIN: It may or may not matter. It’s just one approach to a particular problem. Deontic logic is a logic of prohibitions, obligations and permissions. It is one type of framework that is more transparent in order to get a better understanding of the way in which systems operate in the real world. Then it explains its reasoning to a human being, as compared to say deep learning or other forms of neural networks which may indeed provide good results but provide no explanatory capability to humans after the decision has been made.

The notion of prohibitions basically tells the robot some things it cannot do. The notion of permissions describe the things it is potentially allowed to do. And the notion of obligations says things that the robot must do under specific circumstances. It provides a nice framework of being able to explore morality. Others have already explored this, like Bringsjord at RPI who has applied deontic logic in this framework. We are certainly not the first in that particular space. Much of the work we have done has dealt with first-order predicate logic as a basis. It doesn’t have to be deep because it is so explicit in many cases.

Q6 WENG: Your hybrid embedded ethics framework aims for governing lethal behavior of military robots. Can we apply the same technical framework to service robots? For example, personal care robots might have some non-lethal, but still risky autonomous behaviors that need to be regulated by machine ethics.

ARKIN: I believe so. Indeed the whole point of the architecture is trying to design the systems to make them generalizable and not ad hoc for particular sets of circumstances. We have yet to look at service robotics in a traditional sense, but we have already developed something we refer to as the intervening Ethical Governor, which we are using for health care applications—particularly for assisting the management of two human beings in a particular relationship, like a patient and their care giver.

A robot hands a medication bottle to a person. Photo credit: Keith Bujak. Source: Georgia Tech News Center

We are doing work for the National Science Foundation, joint with Tufts University, looking at how in early stage Parkinson’s there is a condition called the facial masking. The loss of muscle movements in the face of the patient can produce stigmatization, and the patient-care giver relationship deteriorates due to the loss of non-verbal communication. We just had a paper accepted at ICRA describing the architecture that we have and how it has been modified from the military version. We generalized it from the military constraint-based version to a rule-based version.

But also, instead of using the international humanitarian law, we are using occupational therapy guides as a basis for generating the responses when two human beings in the relationship step outside of acceptable bounds. So the acceptable bounds now are not defined by international humanitarian law, but they are defined by occupational therapy—such as when a patient or a care giver start to yell at each other, or one starts to leave the room. The hard part is recognizing when dignity is being abridged, particularly in the patient. So we model the moral emotions of empathy, shame and embarrassment. We tried to ensure that the empathy level in the care giver remains within acceptable bounds and that the patient shame and embarrassment doesn’t rise to unacceptable levels. So the intervening Ethical Governor intervenes when a really inappropriate event occurs. It then provides strong corrective action by the robot to encourage people to get along better. The moral adaptor component which we used in the ethical architecture—initially with the guilt model—has now been extended to other moral emotions like shame, embarrassment, guilt and empathy [in order to] try and provide more subtle non-verbal queues [to help the] robot steer the relationship in the correct way before it get out of hand.

Q7 WENG: Regarding your research on machine ethics, what kind of indicators you use to evaluate the final outputs of your project?

ARKIN: In the first one, we were designing a prototype to prove the concept of the architecture. We never make the claim that this is the best way to do it. It is just one way to do it. Then we demonstrate that particular capability through the use of simulations and other aspects. The system was not intended to be fielded, but hopefully others could take those core ideas and potentially put them to use in the real world if they so choose. For the intervening governor, we’re running human subject tests right now—not with Parkinson’s patients yet, but with the elderly population, 50 or older. We’re trying to get a better understanding of the correct sets of rules, and to correct the interventions in a normative fashion, so we can ultimately move on in the final year of the grant to work with actual Parkinson’s patients. Again, using standard HRI technics, such as recording the kinds of reactions they have and parsing the data from the sessions and using pre- and post- evaluation questionnaires.

Q8 WENG: Let’s talk about your proposed machine ethics framework. You had mentioned that it includes four main parts: (1) Ethical Governor (2) Ethical Behavior Control (3) Ethical Adaptor (4) Responsibility Advisor. First of all, what is the major difference between Ethical Governor and Ethical Behavior Control? Is it the difference between deliberative and reactive control, or something else?

ARKIN: The Ethical Governor is basically a standalone component to the architecture which is a bolt on component at the end of the existing system. In the case of military scenario: asking for permission—should it fire or not. This is an intervening stage which either grants or denies permission through the validation of permission-to-fire variable. So it is straightforward thing to be able to do. The Behavior Ethical Control is a different kind of thing as it works purely with the reactive component. What it tries to do is, for each of the individual behaviors themselves, to respond whether or not to engage a target. This is prior to dealing with the Ethical Governor making the decision; trying to make the decision correctly before it gets to the Ethical Governor, where individual behaviors are given ethical guidelines to try and ensure that the behavior is correct itself. Each of the behaviors is provided with appropriate ethical constraints. So the difference is: one is located within the reactive component in each of the individual behaviors that makes the activity of the architecture, and the Ethical Governor is at the end of the architectural chain.

Q9 WENG: What is the importance of Ethical Adaptor? What kinds of machine learning skill that you chose to improve its adaptiveness?

ARKIN: Machine learning is something we haven’t generally explored in this particular context. Especially the military context, we kind of engineer out the adaptiveness and machine learning for fear that the robot would be granted the authority to start to generalize the targets in the field, which we feel is inappropriate. Learning after action is okay. but that’s a different story. The importance of the Ethical Adaptor, and its cognitive models of guilt that we use, is the fact that even the actions the agent takes may inadvertently end up unethical. This is referred to as a “moral luck”. It could be good luck or bad luck. But if it is bad luck, you want to be able to ensure proactively that the bad action doesn’t occur again. That’s actually the functions of cognitive guilt for people as well to reduce the likelihood of reoccurrence of a behavior that resulted in bad actions. By giving this to a robot we hope that if it makes a mistake it will not repeat it. The probability of that behavior reoccurring would be reduced.

The best example is the case of battle damage assessment for a military scenario. For example, if a robot does its assessment and it looks like a released weapon wouldn’t create too much collateral damage, within acceptable limits, then it will release the weapon. But after it does it, the battle damage action is actually more severe then, it initially appeared to be. Under those circumstances, we want the robot in the field to reduce the probability of releasing the weapon again under similar circumstances. That’s, in essence, the analogue of guilt we use, and eventually, it will get to the point that the robot will not engage targets at all. Again, the software is updated to make sure the use is consistent with Geneva convention.

Q10 WENG: The Laws of War (LOW) and Rules of Engagement (ROE) are written in the form of natural language. How did you transfer these norms into machine readable formats?

ARKIN: It’s important to keep it in mind that we do not try and capture all the Rules of Engagement and all the Laws of War. We always advocate that the system should be used in very narrow circumstances and very narrow situations, such as building clearing operations or counter-sniper operations. It’s impossible to capture all human morality and all the Laws of War at this particular point of time. So, the question is what are the relevant Laws of War and Rules of Engagement for these particular narrow situations that the robot will find itself in.

Let’s take a room clearing operation. We try and do this using these constraint-based approaches. The Laws of War basically say that if you see a civilian, you should not shoot them; if you see a surrendered person, you should not shoot them. For aerial vehicles, if you are outside the kill zone (this is the simulations we did) which is designated by the Rules of Engagement, you cannot fire your weapon—explain simply why you cannot do that. If you release the weapon a toward a legitimate target but it would result in damage of cultural property, like a mosque or church or some similar circumstance, that is illegal as well. You are forbidden from doing those sort of things.

So, the initial work we did was more for aerial circumstances. We also computed the collateral damage using a battle damage assessment strategy to determine whether if it is within acceptable limits for a particular target and to restrain the weapon for proportionality. To be able to do this you have to be able to understand the military necessity of the target, which is assigned by a human being. You have to be able to understand the proportionality of the weapon. In other words: Which weapon is sufficient to take out the target but to do no more damage of the surroundings than is absolutely necessary? You have to ensure that you don’t destroy cultural property as well. Basically, adhere to all necessary principles, such as discrimination which is to be able to tell apart combatants and non-combatants. We also state we should restrict the use of these particular systems for high-intensity combat operations, or inter-state warfare, and not the kind of circumstances you find in modern counter- insurgency operations. They should be used with great care and great caution in very narrow circumstances where the predictability of their performance is far more likely to be good.

Q11 WENG: Why did you choose case-based reasoning (CBR) but not rule-based reasoning (RBR) when you designed Responsibility Advisor?
ARKIN: Historically we used case-based reasoning a lot for a variety of different reasons. Part of the reason is to gain information from previous experts. One of the examples in the past was to work with snipers in the laboratory. We can work with experts. It can be legal experts, it can be medical experts, it can be military experts. Again, the primary reason is to be able to readily use the experience of warfighters to guide us in operations.

Q12 WENG: Finally, your proposal “Governing Lethal Behavior: Embedding Ethics in a Hybrid Deliberative/Reactive Robot Architecture” was published in 2007. There have been great technological improvements in artificial intelligence in the past decade. Are there any recent updates from your robot ethics research, or anything that you want to share with us?

ARKIN: The project ended around 2010, so there is additional information. As I say, I would encourage those who are really interested to take a look at the book entitled “Governing Lethal Behavior in Autonomous Robots” . That represents the final statement and outcome for that particular project and includes the significant technical details describing that work. Recent advances include the application in other domains, such as the caregiver-patient relationship [discussed above] using the same underlying principles, including both the Moral Adaptor and Ethical Governor in a slightly different. If we were successful, it may also be useful for assisting other human-human relationships, such as a parent and a teenager who may be at odds with one another or even in marriage consulting. We hope to use robots for the benefit of mankind, whether it’s to make sure that warfare is conducted within internationally agreed upon laws, or whether it’s to ensure that people behave appropriately toward each other

If you enjoyed this article, you may also enjoy:

• How do we regulate robo-morality?
• One being for two origins: A new perspective on roboethics
• Considering robot care, ethics & future tech at #ERW2016 European Robotics Week
• On the ethics of research in robotics, with Raja Chatila
• A European perspective on robot law: Interview with Mady Delvaux-Stehres
• Open Roboethics initiative delivers statement to United Nations CCW

See all the latest robotics news on Robohub, or sign up for our weekly newsletter.

National Grid, the UK grid operator, has several ways of ensuring supply can always meet demand. For shorter gaps in generation, it asks electricity suppliers to run their conventional power stations at below maximum potential output and ramp up as needed.

For longer gaps, it ensures power stations, particularly gas-based ones, are kept on standby. Some stations may only be asked to generate power for between several dozen and a few hundred hours a year. Besides contributing to carbon emissions, operating power plants for such short interventions is expensive.

The question is what to do about this problem. We could build less renewable power and make conventional power “greener” instead by removing the CO₂ and burying it underground. Opinion divides on when these carbon capture technologies can be made commercially viable on a large scale. In the UK, unfortunately two government kickstarter projects have floundered due to concerns about costs and departmental disagreements.

An alternative is to install very big (“grid scale”) batteries capable of storing renewable power to be released when required. This has generated a lot of interest lately. But given the costs of current battery technology, grid-scale storage requires expensive upfront investments.

Solutions on demand

While researchers study these problems, the UK is developing an alternative known as demand-side response. One aspect involves rewarding certain electricity consumers for reducing their usage at short notice. This can range from large industrial customers to smaller consumers using power for heating rooms, cooling, lighting or even refrigeration.

Solar thermal energy system in a modern house.

The other aspect of demand response involves asking customers who own equipment that can store power to help balance surges in demand. For example, the owners of a house equipped with solar panels and corresponding battery storage might reduce repayment costs on the equipment by making the battery units available to the grid. Other equipment in this category includes electric vehicles and hospital/university uninterruptible power supply (UPS) units.

Both types of demand response are happening already. Some industrial power customers and certain other companies such as hotel operators have contracts for reducing power, while National Grid has been attracting much bidder interest for power storage schemes and has some underway in parts of the country. This storage is an alternative to deploying large-scale batteries, and promises to be much more economical if we can make it work on a large enough scale.

Once a pool of customers is set up, some devices might not always be available for storage or reducing demand when needed. This needs to be factored into the calculations both to minimise grid disruption and incentivise customers to participate at these times.The problem is that these schemes get more complicated once the pool of customers gets beyond a certain size. Knowing which customers to sign up and what tariffs to offer requires understanding to what extent devices will be available and at what cost, for example.

There can also be undesired effects, such as large simultaneous rebounds in consumption. For example many refrigerators will draw extra power to get their internal temperature below the required level when a demand response period ends.

Finally there’s a potential major security issue: a central system that collects data about energy usage from many devices may be prone to malicious attacks and information tampering. This could undermine both grid balancing and keeping track of what customers are owed.

How AI can help

Emerging artificial intelligence technologies look like providing answers to these challenges. To select the best participants, for example, grid operators will be able to use sophisticated machine-learning techniques to model the behaviour of individual devices and battery storage units by reviewing data from smart meters and sensors.

Smart Grid concept Industrial and smart grid devices in a connected network. Renewable Energy and Smart Grid Technology Smart city design with future technology for living.

Once signed up for grid storage, it should be possible to estimate the useful lifetime of a battery pack or unit by applying prognostic algorithms to its charging/discharging data. Owners will then receive appropriate compensation, plus the added incentive of knowing how long their battery will last.

When it comes to managing devices in the pool, people used to think we could use individual smart meters or control devices to feed a central server in the cloud. But meters are expensive and the short response times require the cloud server to analyse data in milliseconds, which looks unfeasible once many thousands of units are in a pool.

An alternative is to have metering devices which detect demand levels on the grid themselves and reduce power accordingly. These take pressure off the central server and it only requires metering at site level, rather than for every electrical device. But it still leaves you with a complex control problem in coordinating all these individual decisions. We at Heriot-Watt are working on a solution to this using AI-based algorithms.

Another line of AI research draws on insights from algorithmic game theory to develop reward/penalty mechanisms which ensure enough customers in the pool are willing to participate, and actually respond when necessary. Researchers are also optimistic that blockchain protocols, using the same technology as Bitcoin, could underpin a decentralised ledger system that would get round the security risk of having a single storage point for user data.

Numerous AI research groups, both in the UK and elsewhere, have been addressing these challenges, while a number of start-ups have started developing such systems in practice – relatively simple versions of machine learning are now beginning to be used, for instance. The UK has a good chance to be at the forefront of international efforts to make smarter demand response a reality over the next few years.

This article was originally published on The Conversation.

You might also enjoy the following articles:

• Wearable AI that can detect the tone of a conversation
• Artificial intelligence and ethics: Who does the thinking?
• Soft exosuit economies: Understanding the costs of lightening the load
• MIT Media Lab to participate in $27 million initiative on AI ethics and governance

See all the latest robotics news on Robohub, or sign up for our weekly newsletter.

Part 2: Autonomous Systems and Transparency

In my previous post I argued that a wide range of AI and Autonomous Systems (from now on I will just use the term AS as shorthand for both) should be regarded as Safety Critical. I include both autonomous software AI systems and hard (embodied) AIs such as robots, drones and driverless cars. Many will be surprised that I include in the soft AI category apparently harmless systems such as search engines. Of course no-one is seriously inconvenienced when Amazon makes a silly book recommendation, but consider very large groups of people. If a truth (such as global warming) is – because of accidental or willful manipulation – presented as false, and that falsehood is believed by a very large number of people, then serious harm to the planet (and we humans who depend on it) could result.

I argued that the tools barely exist to properly assure the safety of AS, let alone the standards and regulation needed to build public trust, and that political pressure is needed to ensure our policymakers fully understand the public safety risks of unregulated AS.

In this post I will outline the case that transparency is a foundational requirement for building public trust in AS, based on the radical proposition that it should always be possible to find out why an AS made a particular decision.

Transparency is not one thing. Clearly your elderly relative doesn’t require the same level of understanding of her care robot as the engineer who repairs it. Nor would you expect the same appreciation of the reasons a medical diagnosis AI recommends a particular course of treatment as your doctor. Broadly (and please understand this is a work in progress) I believe there are five distinct groups of stakeholders, and that AS must be transparent to each, in different ways and for different reasons. These stakeholders are: (1) users, (2) safety certification agencies, (3) accident investigators, (4) lawyers or expert witnesses and (5) wider society.

1. For users, transparency is important because it builds trust in the system, by providing a simple way for the user to understand what the system is doing and why.
2. For safety certification of an AS, transparency is important because it exposes the system’s processes for independent certification against safety standards.
3. If accidents occur, AS will need to be transparent to an accident investigator; the internal process that led to the accident need to be traceable.
4. Following an accident lawyers or other expert witnesses, who may be required to give evidence, require transparency to inform their evidence. And
5. for disruptive technologies, such as driverless cars, a certain level of transparency to wider society is needed in order to build public confidence in the technology.

Of course the way in which transparency is provided is likely to be very different for each group. If we take a care robot as an example transparency means the user can understand what the robot might do in different circumstances; if the robot should do anything unexpected she should be able to ask the robot ‘why did you just do that?’ and receive an intelligible reply. Safety certification agencies will need access to technical details of how the AS works, together with verified test results. Accident investigators will need access to data logs of exactly what happened prior to and during an accident, most likely provided by something akin to an aircraft flight data recorder (and it should be illegal to operate an AS without such a system). And wider society would need accessible documentary-type science communication to explain the AS and how it works.

Source: ROBOTT-NET

Samsung Simband. Image: Jason Dorfman, MIT CSAIL

It’s a fact of nature that a single conversation can be interpreted in very different ways. For people with anxiety or conditions like Asperger’s, this can make social situations extremely stressful. But what if there was a more objective way to measure and understand our interactions?

Researchers from MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) say that they’ve gotten closer to a potential solution: an artificially intelligent wearable system that can predict if a conversation is happy, sad or neutral based on a person’s speech patterns and vitals.

“Imagine if, at the end of a conversation, you could rewind it and see the moments when the people around you felt the most anxious,” says graduate student Tuka Alhanai, who co-authored a related paper with PhD candidate Mohammad Ghassemi that they will present at next week’s Association for the Advancement of Artificial Intelligence (AAAI) conference in San Francisco. “Our work is a step in this direction, suggesting that we may not be that far away from a world where people can have an AI social coach right in their pocket.”

As a participant tells a story, the system can analyze audio, text transcriptions and physiological signals to determine the overall tone of the story with 83 percent accuracy. Using deep-learning techniques, the system can also provide a specific sentiment score for five-second intervals within a conversation.

“As far as we know, this is the first experiment that collects both physical data and speech data in a passive but robust way, even while subjects are having natural, unstructured interactions,” says Ghassemi. “Our results show that it’s possible to classify the emotional tone of conversations in real-time.”

The team is keen to point out that they developed the system with privacy strongly in mind: the algorithm runs locally on a user’s device as a way of protecting personal information. (Alhanai says that a consumer version would obviously need clear protocols for getting consent from the people involved in the conversations.)

How it works

Past studies in this area would often show participants “happy” and “sad” videos, or ask them to artificially act out specific emotive states. But in an effort to elicit more organic emotions, the team instead asked subjects to tell a happy or sad story of their own choosing.

Subjects wore a Samsung Simband, a device that captures high-resolution physiological waveforms to measure features like movement, heart rate, blood pressure, blood flow and skin temperature,. The system also captured audio data and text transcripts to analyze the speaker’s tone, pitch, energy, and vocabulary.

Mohammad Ghassemi and Tuka Alhanai converse with the wearable. Image: Jason Dorfman MIT CSAIL

“The team’s usage of consumer market devices for collecting physiological data and speech data shows how close we are to having such tools in everyday devices,” says Björn Schuller, professor and chair of Complex and Intelligent Systems at the University of Passau in Germany who was not involved in the research. “Technology could soon feel much more emotionally intelligent, or even ‘emotional’ itself.”

After capturing 31 different conversations of several minutes each, the team trained two algorithms on the data: one classified the overall nature of a conversation as either happy or sad, while the second classified each five-second block of every conversation as either positive, negative or neutral.

Alhanai notes that, in traditional neural networks, all features about the data are provided to the algorithm at the base of the network. In contrast, her team found that they could improve performance by organizing different features at the various layers of the network.

“The system picks up on how, for example, the sentiment in the text transcription was more abstract than the raw accelerometer data, says Alhanai. “It’s quite remarkable that a machine could approximate how we humans perceive these interactions, without significant input from us as researchers.”

Results

Indeed, the algorithm’s findings align well with what we humans might expect to observe. For instance, long pauses and a monotonous vocal tones were associated with sadder stories, while more energetic, varied speech patterns were associated with happier ones. In terms of body language, sadder stories were also strongly associated with increased fidgeting and cardiovascular activity, as well as certain postures like putting one’s hands on one’s face.

Graph showing real-time emotion detection. Image: MIT CSAIL

On average, the model could classify the mood of each five-second interval with an accuracy that was approximately 18 percent above chance, and a full 7.5 percent better than existing approaches. In future work, the team hopes to collect data on a much larger scale, potentially using commercial devices like the Apple Watch that would allow them to more easily deploy the system out in the world.

“Our next step is to improve the algorithm’s emotional granularity so it can call out boring, tense, and excited moments with greater accuracy instead of just labeling interactions as ‘positive’ or ‘negative’,” says Alhani. “Developing technology that can take the pulse of human emotions has the potential to dramatically improve how we communicate with each other.”

Click here to download the research paper.

You might also enjoy the following articles:

• Artificial intelligence and ethics: Who does the thinking?
• Soft exosuit economies: Understanding the costs of lightening the load
• MIT Media Lab to participate in $27 million initiative on AI ethics and governance
• How artificial intelligence is changing our Christmas shop
• White House report: Artificial intelligence, automation, and the economy

See all the latest robotics news on Robohub, or sign up for our weekly newsletter.

bots_alive Gotcha

Modern character AI could be better.

I’m unaware of any NPCs or electronic toy characters that can sustain an illusion of life over more than an hour. They suffer from predictability, simplicity, and inauthentic verbal and nonverbal behavior.

This post is about a new form of character AI. Much like motion capture for scripted animation, this new technique may revolutionize how interactive characters are created, through observation of authentic human-generated behavior.

I’ll explain via a product on which we’re pioneering this technique.

Bots_alive! A robot that plays and finds its way

We launched bots_alive last week on Kickstarter (campaign here). Here’s a 1-minute teaser video with mostly scripted motion.

The product is a smartphone kit that gives simple but lifelike autonomy to a popular RC toy, the Hexbug® Spider.

That video shows the product as we envision it at shipping time. Here you can see how it acts now, with this new character AI technique driving its core functionality already.

Training data to build a character

The character AI for these robots is created in a new way. Here’s how it works.

A human teleoperates the robot through many different scenarios. The teleoperator acts as an improvising puppeteer; he or she follows heuristics yet also empathizes with the robot and authentically responds to what happens, whether it be with curiosity, fear, jubilation, or another emotion.

For bots_alive, the teleoperator watches a screen display of what the system sees and pushes buttons to send commands of forward, back, left, right, forward-right, forward-left, back-right, or back-left. We also consider not pushing a button to be a no-action command.

From these teleoperation sessions, we gather training data containing (a) teleoperation commands and (b) information about the context in which each command was given. A category of machine learning called supervised learning is applied to create a model of the puppeteer, which effectively answers the question: In context X, what is the probability that the teleoperator would have given command Y?

Creating a behavioral model

All of that happens during development. In the end user’s hands, the robot is autonomously controlled by the model. The model of the puppeteer is the character AI.

Let’s assume that the contextual information in the training data is sufficient to justify why the teleoperator chooses one action over another. Then the data contains the spontaneity, uncertainty, and social authenticity* expressed by the human puppeteer. And the autonomous robot retains these qualities if the machine learning is effective.

Why go to the trouble of using machine learning?

We expect this process to create character AI that feels more organic and alive than traditional character AI.

Traditional methods consist of imagining and writing rules for behavior or a finite state machine. It’s much more of an abstract exercise than immersing someone in a situation as the puppeteer.

If you play video games, imagine how much you sometimes empathize with the character you’re controlling. You may even forget the separation between the two of you, experiencing the world as that character. Consider whether your control in these moments is different, both at a high level and in tiny movements, than what the character would do if you had simply written down a set of rules for it to act by.

In that difference lies the promise of this method for developing character AI.

In psychology research methodology, there’s a broad consensus that if you want to know what someone would do in a situation, you don’t ask them what they would do. You put them in the situation and observe. Traditional character AI development is like asking the human developer or designer. Our method is instead analogous to observing a human immersed in the experience of the character. That’s why we expect it to create more authentic characters.

Our method is an application of what’s called learning from demonstration.** For times the teleoperation needs to include human interaction with the character, we keep the teleoperation secret to keep the human interaction partner from changing his or her behavior from what would be done with an autonomous character. We coined this secret version learning from the wizard (or LfW) when it was a research project at the MIT Media Lab. The name combines learning from demonstration with the Wizard of Oz experimental paradigm.

Training as interactive machine learning

In our development, I act as the teleoperator and machine learning programmer. The training process is not simply a bunch of demonstration. Rather, it’s an iterative process of:

• demonstrate
• apply machine learning on the demonstration data set
• observe behavior from the learned model
• create more demonstrations in those contexts the robot isn’t acting satisfactorily
• apply machine learning, and so on.

A general representation of interactive machine learning. From Amershi et al., 2015. Power to the People: The Role of Humans in Interactive Machine Learning. In AI Magazine.

Throughout these iterations, the teleoperator and algorithm designer also reflect on what contextual information still needs to be encoded to improve learning, determine what context cannot be encoded and therefore should be ignored by the teleoperator, and find fun and more delightful behavior than originally puppeteered.

How can we know that it’s better?

We can’t. Not with 100% confidence. But we have already seen that it can make compelling character AI.

We saw it in our study at MIT. In a relatively large randomized experiment, kids interacted either with a human-teleoperated robot or one with autonomous behavior learned from previous teleoperation. Compared to a no-robot condition, the teleoperated robot and the autonomous robot programmed by machine learning elicited similar behavior from their human interaction partners. Curiously, when children were asked whether the robot was human-controlled or autonomous, approximately half in each condition thought it was human-controlled. If you’re familiar with the Turing Test, the popularly known test of the effectiveness of an artificial intelligence, you might recognize that the MIT study constitutes passing a certain narrow and social Turing Test.

Compelling character AI was achieved in another such Turing test, using a technique similar to learning from demonstration, in Unreal Tournament 2004. BotPrize is a yearly academic competition for the bot that best passes as human. The 2012 winner was a bot by Jacob Schrum, Igor Karpov, and Risto Miikkulainen. It combined algorithmic optimization—which can result in markedly unnatural behavior that still achieves the specified goal— of its combat effectiveness with selective replay of recorded human behavior. Twenty-five human judges played alongside the Unreal-playing bots, repeatedly judging whether the other characters were human or artificial during combat. The bot by Schrum and colleagues was judged human more than 51.9% of the time. That percentage was just below the two human characters who were judged most human, at 53.3% and 52.2%. The algorithm is described here.

Lastly, we see compelling character AI now, in our playtesting of bots_alive.

The limitations of this approach are (1) sensing and encoding contextual information and (2) applying machine learning effectively. Both academia and industry are intensely researching how to improve those two limitations. Major strides have been made in the past 10 years, most notably through deep learning, and so it seems likely that further improvement is yet to come.

Our ability to write down behavioral rules that encode authentic behavior with an illusion of life? If there’s a reason why that will improve meaningfully, I don’t see it.

The webcast will take place on today from 9am-5:30pm EST and February 1st from 9am-5pm EST. Webcast participants are encouraged to submit questions for the presenters by e-mailing Michelle Schwalbe at mschwalbe@nas.edu who will read them out if time permits.

Machine learning is at the core of many applications that have become part of daily life, from voice recognition to image perception. These technologies, which a few years ago were performing at noticeably below-human levels, can now outperform people at some tasks. As the field continues to evolve, machine learning has the potential to play a transformative role across a diverse range of sectors including transportation, medicine, public services, and finance. This forum will bring together scientists from the UK and the US to explore potential applications for machine learning and discuss the legal and ethical challenges that could arise as machine learning algorithms are implemented.

Watch the live webcast here.

Session outlines below:

Session 1: The Frontiers of Machine Learning

The ubiquity of data, accessibility of computing power, and algorithmic advances have driven rapid progress in machine learning over the past five years. Not only does machine learning now underpin many applications that have become part of daily life, the field continues to evolve quickly, and has the potential to play a transformative role across a diverse range of sectors. This session will explore the frontiers of machine learning, in terms of both cutting-edge technology and near-term applications, and discuss the state of the art of machine learning.

Session 2: Machine learning and society

People and machine learning systems are increasingly interacting through a range of applications or contexts. This expansion of machine learning raises legal and ethical questions, re-frames discussions about uses of data, and poses new challenges for the governance of this technology. The social acceptability of different machine learning applications, desirability of automated decision-making processes, adequacy of processes to manage concerns about statistical stereotyping or privacy, and more, will all influence how and where society has confidence in the deployment of machine learning systems. This session will explore the societal implications of increased use of machine learning, and the opportunities and challenges associated with advances in the field.

Session 3: Machine learning in the research and commercial communities

There are enormous opportunities in machine learning in academia, research labs, and industry. While much of the research and development of machine learning to date has been done in the commercial world, each of these communities will continue advancing this field. Establishing key research challenges and areas of commercial opportunity will, therefore, be important in moving the frontiers of machine learning forward. This session will explore key areas of interest in machine learning in the research and commercial communities.

Twitter feed

#MachineLearning Tweets

About the Sackler Forum

The Raymond and Beverly Sackler U.S.-U.K. Scientific Forum was established to help the scientific leadership of the United Kingdom and the United States forge an enduring and productive partnership on pressing topics of worldwide scientific concern with benefit to all people. These meetings are organized jointly by the National Academy of Sciences and the Royal Society, and alternate between locations in the U.S. and the U.K. This program is made possible by a generous gift from Raymond and Beverly Sackler.

Image: IEEE

On the 15th November 2016, the IEEE’s AI and Ethics Summit posed the question: “Who does the thinking?” In a series of key-note speeches and lively panel discussions, leading technologists, legal thinkers, philosophers, social scientists, manufacturers and policy makers considered such issues as:

• The social, technological and philosophical questions orbiting AI.
• Proposals to program ethical algorithms with human values to machines.
• The social implications of the applications of AI.

Click here to read the final report, and enjoy a selection of videos from the conference below.

Welcome: Marko Delimar
Keynote Speaker: Wojciech Wiewiórowski

Panel 1: Autonomous Systems – Managing Risk and Reward
Panel Members: Raja Chatila, Kay Firth-Butterfield, Juha Heikkilä, Jérôme Perrin

Panel 2: Programming Human Ethics: Cui Bono?
Panel Members: Joanna Bryson, Mady Delvaux, Sarah Spiekermann, Corrine Cath

Panel 3: Social Implications – Perils & Promises of AI
Panel Members: Greg Adamson, Nikolaos Mavridis, Paul Nemitz, Aurélie Pols

Closing Remarks: John C. Havens

You might also enjoy the following articles about AI and AI/robotics policy:

• MIT Media Lab to participate in $27 million initiative on AI ethics and governance
• How artificial intelligence is changing our Christmas shop
• White House report: Artificial intelligence, automation, and the economy
• White House launches public workshops on AI issues
• White House releases reports focusing on opportunities and challenges of AI

See all the latest robotics news on Robohub, or sign up for our weekly newsletter.

Part 1: Autonomous Systems and Safety

We all rely on machines. All aspects of modern life, from transport to energy, work to welfare, play to politics depend on a complex infrastructure of physical and virtual systems. How many of us understand how all of this stuff works? Very few I suspect. But it doesn’t matter, does it? We trust the good men and women (the disgracefully maligned experts) who build, manage and maintain the infrastructure of life. If something goes wrong they will know why. And (we hope) make sure it doesn’t happen again.

All well and good you might think. But the infrastructure of life is increasingly autonomous – many decisions are now made not by a human but by the systems themselves. When you search for a restaurant near you the recommendation isn’t made by a human, but by an algorithm. Many financial decisions are not made by people but by algorithms; and I don’t just mean city investments – it’s possible that your loan application will be decided by an AI. Machine legal advice is already available; a trend that is likely to increase. And of course if you take a ride in a driverless car, it is algorithms that decide when the car turns, brakes and so on. I could go on.

Algorithmic trading is now commonplace

These are not trivial decisions. They affect lives. The real world impacts are human and economic, even political (search engine results may well influence how someone votes). In engineering terms these systems are safety critical. Examples of safety critical systems that we all rely on from time to time include aircraft autopilots or train braking systems. But – and this may surprise you – the difficult engineering techniques used to prove the safety of such systems are not applied to search engines, automated trading systems, medical diagnosis AIs, assistive living robots, delivery drones, or (I’ll wager) driverless car autopilots.

Why is this? Well, it’s partly because the field of AI and autonomous systems is moving so fast. But I suspect it has much more to do with an incompatibility between the way we have traditionally designed safety critical systems, and the design of modern AI systems. There is I believe one key problem: learning. There is a very good reason that current safety critical systems (like aircraft autopilots) don’t learn. Current safety assurance approaches assume that the system being certified will never change, but a system that learns does – by definition – change its behaviour, so any certification is rendered invalid after the system has learned.

And as if that were not bad enough, the particular method of learning which has caused such excitement – and rapid progress – in the last few years is based on Artificial Neural Networks (more often these days referred to as Deep Learning). A characteristic of ANNs is that after the ANN has been trained with datasets, any attempt to examine its internal structure in order to understand why and how the ANN makes a particular decision is impossible. The decision making process of an ANN is opaque. Alphago’s moves were beautiful but puzzling. We call this the black box problem.

Safety critical systems such as aircraft autopilots do not learn

Does this mean we cannot assure the safety of learning autonomous/AI systems at all? No it doesn’t. The problem of safety assurance of systems that learn is hard but not intractable, and is the subject of current research*. The black box problem may be intractable for ANNs, but could be avoided by using approaches to AI that do not use ANNs.

But – here’s the rub. This involves slowing down the juggernaut of autonomous systems and AI development. It means taking a much more cautious and incremental approach, and it almost certainly involves regulation (that, for instance, makes it illegal to run a driverless car unless the car’s autopilot has been certified as safe – and that would require standards that don’t yet exist). Yet the commercial and political pressure is to be more permissive, not less; no country wants to be left behind in the race to cash in on these new technologies.

This is why work toward AI/Autonomous Systems standards is so vital, together with the political pressure to ensure our policymakers fully understand the public safety risks of unregulated AI.

In my next blog post I will describe one current standards initiative, towards introducing transparency in AI and Autonomous Systems based on the simple principle that it should always be possible to find out why an AI/AS system made a particular decision.

The next few years of swimming against the tide is going to be hard work. As Luke
Muehlhauser writes in his excellent essay on transparency in safety-critical systems “…there is often a tension between AI capability and AI transparency. Many of AI’s most powerful methods are also among its least transparent”.

*some, but nowhere near enough. See for instance Verifiable Autonomy.

See all the latest robotics news on Robohub, or sign up for our weekly newsletter.

Forbes “30 under 30″ list celebrates the achievements of 30 game changers in 20 different industries — 600 in total — of the brightest young entrepreneurs and innovators who are challenging conventional wisdom. The list highlights some interesting trends in robotics, AI, intelligence energy storage, and automation. Connected/smart consumer goods, advanced manufacturing, and artificial intelligence are big ticket items for 2017. Forbes also unveiled its second annual “30 Under 30 Europe” list, featuring 300 young innovators, covering 10 categories. The entire list can be viewed here. 

You may also recognise some of these bright young minds from their contributions on Robohub:

Smart design / connected devices / energy storage

• Amrit Robbins & Anthony Diamond, Cofounders, Axiom Exergy
  An industry first, the Refrigeration Battery makes it possible for supermarkets, cold-storage facilities, and food processors to intelligently store and deploy refrigeration. The battery plugs into your refrigeration system, just like a display case. It stores thermal energy produced by your refrigeration system during off-peak hours and deploys it during on-peak hours. As a result, electricity demand is reduced by up to 40 percent during peak hours—a significant cost saving.
• Paige Kassalen, Engineer, Solar Impulse
  Kassalen was an electrical engineer on the ground crew of Solar Impulse, the first airplane to circle the globe powered by the sun. At 22, Paige was the youngest member and only American on the ground crew.
• Justin Barozie, Associate Manager of Manufacturing Engineering, Battery, Tesla Motors
  Barozie is lead manufacturing engineer for Tesla’s energy products and Model 3 battery modules. He leads a team of nine manufacturing engineers, and has designed and delivered more than 50 mass-production, multi-million dollar state-of-the-art automation projects.
• Nishant Garg & Jimit Shah, Cofounders, Flow Labs
  Flow Labs has devised a small wireless-enabled sensor that attaches to water pipes and measures the amount of water flowing through them — ideal for homes and businesses looking for ways to save water and save money.
• Ravi Kurani, CEO, Sutro
  Kurani has built a sensor connected app that monitors the health of your swimming pool, notifies you if maintenance is required and orders pool supplies. Automating water system maintenance saves chemicals, water and energy.
• Maanasa Mendu, student, Mason High school
  On family visits to India, Mendu saw the challenges of energy scarcity. A high school freshman in Ohio, she created a piezoelectric “leaf” device that harvests energy from sun, wind and rain. She’s working to commercialize her device.
• Colin Touhey, CEO, Pvilion
  Touhey’s Pvilion makes flexible photovoltaic solar products. He has led the development of solar-powered fabric for Google, built a sustainable 6,000-square-foot building on the roof of Carnegie Hall, and devised self-erecting robotic Army shelters.
• Richard Wang, CEO, Cuberg
  Wang’s Cuberg is developing a safer and more energy-dense battery to improve the affordability and range of electric vehicles. Wang studied battery degradation mechanisms Tesla Motors and has a PhD in materials science from Stanford.

Manufacturing & Industry

• Rhae Adams, Director of Energy & Mining, Planetary Resources
  Adams is helping to develop a market for mined materials from asteroids including water and industrial metals. Planetary Resources has raised more than $30 million in 2016 for its asteroid prospecting system.
• Chad Ammon, Cofounder, Inovdrone
  INOVA Drone is a TechStars company developing small drones for commercial and governmental applications, including public safety and critical infrastructure inspection. It was one of the first companies in Qualcomm’s Robotics Accelerator. Before founding the company, Amonn was one of the lead drone designers at 3D Robotics.
• Jeremy Blum, Head of Electrical Engineering, Shaper Tools
  Blum is head of electrical engineering at Shaper, a company looking to improve old-fashioned power tools with cutting-edge technology like machine vision. Prior to Shaper, Blum worked as a lead electrical system engineer at GoogleX, and did prototyping on Google Glass. He’s a prolific inventor, with three patents, and hosts a popular YouTube channel where he shows off his electrical engineering projects.
• Stefanie Mueller, Assistant Professor, MIT
  Mueller’s work focuses on the computer science of “physical data,” such as that involved in 3D printing. She leads the HCI Engineering Group at MIT CSAIL, which works on hardware and software that advance personal fabrication technologies, and is the conference General Co-Chair at the ACM SIGGRAPH Symposium on Computational Fabrication 2017.
• Anurag Garg, CEO, Dattus
  Garg is cofounder and CEO of Dattus, an industrial Internet of things company that transforms old manufacturing facilities into smarter factories.
• Hasier Larrea, CEO, Ori [featured on Robohub HERE]
  Larrea is the founder of Ori (which derives its name from origami), a robotic, flexible furniture system for small apartments. The system allows for a number of configurations, from bedroom to office to living room, and back again, all controlled from one control panel.
• Ladislas de Toldi and Marine Couteau, cofounders, Leka [featured on Robohub HERE]
  Couteau and de Toldi’s company, Leka, has a solution for helping children with special needs engage with their education. They manufacture a multi-sensory robotic ball with lights, sounds and vibrations to engage children while tracking their progress.
• Marco Mascorro, Cofounder, Fellow Robots [featured on Robohub HERE]
  Retailers regularly struggle with inventory shrinkage and discrepancies in pricing and product that can cost a small fortune. Mascorro, who previously worked on robotics for BMW Group, cofounded Fellow Robots, which is developing robots that can help retailers with inventory management by scanning the shelves and keeping tabs on product. The company is rolling out its robots at American home improvement chain Lowe’s and Japanese electronics retailer Yamada Denki, among others.

Healthcare / wearables / brain-computer interfaces

• Alessandro Babini, Cofounder, Humon
  Alessandro’s company is building a wearable device that measures oxygen levels in muscles to determine how hard athletes should push themselves, when muscles have recovered and when training can resume. The company has received $1 million in funding.
• Huanyu Cheng, Assistant Professor, Pennsylvania State University
  Cheng researches biologically-inspired electronics for use in robotics, biomedicine and energy. One of his more notable projects was being part of the research team that developed wearable electronic tattoos that can eventually dissolve in the human body.
• Prarthna Desai, Operations, Zipline
  Desai left her Harvard grad school program to use drones to get medication to people in the developing world. In her operations role at Zipline, she is leading efforts to integrate the medicine-delivery-by-drone service with the healthcare system in Rwanda.
• Nicole Moskowitz & Jessica Traver, Cofounders, IntuiTap Medical
  IntuiTap wants to streamline the process by using imaging technology, pressure sensors and predictive analytics not only to make spinal taps more accurate, but also to make the process more pleasant for patients.
• Param Shah and Alex Matthews, Cofounders, Fusiform
  Shah and Mathews are co-founders of Fusiform, which is building an infrastructure for custom fabrication for orthopedic devices. Rather than cast a patient with plaster, Fusiform has developed a way to take a 3D scan and design the orthopedics digitally.
• Lydia Kisley, Postdoc fellow, University of Illinois, Urbana-Champaign
  Kisley focuses on using new technologies to take pictures of chemicals at a nanoscale level in order to better manipulate them. A few projects: showing that proteins clump in the presence of gold nanoparticles only when they are in low concentrations; and new efforts to understand how drug molecules fit together – an effort that could allow scientists to find the porous spaces in any nanotech particle.
• Ashwin Pushpala, Founder, Sano
  Pushpala thinks traditional preventive health is “fundamentally wrong” in its approach.His company, Sano, combines personalized health with a Fitbit approach. It has raised $18.75 million to develop a wearable sensor that, instead of counting steps, assesses things like blood sugar so users can see the impact of dietary changes on individual bodies.
• Niko Skievaski, Cofounder, Redox
  One of the biggest problems in healthcare: getting data from one place to another. Skievaski’s company, Redox, has raised $4 million to make an API that works as a bridge between healthcare data and the software that uses it, translating patients’ records across platforms. So far the company has partnered with more than 1,000 software vendors.
• Daniel Szafir, Assistant Professor, University of Colorado, Boulder
  Szafir’s research sounds almost like science fiction. It’s focused on human-robot interaction and human-computer interaction. Specifically, he’s developing ways to improve the performance of small flying robots so they can operate by themselves. He’s also worked on better ways to interact with computers, including brain-computer interfaces and virtual reality.
• TowerView Health, Cofounders
  After he was diagnosed with cancer, Nick Valilis was surprised by how difficult it was to remember to take his pills. So he teamed up with his college roommates to found TowerView Health, which sells a smart pill box with custom trays of medication.

Big data / machine learning / algorithms

• Michelle Atallah, PhD candidate, Stanford University
  Atallah is designing algorithms to identify mechanisms in tumor-driven immunosuppression as part of her PhD candidacy in the Cancer Biology program at Stanford. Utilizing a new technology called Multiplexed Ion Beam Imaging, she is collecting data about immune cells within tumors in order to better apply therapeutic intervention.
• Xi Chen, Assistant Professor, New York University
  Chen studies machine learning, high-dimensional statistics and operations research. These lead to new learning methods and approaches for big data analysis, with applications for business, medicine, and more.
• Korin Reid, Senior Data Scientist, McKesson
  Reid uses big data technologies to scale predictive modeling and machine learning on billions of healthcare records, reaching over 160 million people. Named one of McKesson’s 2016 Distinguished Technologists, she also mentors youth in STEM skills.
• Justin Solomon, Assistant Professor, MIT
  Solomon researches geometric problems in computer graphics, computer vision and machine learning. Co-inventor of three patents, he also spent five years doing computer research for Pixar, developing new algorithms for image processing in movies.
• Katharina Volz, Founder, Razor, Inc.
  Volz’s company, OccamzRazor, hopes to speed up the processes behind scientific research by addressing information fragmentation and data validation. This, in turn, should make collaboration easier and help build an institutional knowledge base. The company is running pilots at Stanford, Harvard, MIT and biotech companies.

Artificial intelligence and VR

• Marta Jamrozik & Misha Laskin, Cofounders, Claire
  Marta Jamrozik and Misha Laskin built an artificially intelligent chatbot that allows millions of users to provide product feedback in real-time. This helps companies get valuable pre-launch insight, and identify their bestsellers and worst performers early on — which is why Kohl’s, Alex And Ani, Rebecca Minkoff, and Rent the Runway are all Claire customers.
• Scott Britton, Cofounder, Troops.ai
  Britton cofounded Troops.ai with the goal of making work easier for sales teams by providing them with an artificially intelligent assistant that helps them perform their CRM functions in messaging interfaces like Slack.
• Ben Aneesh & Natalie Gray, Cofounders, Cover
  The entrepreneurs behind fashion app Stylekick are back with a less sexy but very useful startup. Mobile app Cover allows you to buy insurance as easily as you’d take a photo: send a snap of your jewelry, car, or home and get a quote.
• Beatty & Scranton, Cofounders, IrisVR
  Beatty and Scranton cofounded New York City-based IrisVR in 2014 to help companies transition their assets to virtual reality (VR) by making software to easily create, share and interact with 3D models on all major VR platforms, like the Oculus Rift and HTC Vive.
• Arruda and Dall’Oglio and Ovbiagele, Cofounders, ROSS intelligence.
  A legal research engine that uses artificial intelligence to automate legal processes making them more efficient and less expensive. Leveraging IBM’s Watson, ROSS uses natural language processing to search and provide legal information from citations to full legal briefs.

Read the entire list here.

The population of the scenic ski-resort Davos, nestled in the Swiss Alps, swelled by nearly +3,000 people between the 17^th and 20^th of January. World leaders, academics, business tycoons, press and interlopers of all varieties were drawn to the 2017 World Economic Forum (WEF) Annual Meeting. The WEF is the foremost creative force for engaging the world’s top leaders in collaborative activities to shape the global, regional and industry agendas for the coming year and beyond. Perhaps unsurprisingly given recent geopolitical events, the theme of this year’s forum was Responsive and Responsible Leadership.

With the onset of the fourth industrial revolution, increasingly discontented segments of society not experiencing congruous economic and social progress are in danger of existential uncertainty and exclusion. Responsive and Responsible Leadership entails inclusive development and equitable growth, both nationally and globally. It also involves working rapidly to close generational divides by exercising shared stewardship of those systems that are critical to our prosperity.

In the end, leaders from all walks of life at the Annual Meeting 2017 must be ready to react credibly and responsibly to societal and global concerns that have been neglected for too long.”

Developing last year’s theme—“The fourth industrial revolution”—this year’s luminaries posited questions, among many others, concerning incipient robotics and artificial intelligence technologies set to have a pronounced impact on the global economy and global consciousness alike. What can we learn from the first wave of AI? How can the humanitarian sector benefit from big data algorithms? How will drone technology change the face of warfare? Can AI and computational tech help foster responsive and responsible leadership? What are the downsides of technology in the fourth industrial revolution?

Enjoy a selection of tech-themed videos below.

And a bit about global science including big data, open source science and education.

Credit: Catalia Health

Catalia Health is leading the surge in social robotics, with Mabu, their patient care management system. Catalia Health likes to be seen primarily as a health company that utilizes robots, rather than a robotics company. This focus on solving real world problems while shipping a product has seen Catalia attract both customers and investors, and recently close their Series A round.

Interview with Cory Kidd, Founder & CEO of Catalia Health

(edited for clarity)

What is Catalia Health?

Catalia Health is a patient care management company. We focus on helping patients adhere to their treatment, whether that be taking medication, or managing chronic disease over the long term. That’s the focus of what we do, and part of how we deliver this to patients is through a cute little robot called Mabu who engages with patients through conversation. She’s a little over a foot tall, and can sit wherever you want to put her … on a countertop or bedside table … and she has big eyes that make eye contact with you while you’re talking to her. Conversations with her might last a minute or two, or maybe five or ten minutes; it really depends on the individual patient and what they want to talk about.

Mabu has a touch screen on the front that she can use to display information, but our overall focus is to create an engaging relationship between the technology and the patient. The reason that we use the robot — as opposed to just delivering this through a phone screen or a tablet or PC — is about psychology and not about technology. When we are in front of a robot that has eyes that can look at us and blink, we tend to be more engaged, and we find the robot to be more credible and informative than if the same information were delivered to us through an app. While we have a lot of healthcare applications that we’re looking to build, the core of this is really just basic psychology: how can we create engagement that lasts for a long time? Psychologists have studied the benefits of face-to-face communication for decades.

Is speech the primary interaction that people have with Mabu?

Our platform’s primary means of interaction is conversation, but this can happen in more than one way. For example, when Mabu is talking, she also displays what she is saying on her screen, to make it easy for anyone to understand what’s going on. And when I reply, I can speak back to her, or I can touch a button or location on the screen. And if I’m not at home, I can also get a reminder via text message … in the future this might happen through an app or other desktop interface.

The physical robot is the thing that’s creating the engagement — the relationship — but we can interact with people through other forms of technology as well.

Does the conversation with Mabu end at home? How is information transferred to the healthcare provider?

We do send information summaries back to health care providers — a pharmacist or physician or some other caregiver — but the overall problem we are trying to help with is that the healthcare system simply doesn’t have enough people to manage chronic disease at scale. So while our technology might also enable tele-operation or tele-presence, the focus of our business is to be able carry out an autonomous one-on-one interaction with patients in real time.

Is the patient, or end user, your customer?

Patients get a lot of benefit from our platform, but they are not the ones who are paying for it.

Our direct customers are pharmaceutical manufacturers and healthcare providers. They provide programs to help patients be more effective at taking their medications and managing their conditions, so in their eyes we are another tool in their arsenal.

Can you tell us about your first deployments?

The places where we are rolling out first are where there are existing care management programs already in place, and these tend to be in areas such as oncology and immunology where higher-end drugs are being used. Talking robots are very new and different, so we wanted our contract structure to look as similar as possible to existing offerings. These were the areas where there were already contract types that we could follow into market. We have been rolling out the first several hundred units in the first half of 2016 and are bringing patients onto the platform by the end of this year.

What is your business model? Is it “Robots as a Service”?

In terms of the patient relationship, our robot is key. But in terms of our business model and contracts, we don’t think of our robot as the key piece of what we’re delivering. We use a service model for care management; our customers pay us on a per patient per month basis.

What does interaction with your service look like from the patient’s perspective?

If you want to see what the patient interaction with Mabu actually looks like, we have a short video at cataliahealth.com.

Once the patient plugs Mabu in, the robot comes alive and starts talking. The conversation starts off with greetings and small talk (such as “Good morning, great to see you!”) and then moves on to whatever issue is relevant to the patient at that point in time. Maybe this is simply to check in on whether the patient has taken their medication, or maybe the patient is at a point in their treatment where it’s common to experience certain side effects, and the conversation is about how best to mitigate those for that patient. It really depends on the particular condition or treatment the patient is dealing with. We do a lot of research on each condition before rolling the platform out to patients, in order to build an understanding of common treatment challenges into the application.

In the background, the conversation is being crafted in real time for that patient.

When Mabu first comes out of the box, we know a little about the patient’s medical condition — perhaps what drugs they are on — but we don’t know much else. So from that very first conversation we start learning about and adapting to the patient’s individual personality and the treatment issues they are facing. Mabu largely directs the conversations, but the patient has a lot of say in terms of where that conversation goes. As we build more conversations and more AI into the platform, we are able to craft appropriate conversations for the patient.

This will very quickly become applicable to a lot more drugs and a lot more disease states. Let’s look at side effects, for example. Our first conversations about side effects will be new, but there are many common side effects among drugs. So while we are starting out in just a handful of areas, our goal is to help any patient who’s dealing with a condition on an ongoing basis to better manage their care, and to provide information back to their caregivers so that they can be more effective in supporting them.

Is it valid to be concerned about robots being used to replace human companionship?

We certainly don’t think of this as robots replacing people; we think of it as robots augmenting people.

One of the big challenges in healthcare today is that there are not enough caregivers to deliver healthcare the way we need it. Almost half our population is managing a chronic disease in this country, and there are very similar rates in advanced nations around the world; if we look at the rate of people dealing with health issues on an ongoing basis, it approaches 2/3 to 3/4 of the population.

People might get to see their doctor for fifteen minutes every two months, but that’s not much time, and it’s not an effective way to provide the ongoing care that is needed. We simply don’t have enough people to manage healthcare the way we did 50 or 100 years ago.

Patients need reminders, and they need answers to all the little questions that come up — and that’s where technology like this comes in. We see our service as a way for the people who are providing health care — doctors, nurses, and other trained caregivers — to more effectively reach a larger group of patients. We are not trying to be people’s doctors, we are trying to help their doctor do a much more effective job.

What kind of feedback have you received so far?

Broadly speaking, the feedback has been very positive. People tend to like the interaction right from the very first conversation, and they like how Mabu adapts to them.

We have a great solution that we’ve shown can effectively help many patients, but we still have a lot to learn. We are really excited about the amount of data that we’re going to be getting back from hundreds of person-months of interaction with our platform this year, and how we’re going to use that to improve conversations and personalize them to every patient.

Thanks to social platforms like Siri, Jibo and Amazon’s Echo, we are starting to get used to having conversations with our devices. But you’ve taken a very specialized path into the market. Why did you pick this pathway and business model?

Scalability — being able to provide care to a growing number of patients — is a big challenge in health care. I spent about a year before launching Catalia Health really digging into the US healthcare market to explore the business opportunities. We were thinking broadly around medication adherence and chronic disease management, talking to potential customers and trying to understand where there was a need for this kind of technology. The quick answer was that it is needed pretty much everywhere within the healthcare system. The question of how to provide healthcare in a cost-effective and scalable way is definitely a challenge here in the US, and also in most other nations in the world. We see an enormous opportunity for using technology to provide scalable personalized care.

Of all the robotics and AI movies that have come out in the past five or ten years, Robot and Frank offers a vision comes closest to what we’re doing. The goal of the robot in that movie was to help Frank live healthier by building a relationship with him. We have the same underlying premise in what we’re doing: our technology is focused on building a relationship with the patient, because once we can do that, then we can talk to them about their health care. By comparison, usage rates on healthcare apps are incredibly low; most patients don’t pick them up after the first or second try. But as it turns out, there are particular psychological aspects of how people interact with robots that make them really effective at helping to solve this challenge.

Do you see ways that other robotics companies can leverage what you’ve learned so far?

The broad lesson is to understand where there is a real human or business need. Asking “Where is there a problem that I can solve?” rather than asking “Where can I build a robot?” or “What market can I serve?”

It’s also important to understand what the existing marketplace looks like for those kinds of solutions right now, because the solution today may look very different. Our robot is an alternative to talking to a pharmacist on the phone, and it’s a very different solution, but understanding what the business model is for that kind of service, how those contracts work, who the players are in the space — I think that’s something that any company would be smart to take a look at and understand deeply before trying to compete in those markets.

You’re tackling one of the largest growing areas of our economy, and you’re doing it with a combination of data, AI and robotics. What do you think has changed in the past couple of years to make robotics a viable solution to a broader range of applications?

One of the biggest changes has been in the cost of building both hardware and software. Our robot is pretty simple; we’re not doing anything cutting edge in terms of the physical device that we’re building. But ten years ago producing our device might have cost 100 times what it does today, and that would have limited us to a small set of business models and it would have been very hard to make money.

With the cost of building the technology drastically lowered, it has enabled us to do something very different today than what we could have done five years ago. Today we can build cutting edge technology at a reasonable price point and therefore deliver a cost effective solution.

CATALIA HEALTH

Catalia Health is a patient health management system using social robotics. Founded in 2013 by Dr. Cory Kidd, Catalia Health builds on years of research into Human-Robot Interaction starting at MIT’s Media Lab and continuing with social robot startups like Intuitive Automata. In June 2015, Khosla Ventures led a $1.25 million seed round in Catalia Health for the first trial customer engagements. Catalia Health is on a mission to address both sides of the healthcare equation: improving patients’ health and extending the capabilities and efficiency of healthcare companies.

SILICON VALLEY ROBOTICS

Silicon Valley Robotics is the industry group for robotics and AI companies in the Greater San Francisco Bay Area, and is a not- for-profit (501c6) that supports innovation and commercialization of robotics technologies. We host the Silicon Valley Robot Block Party, networking events, investor forums, a directory, and a jobs board, and we provide additional services and information for members, such as these reports.

We’ll be releasing additional essays from the reports every week or so. You can read full reports by visiting the website.

If you enjoyed this article you may also want to read:

• HAX takes robotics to market in 2017
• Five women robotics experts from Silicon Valley share their stories
• $18.867 billion paid to acquire 50 robotics companies in 2016

See all the latest robotics news on Robohub, or sign up for our weekly newsletter.

In this fascinating animation from Oxford Sparks, we take a look at how statistics and computer science can be used to make machines that learn for themselves, without being explicitly programmed.

Machine learning is a burgeoning breed of artificial intelligence (AI), and it’s all around us already; on our phones, powering social networks, helping the police and doctors, scientists and mayors. But how does it work? Enjoy the video below, and visit Oxford Sparks to discover more science and research from Oxford University.

You might also enjoy the following posts about AI and AI/robotics policy:

• Brian Cox presents Science Matters: Machine learning and artificial intelligence
• MIT Media Lab to participate in $27 million initiative on AI ethics and governance
• How artificial intelligence is changing our Christmas shop
• White House report: Artificial intelligence, automation, and the economy
• White House launches public workshops on AI issues

See all the latest robotics news on Robohub, or sign up for our weekly newsletter.

We’re beginning to see more and more jobs being performed by machines, even creative tasks like writing music or painting can now be carried out by a computer.

But how and when will machines be able to explain themselves? Should we be worrying about an artificial intelligence taking over our world or are there bigger and more imminent challenges that advances in machine learning are presenting here and now?

Join Professor Brian Cox, the Royal Society Professor of Public Engagement, as he brings together experts on AI and machine learning to discuss key issues that will shape our future. Panelists include:

• Professor Jon Crowcroft FRS, Marconi Professor of Networked Systems at the University of Cambridge
• Professor Joanna Bryson, Reader in AI Ethics, University of Bath and Affiliate, Center for Information Technology Policy, Princeton University
• Professor Sabine Hauert, Lecturer in Robotics at the University of Bristol, and President and Co-Founder of Robohub.org
• Professor Murray Shanahan, Professor of Cognitive Robotics at Imperial College London
• Professor Peter Donnelly FRS FMedSci, Director, Wellcome Trust Centre for Human Genetics, and Professor of Statistical Science, University of Oxford

https://www.youtube.com/watch?v=IunNpGGt3H8

You might also enjoy the following articles about AI and AI/robotics policy:

• MIT Media Lab to participate in $27 million initiative on AI ethics and governance
• How artificial intelligence is changing our Christmas shop
• White House report: Artificial intelligence, automation, and the economy
• White House launches public workshops on AI issues
• White House releases reports focusing on opportunities and challenges of AI

See all the latest robotics news on Robohub, or sign up for our weekly newsletter.

Image: Gerd Altmann

The MIT Media Lab and the Berkman Klein Center for Internet and Society at Harvard University will serve as the founding anchor institutions for a new initiative aimed at bridging the gap between the humanities, the social sciences, and computing by addressing the global challenges of artificial intelligence (AI) from a multidisciplinary perspective.

“Artificial intelligence agents will impact every part of our lives in every society on Earth. Technology and commerce will see to that,” says Alberto Ibargüen, president and CEO of the John S. and James L. Knight Foundation, which is among those supporting the initiative.

Initially funded with $27 million from the Knight Foundation; LinkedIn co-founder Reid Hoffman; the Omidyar Network; the William and Flora Hewlett Foundation; and Jim Pallotta, founder of the Raptor Group, the Ethics and Governance of Artificial Intelligence Fund’s mission is to catalyze global research that advances AI for the public interest, with an emphasis on applied research and education. The fund will also seek to advance public understanding of AI.

“AI’s rapid development brings along a lot of tough challenges,” explains Joi Ito, director of the MIT Media Lab. “For example, one of the most critical challenges is how do we make sure that the machines we ‘train’ don’t perpetuate and amplify the same human biases that plague society? How can we best initiate a broader, in-depth discussion about how society will co-evolve with this technology, and connect computer science and social sciences to develop intelligent machines that are not only ‘smart,’ but also socially responsible?”

What makes this new initiative different and necessary is that it’s aimed at transcending barriers and breaking down silos among disciplines. As founding academic institutions, the Media Lab and Berkman Klein Center, along with other potential collaborators from the public and private sectors, will act as a mechanism to reinforce cross-disciplinary work and encourage intersectional peer dialogue and collaboration.

The fund — projected to operate with a phased approach over the next several years — will complement and collaborate with existing efforts and communities, such as the upcoming public symposium “AI Now,” which is scheduled for July 10 at the MIT Media Lab. The fund will also oversee an AI fellowship program, identify and provide support for collaborative projects, build networks out of the people and organizations currently working to steer AI in directions that help society, and also convene a “brain trust” of experts in the field.

A collaborative network

The Media Lab and the Berkman Klein Center for Internet and Society will leverage a network of faculty, fellows, staff, and affiliates who will collaborate on unbiased, sustained, evidenced-based, solution-oriented work that cuts across disciplines and sectors. This research will include questions that address society’s ethical expectations of AI, using machine learning to learn ethical and legal norms from data, and using data-driven techniques to quantify the potential impact of AI, for example, on the labor market.

Work of this nature is already being undertaken at both institutions. The Media Lab has been exploring some of the moral complexities associated with autonomous vehicles in the Scalable Cooperation group, led by Iyad Rahwan. And the Personal Robots group, led by Cynthia Breazeal, is investigating the ethics of human-robot interaction.

“AI could be as big a disruptor to the world as the Industrial Revolution was in the 18th and 19th centuries,” Rahwan says. He cites transportation systems and employment as among the areas that will likely be affected by automation and AI. “What we need is something that puts our entire society in the control loop of these systems … technologists, engineers, the public, ethicists, cognitive scientists, economists, legal scholars, anthropologists, faith leaders, government regulators — everyone crucial to protecting the public interest.”

“Artificial Intelligence provides the potential for deeply personalized learning experiences for people of all ages and stages,” says Breazeal, who emphasizes the need for AI to reach people in developing nations and underserved populations. But she adds that it is also “a kind of double-edged sword. What should it be learning and adapting to benefit you? And what should it do to protect your privacy and your security?”

Shared goals and governance

The Berkman Klein Center has been working to develop public interest-oriented solutions to many of the challenges of the digital age, incubating programs such as Creative Commons and the Digital Public Library of America. And, it is currently collaborating with the Media Lab on the “Assembly” program, which gathers high-level developers and tech industry professionals for a rigorous three-week course at Harvard University, followed by a 12-week collaborative development period to explore difficult problems in cybersecurity.

“The thread running through these otherwise disparate phenomena is a shift of reasoning and judgment away from people,” says Jonathan Zittrain, co-founder of the Berkman Klein Center and professor of law and computer science at Harvard University. “Sometimes that’s good, as it can free us up for other pursuits and for deeper undertakings. And sometimes it’s profoundly worrisome, as it decouples big decisions from human understanding and accountability. A lot of our work in this area will be to identify and cultivate technologies and practices that promote human autonomy and dignity rather than diminish it.”

The Ethics and Governance of Artificial Intelligence Fund will be governed by a small board, consisting of leadership from each participating foundation and institution. Ito says that convening all perspectives now is essential. “Instead of setting up an institution, we’ve decided to create a dynamic network. There are a number of areas where the deployment of machine learning is already starting to pose questions that are best answered in an interdisciplinary way. This project that we’re all embarking on is just the beginning.”

You might also enjoy the following articles about AI and AI/robotics policy:

• How artificial intelligence is changing our Christmas shop
• White House report: Artificial intelligence, automation, and the economy
• White House launches public workshops on AI issues
• White House releases reports focusing on opportunities and challenges of AI
• White House economic report looks to robotics for the future
• We the Geeks: Whitehouse hangouts celebrate robots, maker culture and more

See all the latest robotics news on Robohub, or sign up for our weekly newsletter.

Having been involved in roboethics for some years, I was delighted when the IEEE launched its initiative on Ethical Considerations in AI and Autonomous Systems, early this year. Especially so because of the reach and traction that the IEEE has internationally. (Up until now most ethics initiatives have been national efforts – with the notable exception of the 2006 EURON roboethics roadmap.)

Even better, this is an initiative of the IEEE standards association – the very same that gave the world Wi-Fi (aka IEEE 802.11) 19 years ago. So when I was asked to get involved I jumped at the chance and became co-chair of the General Principles committee. I found myself in good company; many great people I knew but more I did not, and it was a real pleasure when we met face to face in The Hague at the end of August.

Most of our meetings were conducted by phone and it was a very demanding timetable. From nothing to our first publication: Ethically Aligned Design a few days ago is a remarkable achievement, which I think wouldn’t have happened without the extraordinary energy and enthusiasm of the initiative’s executive director John Havens.

I’m not going to describe what’s in that document here; instead I hope you will read it – and return comments. This document is not set in stone, it is – in the best traditions of the RFCs which defined the Internet – a Request for Input

But there are a couple of aspects I will highlight. Like its modest but influential predecessor, the EPSRC/AHRC principles of robotics, the IEEE initiative is hugely multi-disciplinary. It draws heavily from industry and academia, and includes philosophers, ethicists, lawyers, social scientists – as well as engineers and computer scientists – and significantly a number of diplomats and representatives from transnational bodies like the United Nations, US state department and the WEF. This is so important – if the work of this initiative is to make a difference it will need influential advocates. Equally important is that this is not a group dominated by old white men. There are plenty of those for sure, but I reckoned 40% women (should be 50% though!) and plenty of post-docs and PhD students too.

Equally important, the work is open. The publications are released under the creative commons licence. Likewise active membership is open. If you care about the issues and think you could contribute to one or more of the committees – or even if you think there’s a whole area of concern missing that needs to a new committee – get in touch!

The White House released a new report on the ways that artificial intelligence will transform the US economy over the coming years and decades.

Below is shortened version from the original post.

On December 20, In order to ready the United States for a future in which artificial intelligence (AI) plays a growing role, the White House released a report on Artificial Intelligence, Automation, and the Economy. This report follows up on the Administration’s previous report, Preparing for the Future of Artificial Intelligence, which was released in October 2016, and which recommended that the White House publish a report on the economic impacts of artificial intelligence by the end of 2016.

Accelerating AI capabilities will enable automation of some tasks that have long required human labor. These transformations will open up new opportunities for individuals, the economy, and society, but they will also disrupt the current livelihoods of millions of Americans. The new report examines the expected impact of AI-driven automation on the economy, and describes broad strategies that could increase the benefits of AI and mitigate its costs.

AI-driven automation will transform the economy over the coming years and decades. The challenge for policymakers will be to update, strengthen, and adapt policies to respond to the economic effects of AI.

Although it is difficult to predict these economic effects precisely, the report suggests that policymakers should prepare for five primary economic effects:

• Positive contributions to aggregate productivity growth;
• Changes in the skills demanded by the job market, including greater demand for higher-level technical skills;
• Uneven distribution of impact, across sectors, wage levels, education levels, job types, and locations;
• Churning of the job market as some jobs disappear while others are created; and
• The loss of jobs for some workers in the short-run, and possibly longer depending on policy responses.

In the cases where it is possible to direct mitigations to particularly affected places and sectors, those approaches should be pursued. But more generally, the report suggests three broad strategies for addressing the impacts of AI-driven automation across the whole U.S. economy:

1. Invest in and develop AI for its many benefits;
2. Educate and train Americans for jobs of the future; and
3. Aid workers in the transition and empower workers to ensure broadly shared growth.

Read the entire White House summary here.

Download the PDF report here.

You can also read these articles about AI and the White House:

• White House launches public workshops on AI issues
• White House releases reports focusing on opportunities and challenges of AI
• White House economic report looks to robotics for the future
• We the Geeks: Whitehouse hangouts celebrate robots, maker culture and more

See all the latest robotics news on Robohub, or sign up for our weekly newsletter.

British consumers are expected to spend £280 each on gifts over the weeks leading up to Christmas. More than half of these purchases will take place online. Almost a third of people will rely on online reviews to make their buying decisions, although recommendations from friends and family are still the main source of persuasion.

Online shopping is estimated to rise by 24% by the end of this year. However, as consumers are looking for more sensory and immersive shopping experiences, the pressure is on for online retailers to find new ways to excite customers and keep them satisfied – and artificial intelligence (AI) is the new technology they will use. It is allowing businesses to analyse customer behaviour, predict consumer wants and offer tailored customer experiences. In short, AI is expected to make online experiences altogether more personal.

Getting personal

There are already many ways that retailers use AI to interact with their customers. This type of AI is primarily based on learning customer preferences, behaviours and providing tailored recommendations at a mass scale – also referred to as a mass customisation.

The online fashion retailer Stitch Fix gives customers five curated pieces of clothing each month and the customer then decides which pieces to keep. The selection of items is based on customer surveys, Pinterest boards, weather patterns and personal notes to the stylist. From this data, algorithms help the personal stylists to pick out the items the customer is predicted to like the most.

With extensive choice online, retailers try to simplify the buying process. If you are in a shop you are familiar with, browsing is easier. You can glance at a jumper on your way to the toy section, or spot a tea towel down one aisle while spontaneously examining a Star Wars figurine in another.

That dynamic doesn’t work on the internet and so one of the key objectives of using AI in online retail is to assist consumers find what they are looking for and narrow down their potential choice. A new study showed that once consumers have made a decision to purchase within a product category, having a smaller range of potential products to choose from reduces choice overload.

If you are more of a visual shopper, you may like to get your inspiration from Snap Fashion. Using a visual search engine, items from more than 16,000 brands are presented to the consumer based on photos found online or taken on their smart phone. From August 2017, this experience – Snap Fashion InStore – is expected to come to a high street changing room near you.

Chat bots

The next phase of AI in retail will go beyond personalised recommendations to having conversations with customers.

The term “conversational commerce” was coined in 2015 by Chris Messina who leads the development and expansion of Uber’s partner ecosystem. Conversational commerce arises from the convergence of messaging apps, natural language interfaces and brands allowing consumers to chat, message and talk with brands and services with the help of chat bots.

At peer-to-peer online insurance company, Lemonade, customers use the AI bot, Maya, to design their individualised insurance policy. Completed in minutes via the Lemonade conversational app. Customers can also talk to Maya to submit a claim using the in-app video recording feature to describe the incident.

Conversational commerce may move us from being helpless recipients of the work of AI – say, being presented with products we didn’t know we needed – to having direct conversations with AI and becoming active partners in the process. In fact, we may not even be able to tell if we are talking with an AI bot or a human.

Now, few people will buy their mum a Lemonade insurance policy for Christmas, but if you want to get her a new winter jacket, then the outdoor clothing company, The North Face may have the answer. It offers direct interaction between customers and IBM’s Watson-supported natural language questions-and-answers to help identify the most suitable item. Customers respond to a series of questions about how, when and where they expect to use the jacket and particular features that are important to them, for example a pocket that fits the size of your new mobile phone. Based on this information, weather forecasts and delivery requirements, a ranked selection of options is presented.

For running in Snowdonia during December. The North Face/screenshot

This kind of AI-driven personalised shopping is capable of challenging, possibly even replacing, the guidance provided by human in-store sales assistants. It is knowledgeable, fast and able to draw on multiple data points to provide personalised guidance to customers quickly and efficiently. It also goes without saying that AI assistants are less likely to be hungover, bored or tut loudly at your insane questions about a toaster’s bagel feature.

Mass customisation

The use of AI in retail offers opportunities for mass customisation and helps customers to narrow down their choices faster. That has clear benefits for those struggling to grind through the Christmas shopping list. And for companies, it can provide a consistency of service which creates more “sticky” customer experiences online. This is crucial in the face of hyper-competition across all retail sectors.

The trouble is that it also reduces the element of discovery and exploration, which consumers enjoy about the shopping experience, especially during the festive season. It is a streamlining of the buying process, using machine learning and predictive analytics to nudge consumers towards a particular purchase decision.

Consumers’ interactions with brands will become highly personalised when conversational commerce gains traction. These AI systems can be of great help for people looking to solve simple and specific tasks, like buying insurance or gaining technical advice. But for more complex, emotive and subjective buying decisions – like buying gifts for our loved ones – we may still seek that special human touch and personal opinion from friends, family, or even (hungover) sales assistants.

This article was originally published on The Conversation. Read the original article.

MIT researchers have developed a new approach to training speech-recognition systems that doesn’t depend on transcription. Instead, their system analyzes correspondences between images and spoken descriptions of those images, as captured in a large collection of audio recordings. Image: MIT News

Speech recognition systems, such as those that convert speech to text on cellphones, are generally the result of machine learning. A computer pores through thousands or even millions of audio files and their transcriptions, and learns which acoustic features correspond to which typed words.

But transcribing recordings is costly, time-consuming work, which has limited speech recognition to a small subset of languages spoken in wealthy nations.

At the Neural Information Processing Systems conference this week, researchers from MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) are presenting a new approach to training speech-recognition systems that doesn’t depend on transcription. Instead, their system analyzes correspondences between images and spoken descriptions of those images, as captured in a large collection of audio recordings. The system then learns which acoustic features of the recordings correlate with which image characteristics.

“The goal of this work is to try to get the machine to learn language more like the way humans do,” says Jim Glass, a senior research scientist at CSAIL and a co-author on the paper describing the new system. “The current methods that people use to train up speech recognizers are very supervised. You get an utterance, and you’re told what’s said. And you do this for a large body of data.

“Big advances have been made — Siri, Google — but it’s expensive to get those annotations, and people have thus focused on, really, the major languages of the world. There are 7,000 languages, and I think less than 2 percent have ASR [automatic speech recognition] capability, and probably nothing is going to be done to address the others. So if you’re trying to think about how technology can be beneficial for society at large, it’s interesting to think about what we need to do to change the current situation. And the approach we’ve been taking through the years is looking at what we can learn with less supervision.”

Joining Glass on the paper are first author David Harwath, a graduate student in electrical engineering and computer science (EECS) at MIT; and Antonio Torralba, an EECS professor.

Visual semantics

The version of the system reported in the new paper doesn’t correlate recorded speech with written text; instead, it correlates speech with groups of thematically related images. But that correlation could serve as the basis for others.

If, for instance, an utterance is associated with a particular class of images, and the images have text terms associated with them, it should be possible to find a likely transcription of the utterance, all without human intervention. Similarly, a class of images with associated text terms in different languages could provide a way to do automatic translation.

Conversely, text terms associated with similar clusters of images, such as, say, “storm” and “clouds,”  could be inferred to have related meanings. Because the system in some sense learns words’ meanings — the images associated with them — and not just their sounds, it has a wider range of potential applications than a standard speech recognition system.

To test their system, the researchers used a database of 1,000 images, each of which had a recording of a free-form verbal description associated with it. They would feed their system one of the recordings and ask it to retrieve the 10 images that best matched it. That set of 10 images would contain the correct one 31 percent of the time.

“I always emphasize that we’re just taking baby steps here and have a long way to go,” Glass says. “But it’s an encouraging start.”

The researchers trained their system on images from a huge database built by Torralba; Aude Oliva, a principal research scientist at CSAIL; and their students. Through Amazon’s Mechanical Turk crowdsourcing site, they hired people to describe the images verbally, using whatever phrasing came to mind, for about 10 to 20 seconds.

For an initial demonstration of the researchers’ approach, that kind of tailored data was necessary to ensure good results. But the ultimate aim is to train the system using digital video, with minimal human involvement. “I think this will extrapolate naturally to video,” Glass says.

Merging modalities

To build their system, the researchers used neural networks, machine-learning systems that approximately mimic the structure of the brain. Neural networks are composed of processing nodes that, like individual neurons, are capable of only very simple computations but are connected to each other in dense networks. Data is fed to a network’s input nodes, which modify it and feed it to other nodes, which modify it and feed it to still other nodes, and so on. When a neural network is being trained, it constantly modifies the operations executed by its nodes in order to improve its performance on a specified task.

The researchers’ network is, in effect, two separate networks: one that takes images as input and one that takes spectrograms, which represent audio signals as changes of amplitude, over time, in their component frequencies. The output of the top layer of each network is a 1,024-dimensional vector — a sequence of 1,024 numbers.

The final node in the network takes the dot product of the two vectors. That is, it multiplies the corresponding terms in the vectors together and adds them all up to produce a single number. During training, the networks had to try to maximize the dot product when the audio signal corresponded to an image and minimize it when it didn’t.

For every spectrogram that the researchers’ system analyzes, it can identify the points at which the dot-product peaks. In experiments, those peaks reliably picked out words that provided accurate image labels — “baseball,” for instance, in a photo of a baseball pitcher in action, or “grassy” and “field” for an image of a grassy field.

In ongoing work, the researchers have refined the system so that it can pick out spectrograms of individual words and identify just those regions of an image that correspond to them.

“Possibly, a baby learns to speak from its perception of the environment, a large part of which may be visual,” says Lin-shan Lee, a professor of electrical engineering and computer science at National Taiwan University. “Today, machines have started to mimic such a learning process. This work is one of the earliest efforts in this direction, and I was really impressed when I first learned of it.”

“Perhaps even more exciting is just the question of how much we can learn with deep neural networks,” adds Karen Livescu, an assistant professor at the Toyota Technological Institute at the University of Chicago. “The more the research community does with them, the more we realize that they can learn a lot from big piles of data. But it is hard to label big piles of data, so it’s really exciting that in this work, Harwath et al. are able to learn from unlabeled data. I am really curious to see how far they can take that.”

Read the paper here.

In April 2016, British Standards Institution (BSI) published the world’s first ethical standard for the design, production, sale, and usage of social robots. “BS 8611: 2016 Robots and robotic devices” gives guidelines to identify ethical hazards and reduce associated risks to acceptable levels. BS 8611 could be crucial in determining the nature of emerging human-robot relationships. However, the publication also raises many new questions. Are there any conflicts with ISO 13482 when incorporating BS 8611 into the design of safer robots? Should we consider regulating robot ethics? If so, how?

In this interview, we invite Prof. Joanna Bryson to share her insights on how to regulate future human-robot relationships.

Date: December 9^th 2016.

Interviewee: Prof. Joanna J. Bryson, Reader at Department of Computer Science, University of Bath, and affiliate of the Center for Information Technology and Policy, Princeton University.

Interviewer: Dr. Yueh-Hsuan Weng, Co-founder of ROBOLAW.ASIA and Fellow at CLAST AI & Law Research Committee (Beijing) and Tech and Law Center (Milan).

Prof. Joanna Bryson. Image: University of Bath

Q1 WENG: Thank you for agreeing to an interview with us. Can you tell us a little about your background?

BRYSON: My primary academic interest is comparative intelligence.  I was particularly interested in animal behavior as a student, and took a liberal arts degree from Chicago with a major in the Behavioural Sciences.  I now have two degrees in Artificial Intelligence (including my PhD at MIT) and another masters degree in Psychology (both masters are from Edinburgh.)  I’m currently an Associate Professor at Bath where I founded the Intelligent Systems group, and organized Artificial Models of Natural Intelligence, which is basically my own laboratory.

During my PhD I was working on a humanoid robot project and was astonished that people thought they had moral obligations to the robot, even though it did not function at all.  It was just a bunch of motors welded together in the shape of a human.  I began publishing about this problem, thinking that people didn’t know enough about AI, but in recent years I’ve realised they don’t understand enough about ethics.  They are confused about what is really the root of ethical obligation, and they do not understand the nature of artefacts.  They do not understand that we can construct both AI and ethics so that we either are or aren’t obliged to AI.  I have become well-known for arguing that we are ethically obliged to construct AI we are not ethically obliged to.

 Q2 WENG: What is the meaning of “ethical hazards”? Except deception and anthropomorphism, what are other ethical hazards in BS 8611: 2016?

BRYSON: I have to admit I haven’t memorised that document and don’t have time to go back to read it now.  Maybe you are talking about a moral hazard?  A moral hazard is something likely to make you do the wrong thing.  In particular, I have argued that attributing ethical status to AI is a moral hazard, because it allows us to escape the responsibilities we ourselves have as the manufacturers, builders and operators of AI.  This is particularly obvious in the case of military weapons – we have enough trouble with distributing blame between soldiers, their commanders, and the government that commissions them.  Making weapons themselves “sinks” for guilt is not honest or beneficial.  But there are also moral hazards associated with owning “friends” and “lovers” – if children relate mostly with things they own, how will they grow to be able to be equal partners with their spouse or coworkers, or members of a democracy?

Q3 WENG: Some members from Bristol Robotics Laboratory are particularly interested in “embedded intelligence”? Why is this concept important? Does it have any impact on BS 8611: 2016?

BRYSON:  I’m sorry, I don’t think it matters particularly whether AI is embedded in an artefact or “channelled” through the cloud, except obviously there are privacy issues when we connect AI onto the Internet or into any shared digital memory.  In fact, one of the ways I recommend ensuring that we have no ethical obligation to robots or other AI is to ensure there is never a unique copy: that it should be backed up to remote sites at all times.

Q4 WENG: Prof. John Holland said, “..an internal model allows a system to look ahead to the future consequences of current actions, without actually committing itself to those actions”. Can you tell us about the concept of “internal model”? Does a passive dynamic robot have its own internal model as well?

BRYSON:  Ha, that’s a great question!  The idea of an internal model is just a representation that allows, as Prof Holland says, a means for planning around and anticipating the world.  Rodney Brooks famously recommended that AI should not have a model, should use the world as it’s own best model, but in fact that is not practical. There’s not enough information in a sensory signal to disambiguate a world sufficiently to create animal-like intelligence, we need prior expectations to interpret the sensory input.  Still, Brooks was a genius of AI and the field moved forward much faster when we realised we could use minimalist representations; just those necessary for action.  We realised that humans “flush” our memory quickly if sensory information contradicts it, and operate on partial knowledge all the time.  Previously we though AI had to have perfect knowledge so it could make perfect plans, but now we understand that that isn’t tractable.

So does a passive dynamic have an internal model? No.  It is a model of an animal walking system, but to some extent it embodies expectations about for example gravity, but by and large it can do no planning or control. It arrives only where it goes and never knows that it has gotten there or been anywhere else, so certainly cannot anticipate anything.  But its design anticipates particular regularities about the world, so its designers had models that are partly embodied in the walker.

Q5 WENG: I conducted an empirical case study on humanoid robots and regulation with researchers from Waseda University. We found that inside “Tokku” Special Zones there were strong demands for regulating robotics in road traffic, privacy, tele-communication, safety, tax, highway traffic, etc… However, it is not clear that whether robot ethics should be regulated. Do you have any comments on this?

BRYSON: Unfortunately I also haven’t had time to read your study yet, but I can talk a bit about regulation.  There are two important things to recognise.  First, government should not be thought of as an imposition, particularly in a democracy, but even in an autocracy government only functions to the extent it is tolerated.  Government and regulation is the way we citizens coordinate our desires and vision.  Many people in technology seem to think of regulation only as limits on their freedom, and indeed even the word compromise can be used in both a negative and a positive way.  But technologists need to understand that when people come together and work together, they often create something of mutual benefit, public goods that are greater than the summed contribution of those who are cooperating.  Regulations can include those that mandate investment of government resources (our taxes) in innovation.  They can be liberating as well as constraining. Coming to agreements with fellow citizens about essential matters like privacy and the distribution of wealth that AI has huge impact on is essential, and yes, agreements require compromise, but it is through these that we create peace, stability, and affluence.

Having said all that, what I get from surveys of citizens is not information about what regulations would really work.  Governance and regulation are extremely complex and have been studied for centuries; it is as crazy to suggest that ordinary citizens construct regulations as that they construct nuclear weapons or that random people be drafted to represent a country in the World Cup or the Olympics.  We need to respect expertise as well as the desires of ordinary citizens.  What the fears and goals of citizens tell us is how easy or hard it would be to implement particular regulations, how likely they are to be accepted and respected.  Most people on the planet believe that their ancestors still have concerns with their own behaviour, but I wouldn’t want to build policy relying on dead ancestors as actors.  But it does make sense to know people think this if you are governing.

 Q6 WENG: What’s your definition of “ethical robots”? Should we teach them right from wrong?

BRYSON:  I think it makes more sense to program than to teach. We have enough trouble teaching citizens and governments to do good.  Most actions have ethical consequences and there is no problem in building AI to recognise and reason about these, but I would never recommend holding the AI as morally responsible, even though I think AI will necessarily take actions with moral consequences.  As the British EPSRC Principles of Robotics make clear, the responsibility for commercial products lies either in their manufacturer (if they fail to meet an advertised specification) or in their owner / operator (if they are improperly used.)  There are ethical implementations of AI, for example protecting the privacy of the individual, or limiting the damage an individual can do with an autonomous vehicle, but the AI itself should not be designated the responsible party in any circumstance.

 Q7 WENG: Do you believe that Asimov’s Three Laws of Robotics will be guiding principles to ensure the safety and security of human-robot interactions?

BRYSON:  Asimov’s laws certainly impact how people think about AI ethics; the first three (of five) Principles of Robotics reflect (and correct) Asimov’s laws.  But we are working hard to make people understand that these laws are by no means necessary or sufficient for AI ethics.  First of all, Asimov’s laws are intractable. It’s computationally impossible to implement them as described in the books, we can never know all the consequences of our actions.  Second, the books themselves show repeatedly that the laws are insufficient to create truly moral behaviour, that’s the fun of the short stories.  And third, Asimov like most science fiction writers uses AI as a stand-in for humans, for examining what it is to be human. The laws describe the robot as the responsible party, and that’s as I said earlier a mistake.

 Q8 WENG: Do you agree that robots should not be legal persons? Why?

BRYSON:  I don’t know for sure whether people shall make them legal persons, but I am certain that they should not, and in the long run will not be.  A legal person is something that is subject to human justice, and I do not believe it is possible in a well-designed system to create the kind of suffering that is intrinsic to human punishment.  There is no necessity for robots or AI to know shame, pain, regret for the passage of time.  These things are completely intrinsic to being human, we do not consider people with deficits in these concerns to be mentally healthy.  This is because we have evolved to be social organisms and there is no way to extract these properties from our intelligence.  But well-designed AI is modular, anything that is deliberately added by a designer can be removed.  So that’s one reason it would be incoherent for AI to be a legal person.  Another reason is that there is no discrete unit of AI or robots like there is for humans.  We know what it is to be a human being, though we are sometimes a little challenged by people who are brain dead or conjoined twins.  But AI cannot be neatly counted like humans, and therefore the idea of taxing a robot like a person is ludicrous.  And, again, it is a moral hazard.  The companies who decide to employ robotics should be the ones who have legal and economic liability for whatever their robots do.  We should not let them wiggle out of their responsibility because of our love for science fiction or our deep desire to construct perfect, obedient, cyber “children”.

Q9 WENG: What is the definition of “well-designed robots” in your opinion?

BRYSON: A robot is well-designed if the company that produces it is willing to take full responsibility for its actions, and further if governments are willing to license it as safe for human society.  That second requirement is necessary because some corporations have more resources than many countries, so may be willing to accept risks that individual citizens would prefer not to be subject to.

Q10 WENG: Finally, what will be the biggest challenge to developing a human-robot co-existence Society in the following decade?

BRYSON:  We still have many challenges.  I think we will solve dexterity and physical safety before we solve cyber security.  I don’t think we have a full grasp yet about what impact we are already having on human society and human autonomy.  It’s difficult to measure this because so many things are changing at once; we also have the problems of climate change, of massive increases in human numbers and access to communication.  We are currently in a period of grotesque income inequality and political polarisation – some people blame these on AI, but in fact this happened only a century ago, and probably contributed to the two world wars and the great depression. So getting on top of understanding how we are influencing each other and compromising our health with our planets’ is essential to answering this question.

If you enjoyed this article, you may also also enjoy:

• One being for two origins: A new perspective on roboethics
• Considering robot care, ethics & future tech at #ERW2016 European Robotics Week
• On the ethics of research in robotics, with Raja Chatila
• A European perspective on robot law: Interview with Mady Delvaux-Stehres
• Open Roboethics initiative delivers statement to United Nations CCW

See all the latest robotics news on Robohub, or sign up for our weekly newsletter.

Image: Massachusetts Institute of Technology

MIT researchers and their colleagues have developed a new computational model of the human brain’s face-recognition mechanism that seems to capture aspects of human neurology that previous models have missed.

The researchers designed a machine-learning system that implemented their model, and they trained it to recognize particular faces by feeding it a battery of sample images. They found that the trained system included an intermediate processing step that represented a face’s degree of rotation — say, 45 degrees from center — but not the direction — left or right.

This property wasn’t built into the system; it emerged spontaneously from the training process. But it duplicates an experimentally observed feature of the primate face-processing mechanism. The researchers consider this an indication that their system and the brain are doing something similar.

“This is not a proof that we understand what’s going on,” says Tomaso Poggio, a professor of brain and cognitive sciences at MIT and director of the Center for Brains, Minds, and Machines (CBMM), a multi-institution research consortium funded by the National Science Foundation and headquartered at MIT. “Models are kind of cartoons of reality, especially in biology. So I would be surprised if things turn out to be this simple. But I think it’s strong evidence that we are on the right track.”

Indeed, the researchers’ new paper includes a mathematical proof that the particular type of machine-learning system they use, which was intended to offer what Poggio calls a “biologically plausible” model of the nervous system, will inevitably yield intermediary representations that are indifferent to angle of rotation.

Poggio, who is also a primary investigator at MIT’s McGovern Institute for Brain Research, is the senior author on a paper describing the new work, which appeared today in the journal Computational Biology. He’s joined on the paper by several other members of both the CBMM and the McGovern Institute: first author Joel Leibo, a researcher at Google DeepMind, who earned his PhD in brain and cognitive sciences from MIT with Poggio as his advisor; Qianli Liao, an MIT graduate student in electrical engineering and computer science; Fabio Anselmi, a postdoc in the IIT@MIT Laboratory for Computational and Statistical Learning, a joint venture of MIT and the Italian Institute of Technology; and Winrich Freiwald, an associate professor at the Rockefeller University.

Emergent properties

The new paper is “a nice illustration of what we want to do in [CBMM], which is this integration of machine learning and computer science on one hand, neurophysiology on the other, and aspects of human behavior,” Poggio says. “That means not only what algorithms does the brain use, but what are the circuits in the brain that implement these algorithms.”

Poggio has long believed that the brain must produce “invariant” representations of faces and other objects, meaning representations that are indifferent to objects’ orientation in space, their distance from the viewer, or their location in the visual field. Magnetic resonance scans of human and monkey brains suggested as much, but in 2010, Freiwald published a study describing the neuroanatomy of macaque monkeys’ face-recognition mechanism in much greater detail.

Freiwald showed that information from the monkey’s optic nerves passes through a series of brain locations, each of which is less sensitive to face orientation than the last. Neurons in the first region fire only in response to particular face orientations; neurons in the final region fire regardless of the face’s orientation — an invariant representation.

But neurons in an intermediate region appear to be “mirror symmetric”: That is, they’re sensitive to the angle of face rotation without respect to direction. In the first region, one cluster of neurons will fire if a face is rotated 45 degrees to the left, and a different cluster will fire if it’s rotated 45 degrees to the right. In the final region, the same cluster of neurons will fire whether the face is rotated 30 degrees, 45 degrees, 90 degrees, or anywhere in-between. But in the intermediate region, a particular cluster of neurons will fire if the face is rotated by 45 degrees in either direction, another if it’s rotated 30 degrees, and so on.

This is the behavior that the researchers’ machine-learning system reproduced. “It was not a model that was trying to explain mirror symmetry,” Poggio says. “This model was trying to explain invariance, and in the process, there is this other property that pops out.”

Neural training

The researchers’ machine-learning system is a neural network, so called because it roughly approximates the architecture of the human brain. A neural network consists of very simple processing units, arranged into layers, that are densely connected to the processing units — or nodes — in the layers above and below. Data are fed into the bottom layer of the network, which processes them in some way and feeds them to the next layer, and so on. During training, the output of the top layer is correlated with some classification criterion — say, correctly determining whether a given image depicts a particular person.

In earlier work, Poggio’s group had trained neural networks to produce invariant representations by, essentially, memorizing a representative set of orientations for just a handful of faces, which Poggio calls “templates.” When the network was presented with a new face, it would measure its difference from these templates. That difference would be smallest for the templates whose orientations were the same as that of the new face, and the output of their associated nodes would end up dominating the information signal by the time it reached the top layer. The measured difference between the new face and the stored faces gives the new face a kind of identifying signature.

In experiments, this approach produced invariant representations: A face’s signature turned out to be roughly the same no matter its orientation. But the mechanism — memorizing templates — was not, Poggio says, biologically plausible.

So instead, the new network uses a variation on Hebb’s rule, which is often described in the neurological literature as “neurons that fire together wire together.” That means that during training, as the weights of the connections between nodes are being adjusted to produce more accurate outputs, nodes that react in concert to particular stimuli end up contributing more to the final output than nodes that react independently (or not at all).

This approach, too, ended up yielding invariant representations. But the middle layers of the network also duplicated the mirror-symmetric responses of the intermediate visual-processing regions of the primate brain.

“I think it’s a significant step forward,” says Christof Koch, president and chief scientific officer at the Allen Institute for Brain Science. “In this day and age, when everything is dominated by either big data or huge computer simulations, this shows you how a principled understanding of learning can explain some puzzling findings.”

“They’re very careful,” Koch adds. “They’re only looking at the feed-forward pathway — in other words, the first 80, 100 milliseconds. The monkey opens its eyes, and within 80 to 100 milliseconds, it can recognize a face and push a button signaling that. The question is what goes on in those 80 to 100 milliseconds, and the model that they have seems to explain that quite well.”

If you enjoyed this article, you might also be interested in:

• The human brain vs. the digital brain: A case for visual inspections
• Assistive robot operated via a brain-computer interface
• Virtual race: Competing in Brain Computer Interface at Cybathlon
• Can a brain-computer interface convert your thoughts to text?

See all the latest robotics news on Robohub, or sign up for our weekly newsletter.

U.S. Sen. Ted Cruz (R-Texas), chairman of the Subcommittee on Space, Science, and Competitiveness will convene a hearing today, 30 November, at 2:30 p.m. EST on “The Dawn of Artificial Intelligence.” The hearing will conduct a broad overview of the state of artificial intelligence, including policy implications and effects on commerce.

On October 12, 2016, the White House Office of Science and Technology Policy released a report entitled, Preparing for the Future of Artificial Intelligence. The report outlined opportunities for artificial intelligence, including recommendations on how the technology can be used to advance social good and improve government operations. The report also proposed 23 policy recommendations and a companion strategic plan that identified priorities for federally-funded research and development in artificial intelligence.

At today’s Senate hearing, which will be livestreamed, five experts will be present:

• Dr. Eric Horvitz, Interim Co-Chair, Partnership on Artificial Intelligence; Managing Director, Microsoft Research Lab
• Dr. Andrew Moore, Dean, School of Computer Science, Carnegie Mellon University
• Dr. Andrew Futreal, Professor, Department of Genomic Medicine, University of Texas MD Anderson Cancer Center
• Mr. Greg Brockman, Cofounder and Chief Technology Officer, OpenAI
• Dr. Steve Chien, Senior Research Scientist, Autonomous Space Systems and Technical Group Supervisor, Artificial Intelligence Group, NASA Jet Propulsion Laboratory, California Institute of Technology

* Witness list subject to change

The common, and recurring, view of the latest breakthroughs in artificial intelligence research is that sentient and intelligent machines are just on the horizon. Machines understand verbal commands, distinguish pictures, drive cars and play games better than we do. How much longer can it be before they walk among us?

The new White House report on artificial intelligence takes an appropriately skeptical view of that dream. It says the next 20 years likely won’t see machines “exhibit broadly-applicable intelligence comparable to or exceeding that of humans,” though it does go on to say that in the coming years, “machines will reach and exceed human performance on more and more tasks.” But its assumptions about how those capabilities will develop missed some important points.

As an AI researcher, I’ll admit it was nice to have my own field highlighted at the highest level of American government, but the report focused almost exclusively on what I call “the boring kind of AI.” It dismissed in half a sentence my branch of AI research, into how evolution can help develop ever-improving AI systems, and how computational models can help us understand how our human intelligence evolved.

The report focuses on what might be called mainstream AI tools: machine learning and deep learning. These are the sorts of technologies that have been able to play “Jeopardy!” well, and beat human Go masters at the most complicated game ever invented. These current intelligent systems are able to handle huge amounts of data and make complex calculations very quickly. But they lack an element that will be key to building the sentient machines we picture having in the future.

We need to do more than teach machines to learn. We need to overcome the boundaries that define the four different types of artificial intelligence, the barriers that separate machines from us – and us from them.

Type I AI: Reactive machines

The most basic types of AI systems are purely reactive and have the ability neither to form memories nor to use past experiences to inform current decisions. Deep Blue, IBM’s chess-playing supercomputer, which beat international grandmaster Garry Kasparov in the late 1990s, is the perfect example of this type of machine.

Deep Blue can identify the pieces on a chess board and know how each moves. It can make predictions about what moves might be next for it and its opponent. And it can choose the most optimal moves from among the possibilities.

But it doesn’t have any concept of the past, nor any memory of what has happened before. Apart from a rarely used chess-specific rule against repeating the same move three times, Deep Blue ignores everything before the present moment. All it does is look at the pieces on the chess board as it stands right now, and choose from possible next moves.

This type of intelligence involves the computer perceiving the world directly and acting on what it sees. It doesn’t rely on an internal concept of the world. In a seminal paper, AI researcher Rodney Brooks argued that we should only build machines like this. His main reason was that people are not very good at programming accurate simulated worlds for computers to use, what is called in AI scholarship a “representation” of the world.

The current intelligent machines we marvel at either have no such concept of the world, or have a very limited and specialized one for its particular duties. The innovation in Deep Blue’s design was not to broaden the range of possible movies the computer considered. Rather, the developers found a way to narrow its view, to stop pursuing some potential future moves, based on how it rated their outcome. Without this ability, Deep Blue would have needed to be an even more powerful computer to actually beat Kasparov.

Similarly, Google’s AlphaGo, which has beaten top human Go experts, can’t evaluate all potential future moves either. Its analysis method is more sophisticated than Deep Blue’s, using a neural network to evaluate game developments.

These methods do improve the ability of AI systems to play specific games better, but they can’t be easily changed or applied to other situations. These computerized imaginations have no concept of the wider world – meaning they can’t function beyond the specific tasks they’re assigned and are easily fooled.

They can’t interactively participate in the world, the way we imagine AI systems one day might. Instead, these machines will behave exactly the same way every time they encounter the same situation. This can be very good for ensuring an AI system is trustworthy: You want your autonomous car to be a reliable driver. But it’s bad if we want machines to truly engage with, and respond to, the world. These simplest AI systems won’t ever be bored, or interested, or sad.

Type II AI: Limited memory

This Type II class contains machines can look into the past. Self-driving cars do some of this already. For example, they observe other cars’ speed and direction. That can’t be done in a just one moment, but rather requires identifying specific objects and monitoring them over time.

These observations are added to the self-driving cars’ preprogrammed representations of the world, which also include lane markings, traffic lights and other important elements, like curves in the road. They’re included when the car decides when to change lanes, to avoid cutting off another driver or being hit by a nearby car.

But these simple pieces of information about the past are only transient. They aren’t saved as part of the car’s library of experience it can learn from, the way human drivers compile experience over years behind the wheel.

So how can we build AI systems that build full representations, remember their experiences and learn how to handle new situations? Brooks was right in that it is very difficult to do this. My own research into methods inspired by Darwinian evolution can start to make up for human shortcomings by letting the machines build their own representations.

Type III AI: Theory of mind

We might stop here, and call this point the important divide between the machines we have and the machines we will build in the future. However, it is better to be more specific to discuss the types of representations machines need to form, and what they need to be about.

Machines in the next, more advanced, class not only form representations about the world, but also about other agents or entities in the world. In psychology, this is called “theory of mind” – the understanding that people, creatures and objects in the world can have thoughts and emotions that affect their own behavior.

This is crucial to how we humans formed societies because they allowed us to have social interactions. Without understanding each other’s motives and intentions, and without taking into account what somebody else knows either about me or the environment, working together is at best difficult, at worst impossible.

If AI systems are indeed ever to walk among us, they’ll have to be able to understand that each of us has thoughts and feelings and expectations for how we’ll be treated. And they’ll have to adjust their behavior accordingly.

Type IV AI: Self-awareness

The final step of AI development is to build systems that can form representations about themselves. Ultimately, we AI researchers will have to not only understand consciousness, but build machines that have it.

This is, in a sense, an extension of the “theory of mind” possessed by Type III artificial intelligences. Consciousness is also called “self-awareness” for a reason. (“I want that item” is a very different statement from “I know I want that item.”) Conscious beings are aware of themselves, know about their internal states, and are able to predict feelings of others. We assume someone honking behind us in traffic is angry or impatient, because that’s how we feel when we honk at others. Without a theory of mind, we could not make those sorts of inferences.

While we are probably far from creating machines that are self-aware, we should focus our efforts toward understanding memory, learning and the ability to base decisions on past experiences. This is an important step to understand human intelligence on its own. And it is crucial if we want to design or evolve machines that are more than exceptional at classifying what they see in front of them.

This article was originally published on The Conversation. Read the original article.

If you enjoyed this article, you may also want to read these articles about AI: 

• President Obama discusses artificial intelligence with Media Lab Director Joi Ito
• What artificial intelligence will look like in 2030
• Artificial intelligence expedites breast cancer risk prediction
• Robohub roundtable: Public perceptions of robotics and AI, what’s your pet peeve?
• Why watching Westworld’s robots should make us question ourselves

See all the latest robotics news on Robohub, or sign up for our weekly newsletter.

Credit: Carl Vondrick, MIT CSAIL

Living in a dynamic physical world, it’s easy to forget how effortlessly we understand our surroundings. With minimal thought, we can figure out how scenes change and objects interact.

But what’s second nature for us is still a huge problem for machines. With the limitless number of ways that objects can move, teaching computers to predict future actions can be difficult.

Recently, researchers from MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) have gotten a step closer, developing a deep-learning algorithm that, given still images from a scene, can create brief videos that simulate the future of that scene.

Trained on two million unlabeled videos that include a year’s worth of footage, the algorithm generated videos that human subjects deemed to be realistic 20 percent more often than a baseline model.

To be clear, at this point the videos are still relatively low-resolution and only 1-1.5 seconds in length. But the team is hopeful that future versions could be used for everything from improved security tactics to safer self-driving cars.

According to CSAIL PhD student and first author Carl Vondrick, the algorithm can also help machines recognize people’s activities without expensive human annotations.

“These videos show us what computers think can happen in a scene,” says Vondrick. “If you can predict the future, you must have understood something about the present.”

Vondrick wrote the paper with MIT professor Antonio Torralba and Hamed Pirsiavash, a former CSAIL postdoctoral associate who is now a professor at the University of Maryland, Baltimore County. The work will be presented at next week’s Neural Information Processing Systems (NIPS) conference in Barcelona.

How it works
Multiple researchers have tackled similar topics in computer vision, including MIT professor Bill Freeman, whose new work on “visual dynamics” also creates future frames in a scene. But where his model focuses on extrapolating videos into the future, Torralba’s model can also generate completely new videos that haven’t been seen before.

Credit: Carl Vondrick, MIT CSAIL

Previous systems build up scenes frame by frame, which creates a large margin for error. In contrast, this work focuses on processing the entire scene at once, with the algorithm generating as many as 32 frames from scratch per second.

“Building up a scene frame-by-frame is like a big game of ‘Telephone,’ which means that the message falls apart by the time you go around the whole room,” says Vondrick. “By instead trying to predict all frames simultaneously, it’s as if I’m talking to everyone in the room at once.”

Of course, there’s a trade-off to generating all frames simultaneously: while it becomes more accurate, the computer model also becomes more complex for longer videos.

To create multiple frames, researchers taught the model to generate the foreground separate from the background, and to then place the objects in the scene to let the model learn which objects move and which objects don’t.

The team used a deep-learning method called “adversarial learning” that involves training two competing neural networks. One network generates video, and the other discriminates between the real and generated videos. Over time, the generator learns to fool the discriminator.

From that, the model can create videos resembling scenes from beaches, train stations, hospitals, and golf courses.  For example, the beach model produced beaches with crashing waves, and the golf model had people walking on grass.

Testing the scene
The team compared the videos against a baseline of generated videos and asked subjects which they thought were more realistic. From over 13,000 opinions of 150 users, subjects chose the generative model videos 20 percent more often than the baseline.

To be clear, the the model still lacks some fairly simple common-sense principles. For example, it often doesn’t understand that objects are still there when they move, like when a train passes through a scene. The model also tends to make humans and objects look much larger in size than reality.

As mentioned before, another limitation is that the generated videos are just one and a half seconds long, which the team hopes to be able to increase in future work. The challenge is that this requires tracking longer dependencies to ensure that the scene still makes sense over longer time periods. One way to do this would be to add human supervision.

“It’s difficult to aggregate accurate information across long time periods in videos,” says Vondrick. “If the video has both cooking and eating activities, you have to be able to link those two together to make sense of the scene.”

These types of models aren’t limited to predicting the future. Generative videos can be used for adding animation to still images, like the animated newspaper from the Harry Potter books. They could also help detect anomalies in security footage and compress data for storing and sending longer videos.

“In the future, this will let us scale up vision systems to recognize objects and scenes without any supervision, simply by training them on video,” says Vondrick.

This work was supported by the National Science Foundation, the START program at UMBC, and a Google PhD fellowship.

Read the research paper.

Information extraction — or automatically classifying data items stored as plain text — is a major topic of artificial-intelligence research. Image: MIT News

Of the vast wealth of information unlocked by the Internet, most is plain text. The data necessary to answer myriad questions — about, say, the correlations between the industrial use of certain chemicals and incidents of disease, or between patterns of news coverage and voter-poll results — may all be online. But extracting it from plain text and organizing it for quantitative analysis may be prohibitively time-consuming.

Information extraction — or automatically classifying data items stored as plain text — is thus a major topic of artificial-intelligence research. Last week, at the Association for Computational Linguistics’ Conference on Empirical Methods on Natural Language Processing, researchers from MIT’s Computer Science and Artificial Intelligence Laboratory won a best-paper award for a new approach to information extraction that turns conventional machine learning on its head.

Most machine-learning systems work by combing through training examples and looking for patterns that correspond to classifications provided by human annotators. For instance, humans might label parts of speech in a set of texts, and the machine-learning system will try to identify patterns that resolve ambiguities — for instance, when “her” is a direct object and when it’s an adjective.

Typically, computer scientists will try to feed their machine-learning systems as much training data as possible. That generally increases the chances that a system will be able to handle difficult problems.

In their new paper, by contrast, the MIT researchers train their system on scanty data — because in the scenario they’re investigating, that’s usually all that’s available. But then they find the limited information an easy problem to solve.

“In information extraction, traditionally, in natural-language processing, you are given an article and you need to do whatever it takes to extract correctly from this article,” says Regina Barzilay, the Delta Electronics Professor of Electrical Engineering and Computer Science and senior author on the new paper. “That’s very different from what you or I would do. When you’re reading an article that you can’t understand, you’re going to go on the web and find one that you can understand.”

Confidence boost

Essentially, the researchers’ new system does the same thing. A machine-learning system will generally assign each of its classifications a confidence score, which is a measure of the statistical likelihood that the classification is correct, given the patterns discerned in the training data. With the researchers’ new system, if the confidence score is too low, the system automatically generates a web search query designed to pull up texts likely to contain the data it’s trying to extract.

It then attempts to extract the relevant data from one of the new texts and reconciles the results with those of its initial extraction. If the confidence score remains too low, it moves on to the next text pulled up by the search string, and so on.

“The base extractor isn’t changing,” says Adam Yala, a graduate student in the MIT Department of Electrical Engineering and Computer Science (EECS) and one of the coauthors on the new paper. “You’re going to find articles that are easier for that extractor to understand. So you have something that’s a very weak extractor, and you just find data that fits it automatically from the web.” Joining Yala and Barzilay on the paper is first author Karthik Narasimhan, also a graduate student in EECS.

Remarkably, every decision the system makes is the result of machine learning. The system learns how to generate search queries, gauge the likelihood that a new text is relevant to its extraction task, and determine the best strategy for fusing the results of multiple attempts at extraction.

Just the facts

In experiments, the researchers applied their system to two extraction tasks. One was the collection of data on mass shootings in the U.S., which is an essential resource for any epidemiological study of the effects of gun-control measures. The other was the collection of similar data on instances of food contamination. The system was trained separately for each task.

In the first case — the database of mass shootings — the system was asked to extract the name of the shooter, the location of the shooting, the number of people wounded, and the number of people killed. In the food-contamination case, it extracted food type, type of contaminant, and location. In each case, the system was trained on about 300 documents.

From those documents, it learned clusters of search terms that tended to be associated with the data items it was trying to extract. For instance, the names of mass shooters were correlated with terms like “police,” “identified,” “arrested,” and “charged.” During training, for each article the system was asked to analyze, it pulled up, on average, another nine or 10 news articles from the web.

The researchers compared their system’s performance to that of several extractors trained using more conventional machine-learning techniques. For every data item extracted in both tasks, the new system outperformed its predecessors, usually by about 10 percent.

“One of the difficulties of natural language is that you can express the same information in many, many different ways, and capturing all that variation is one of the challenges of building a comprehensive model,” says Chris Callison-Burch, an assistant professor of computer and information science at the University of Pennsylvania. “[Barzilay and her colleagues] have this super-clever part of the model that goes out and queries for more information that might result in something that’s simpler for it to process. It’s clever and well-executed.”

Callison-Burch’s group is using a combination of natural-language processing and human review to build a database of information on gun violence, much like the one that the MIT researchers’ system was trained to produce. “We’ve crawled millions and millions of news articles, and then we pick out ones that the text classifier thinks are related to gun violence, and then we have humans start doing information extraction manually,” he says. “Having a model like Regina’s that would allow us to predict whether or not this article corresponded to one that we’ve already annotated would be a huge time savings. It’s something that I’d be very excited to do in the future.”

Read the paper here.

Frame of the tutoring experience of the student with low motivation and high competence. Reproduced with permission from the parents of the children. Courtesy of Dr Imbernon Cuadrado

By: Ian Salter

The use of robotic tutors in primary school classrooms is one step closer according to research recently published in the open access journal Frontiers in Computational Neuroscience.

Dr Imbernòn Cuadrado and his co-workers at the Department of Artificial Intelligence in Madrid have developed an integrated computational architecture (ARTIE) for use with software applications in schools.

“The main goal of our work was to design a system that can detect the emotional state of primary school children interacting with educational software and make pedagogic interventions with a robot tutor that can ultimately improve the learning experience,” says Luis Imbernòn Cuadrado.

Online educational resources are becoming increasingly common in the classroom, although they have not taken into sufficient account that the learning ability of primary school children is particularly sensitive to their emotional state. This is perhaps where robot tutors can step in to assist teachers.

Rather than focusing on specific emotions, the researchers first identified three cognitive states (concentrating, distracted and inactive) known to influence the course of learning.  Keyboard strokes and mouse actions of children using educational software were used to predict which of these cognitive states the child is experiencing and subsequently linked to an algorithm that chooses the correct form of pedagogic intervention. These could be words and gestures of encouragement or attempts to raise interest and motivation for a specific learning objective, all of which can be delivered by a robot tutor.

In the Frontiers research article, Dr. Imbernón Cuadrado and co-workers describe an example of a robotic tutor platform, MONICA, which integrates the educational software Scratch to a commercially available robot through the computational architecture developed in their paper.

Reproduced with permission from the parents of the children. Courtesy of Dr Imbernon Cuadrado

MONICA was tested out on two primary school volunteers known to have different cognitive learning states. Whilst the children enjoyed having a robot guide their learning, and preferred it to working alone, they both felt they would have learned more with their normal teacher, despite feeling more relaxed in the presence of a robot tutor.

“Our first prototype was designed to demonstrate that the architecture works in detecting simplified emotional states,” says Dr. Imbernón Cuadrado. “The next step will be to implement methods for detecting a more complex range of emotions with cameras and microphones and to test the longer term impact of robot tutors on children’s learning curves.”

It seems however that robot tutors are not ready to replace school teachers just yet.  “We consider that robot tutors could have an effective support role to play in the primary school classroom in helping children reach their specific learning objectives,” concludes Dr. Imbernón Cuadrado.

Click here to read the research paper

If you liked this article, you may also want to read these other articles by Frontiers:

• Living with a prosthesis that learns: A case-study in translational medicine
• Putting humanoid robots in contact with their environment
• Robotic research: Are we applying the scientific method?
• Robots can successfully imitate human motions in the operating room

See all the latest robotics news on Robohub, or sign up for our weekly newsletter.

Source: YouTube

In October, the U.S.-China Economic and Security Review Commission released a report, China’s Industrial and Military Robotics Development, prepared by the Defense Group, Inc. at the Commission’s request. The report examines the development of China’s unmanned industrial, service, and military robotics systems, such as drones and driverless cars, and the economic and national security implications of these trends for the United States.

China’s rising demand for high-end robotic components and service robots as well as U.S.-China bilateral artificial intelligence research present potential markets and collaboration opportunities for the United States. But the inherently dual-use functions of industrial and service robotics strengthen China’s commercial and military production and robotic capabilities, potentially improving China’s defense industry and eroding U.S. military advantages. In addition, the application of artificial intelligence can enhance the learning capabilities and the military effectiveness of unmanned systems such as drones, which may undermine U.S. technological and military advantages.

Overall, the report advocates that the U.S. government:

1. implement the recommendations of the Advanced Manufacturing Partnership 2.0 Steering Committee Report and expand the National Network for Manufacturing Innovation;
2. monitor and account for Chinese advances in unmanned systems and electronic countermeasures;
3. conduct an interagency review with economic, scientific, and regional experts to assess U.S.-China cooperation and bilateral investments in artificial intelligence;
4. increase awareness among federal agencies, defense contractors, and research universities that Chinese research institutes actively collect their published materials, designs, specifications, and graphics;
5. fully implement the Cybersecurity National Action Plan;
6. use China’s state plans, procurement practices, defense plans, and other Chinese language materials to identify technologies that the PRC is seeking to acquire;
7. consider requirements to more thoroughly vet foreign participants for military or other undisclosed defense affiliations in academic exchanges and research in emerging technologies; and
8. monitor and, when necessary, investigate China’s growing foreign investments in robotics and artificial intelligence companies.

The report was authored by Jonathan Ray, Katie Atha, Edward Francis, Caleb Dependahl, Dr. James Mulvenon, Daniel Alderman, and Leigh Ann Ragland-Luce.

The report can be viewed here.

Researchers from MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) have devised a way to train neural networks so that they provide not only predictions and classifications but rationales for their decisions. Illustration: Christine Daniloff/MIT

In recent years, the best-performing systems in artificial-intelligence research have come courtesy of neural networks, which look for patterns in training data that yield useful predictions or classifications. A neural net might, for instance, be trained to recognize certain objects in digital images or to infer the topics of texts.

But neural nets are black boxes. After training, a network may be very good at classifying data, but even its creators will have no idea why. With visual data, it’s sometimes possible to automate experiments that determine which visual features a neural net is responding to. But text-processing systems tend to be more opaque.

At the Association for Computational Linguistics’ Conference on Empirical Methods in Natural Language Processing, researchers from MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) will present a new way to train neural networks so that they provide not only predictions and classifications but rationales for their decisions.

“In real-world applications, sometimes people really want to know why the model makes the predictions it does,” says Tao Lei, an MIT graduate student in electrical engineering and computer science and first author on the new paper. “One major reason that doctors don’t trust machine-learning methods is that there’s no evidence.”

“It’s not only the medical domain,” adds Regina Barzilay, the Delta Electronics Professor of Electrical Engineering and Computer Science and Lei’s thesis advisor. “It’s in any domain where the cost of making the wrong prediction is very high. You need to justify why you did it.”

“There’s a broader aspect to this work, as well,” says Tommi Jaakkola, an MIT professor of electrical engineering and computer science and the third coauthor on the paper. “You may not want to just verify that the model is making the prediction in the right way; you might also want to exert some influence in terms of the types of predictions that it should make. How does a layperson communicate with a complex model that’s trained with algorithms that they know nothing about? They might be able to tell you about the rationale for a particular prediction. In that sense, it opens up a different way of communicating with the model.”

Virtual brains

Neural networks are so called because they mimic — approximately — the structure of the brain. They are composed of a large number of processing nodes that, like individual neurons, are capable of only very simple computations but are connected to each other in dense networks.

In a process referred to as “deep learning,” training data is fed to a network’s input nodes, which modify it and feed it to other nodes, which modify it and feed it to still other nodes, and so on. The values stored in the network’s output nodes are then correlated with the classification category that the network is trying to learn — such as the objects in an image, or the topic of an essay.

Over the course of the network’s training, the operations performed by the individual nodes are continuously modified to yield consistently good results across the whole set of training examples. By the end of the process, the computer scientists who programmed the network often have no idea what the nodes’ settings are. Even if they do, it can be very hard to translate that low-level information back into an intelligible description of the system’s decision-making process.

In the new paper, Lei, Barzilay, and Jaakkola specifically address neural nets trained on textual data. To enable interpretation of a neural net’s decisions, the CSAIL researchers divide the net into two modules. The first module extracts segments of text from the training data, and the segments are scored according to their length and their coherence: The shorter the segment, and the more of it that is drawn from strings of consecutive words, the higher its score.

The segments selected by the first module are then passed to the second module, which performs the prediction or classification task. The modules are trained together, and the goal of training is to maximize both the score of the extracted segments and the accuracy of prediction or classification.

One of the data sets on which the researchers tested their system is a group of reviews from a website where users evaluate different beers. The data set includes the raw text of the reviews and the corresponding ratings, using a five-star system, on each of three attributes: aroma, palate, and appearance.

What makes the data attractive to natural-language-processing researchers is that it’s also been annotated by hand, to indicate which sentences in the reviews correspond to which scores. For example, a review might consist of eight or nine sentences, and the annotator might have highlighted those that refer to the beer’s “tan-colored head about half an inch thick,” “signature Guinness smells,” and “lack of carbonation.” Each sentence is correlated with a different attribute rating.

Validation

As such, the data set provides an excellent test of the CSAIL researchers’ system. If the first module has extracted those three phrases, and the second module has correlated them with the correct ratings, then the system has identified the same basis for judgment that the human annotator did.

In experiments, the system’s agreement with the human annotations was 96 percent and 95 percent, respectively, for ratings of appearance and aroma, and 80 percent for the more nebulous concept of palate.

In the paper, the researchers also report testing their system on a database of free-form technical questions and answers, where the task is to determine whether a given question has been answered previously.

In unpublished work, they’ve applied it to thousands of pathology reports on breast biopsies, where it has learned to extract text explaining the bases for the pathologists’ diagnoses. They’re even using it to analyze mammograms, where the first module extracts sections of images rather than segments of text.

“There’s a lot of hype now — and rightly so — around deep learning, and specifically deep learning for natural-language processing,” says Byron Wallace, an assistant professor of computer and information science at Northeastern University. “But a big drawback for these models is that they’re often black boxes. Having a model that not only makes very accurate predictions but can also tell you why it’s making those predictions is a really important aim.”

“As it happens, we have a paper that’s similar in spirit being presented at the same conference,” Wallace adds. “I didn’t know at the time that Regina was working on this, and I actually think hers is better. In our approach, during the training process, while someone is telling us, for example, that a movie review is very positive, we assume that they’ll mark a sentence that gives you the rationale. In this way we train the deep-learning model to extract these rationales. But they don’t make this assumption, so their model works without using direct annotations with rationales, which is a very nice property.”

For a sci-fi fan like me, fascinated by the nature of human intelligence and the possibility of building life-like robots, it’s always interesting to find a new angle on these questions. As a re-imagining of the original 1970s science fiction film set in a cowboy-themed, hyper-real adult theme park populated by robots that look and act like people, Westworld does not disappoint.

Westworld challenges us to consider the difference between being human and being a robot. From the beginning of this new serialisation on HBO we are confronted with scenes of graphic human-on-robot violence. But the robots in Westworld have more than just human-like physical bodies, they display emotion including extreme pain, they see and recognise each other’s suffering, they bleed and even die. What makes this acceptable, at least within Westworld’s narrative, is that they are just extremely life-like human simulations; while their behaviour is realistically automated, there is “nobody home”.

But from the start, this notion that a machine of such complexity is still merely a machine is undermined by constant reminders that they are also so much like us. The disturbing message, echoing that of previous sci-fi classics such as Blade Runner and AI, is that machines could one day be so close to human as to be indistinguishable – not just in intellect and appearance, but also in moral terms.

At the same time, by presenting an alternate view of the human condition through the technological mirror of life-like robots, Westworld causes us to reflect that we are perhaps also just sophisticated machines, albeit of a biological kind – an idea that has been forcefully argued by the philosopher Daniel Dennett.

The unfortunate robots in Westworld have, at least initially, no insight into their existential plight. They enter into each new day programmed with enthusiasm and hope, oblivious to its pre-scripted violence and tragedy. We may pity these automatons their fate – but how closely does this blinkered ignorance, belief in convenient fictions, and misguided presumption of agency resemble our own human predicament?

Westworld arrives at a time when people are already worried about the real-world impact of advances in robotics and artificial intelligence. Physicist Stephen Hawking and technologist Elon Musk are among the powerful and respected voices to have expressed concern about allowing the AI genie to escape the bottle. Westworld’s contribution to the expanding canon of science fiction dystopias will do nothing to quell such fears. Channelling Shakespeare’s King Lear, a malfunctioning robot warns us in chilling terms: “I shall have such revenges on you both. The things I will do, what they are, yet I know not. But they will be the terrors of the Earth.”

But against these voices are other distinguished experts trying to quell the panic. For Noam Chomsky, the intellectual godfather of modern AI, all talk of matching human intelligence in the foreseeable future remains fiction, not science. One of the world’s best-known roboticists, Rodney Brooks has called on us to relax: AI is just a tool, not a threat.

As a neuroscientist and roboticist, I agree that we are far from being able to replicate human intelligence in robot form. Our current systems are too simple, probably by several orders of magnitude. Building human-level AI is extremely hard; as Brooks says, we are just at the beginning of a very long road. But I see the path along which we are developing AI as one of symbiosis, in which we can use robots to benefit society and exploit advances in artificial intelligence to boost our own biological intelligence.

More than just a tool

Nevertheless, in recent years the robots and AI are “just tools” line of argument has begun to frustrate me. Partly because it has failed to calm the disquiet around AI, and partly because there are good reasons why these technologies are different from others in the past.

Even if robots are just tools, people will see them as more than that. It seems natural for people to respond to robots – even some of the more simple, non-human robots we have today – as though they have goals and intentions. It may be an innate tendency of our profoundly social human minds to see entities that act intelligently in this way. More importantly, people may see robots as having psychological properties such as the ability to experience suffering.

Being human is about more than just looking the part.

It may be difficult to persuade them to see otherwise, particularly if we continue to make robots more life-like. If so, we may have to adapt our ethical frameworks to take this into account. For instance, we might consider violence towards a robot as wrong, even though the suffering is imagined rather than real. Indeed, faced with violence towards a robot some people show this sort of ethical response spontaneously. We will have to deal with these issues as we learn to live with robots.

As AI and robot technology becomes more complex, robots may come to have interesting psychological properties that make them more than just tools. The fictional robots of Westworld are clearly in this category, but already real robots are being developed that have artificial drives and motivations, that are aware of their own bodies as distinct from the rest of the world, that are equipped with internal models of themselves and others as social entities, and that are able to think about their own past and future.

These are not properties that we find in drills and screwdrivers. They are embryonic psychological capacities that, so far, have only been found in living, sentient entities such as humans and animals. Stories such as Westworld remind us that as we progress toward ever more sophisticated AI, we should consider that this path might lead us both to machines that are more like us, and to seeing ourselves as more like machines.

This article was originally posted on The Conversation

If you enjoyed this article, you might also be interested in:

• The new Westworld: Humanizing the un-human, or dehumanizing humankind?
• Ex Machina: When Turing meets Bechdel test
• President Obama discusses artificial intelligence with Media Lab Director Joi Ito

See all the latest robotics news on Robohub, or sign up for our weekly newsletter.

Photo: Gretchen Ertl

Sarah Hensley is preparing an astronaut named Valkyrie for a mission to Mars. It is 6 feet tall, weighs 300 pounds, and is equipped with an extended chest cavity that makes it look distinctly female. Hensley spends much of her time this semester analyzing the movements of one of Valkyrie’s arms.

As a fourth-year electrical engineering student at MIT, Hensley is working with a team of researchers to prepare Valkyrie, a humanoid robot also known as R5, for future space missions. As a teenager in New Jersey, Hensley loved to read in her downtime, particularly Isaac Asimov’s classic robot series. “I’m a huge science fiction nerd — and now I’m actually getting to work with a robot that’s real and not just in books. That’s like, wow.”

Hensley is studying Valkyrie for an advanced independent research program, or SuperUROP, as one of only three undergraduate students in the Robot Locomotion Group in MIT’s Computer Science and Artificial Intelligence Laboratory. Most of her colleagues are graduate-level researchers and postdocs with extensive experience working on complex humanoids. The group is led by professor of electrical engineering and computer science Russ Tedrake, who successfully programmed Valkyrie’s predecessor (named Atlas) to open doors, turn valves, drill holes, climb stairs, and drive a car for the DARPA Robotics Challenge in 2015.

Valkyrie has 28 torque-controlled joints, four body cameras, and more than 200 individual sensors, Hensley says. The robot can walk, bend its joints, and turn a door handle. “This is one of the most advanced robots in the world. And it’s 20 feet from my desk,” she adds.

That’s largely because Valkyrie has a long way to go before it leaves for Mars. MIT is one of three institutions, including Northeastern University and the University of Edinburgh, that NASA selected to develop software enabling the robot to perform space-related tasks — open airlock hatches, attach and remove power cables, repair equipment, and retrieve samples. Oh yeah, and get to its feet when it falls down.

Hensley, who started in the lab over the summer, is intrigued by the challenge of harmonizing the movements of such a highly complex system. “I am trying to solve a very tricky problem,” she says. She’s working out how best to control Valkyrie’s elbow movements by comparing two potential approaches. One uses a main controller to gather information from the various motor systems within the arm, and then uses that data to make accurate movement decisions. The other approach is decentralized, and leaves it to each motor system to decide and act on its own.

Hensley gets animated discussing the alternatives. “Is it better to have multiple decision makers with access to different information? Or is better to have one decision maker choosing all of the motor inputs?” she asks. Hensley has already been accepted into a master’s degree program in electrical engineering at MIT. She hopes to be able to continue her work on Valkyrie.

Every day, Hensley leaves Tau Epsilon Phi, her co-ed fraternity house in the Back Bay; walks across the Massachusetts Ave. bridge to the Stata Center; and plants herself in front of two large monitors in the robotics lab. She analyzes a wealth of scientific literature and writes code for computer simulations of the equations that move the robot’s arm. Sometimes she gets up for peppermint tea, or to peer around the corner of her cubicle at Valkyrie.

One thing is certain, says Hensley. Pop culture fears that machines may soon prove superior to humans are laughable. When Valkyrie is turned on and moves, Hensley says, it often “kind of shivers and falls down. One thing you realize working in this lab is that we are really far away from the robot apocalypse,” she quips. “Sometimes robots work, and sometimes they don’t. That’s our challenge.”

MIT Media Lab Director Joi Ito (left), WIRED Editor-in-Chief Scott Dadich (center), and U.S. President Barack Obama confer in the Roosevelt Room of the White House. Photo: WIRED

When President Barack Obama agreed to guest-edit the November issue of WIRED, he selected MIT Media Lab Director Joi Ito for an exchange of ideas about artificial intelligence (AI). Their recent interview at the White House is featured in the latest online issue of WIRED, published on Oct. 12.

The one-on-one conversation, moderated by WIRED Editor-in-Chief Scott Dadich, ran the gamut of topics at the intersection of societal needs, ethics, and technology — from cybersecurity to self-driving cars; from the roles of government, industry, and academia to the lack of diversity in tech; from “moonshot” motivations to innovation at the margins; and from neurodiversity to Star Trek. All this was covered in the context of AI and extended intelligence (EI), which uses machine learning to augment human capabilities.

Ito says his overarching message for Obama in their conversation was that AI — and the space it occupies — is no longer just a computer science issue. “It’s also very much a societal thing.” And we shouldn’t underestimate the difficulties, he adds. “We can’t think that machines will just figure it all out for us. Everyone needs to recognize the importance of understanding how AI behaves, and we have to address the critical need to build societal values into AI.” Ito is encouraged by what he characterizes as the president’s “amazing team,” which includes U.S. Chief Technology Officer Megan Smith ’86, SM ’88, along with Deputy U.S. CTOs Alexander Macgillivray, formerly a lawyer for Twitter, and Ed Felten of Princeton University.

“And the role of the Media Lab is to be a connective tissue between computer science, and the social sciences, and the lawyers, and the philosophers,” says Ito. “What’s cool is that President Obama gets that.”

“Where should the center of research live, if there even is a center?” Dadich asked both Obama and Ito. In the article, the president noted that “part of the problem that we’ve seen is that our general commitment as a society to basic research has diminished. Our confidence in collective action has been chipped away, partly because of ideology and rhetoric.” Ito said, “Historically, it probably would have been a group of academics with help from a government. But right now, most of the billion-dollar labs are in the business.” He later explained that the Media Lab works in the space between the disciplines — the antidisciplinary space between humans and computers, and between networks and society: “MIT has been and continues to be a center for AI research, and now we see the need for a more fully integrated approach, cutting across all disciplines. I hope the Media Lab can make a meaningful contribution to that with our method and experience.”

What’s next?

Above all, says Ito, “What’s important is to find the people who want to use AI for good — communities and leaders — and figure out how to help them use it.” To that end, Ito says, the MIT Media Lab is committed to not only exploring the technology of AI and EI but also addressing their ramifications for humankind.

“We’re already talking about these issues with people across the full spectrum of society,” Ito adds. “This is a crucial area that cannot be ignored.”

While the future of artificial intelligence is probably going to be driven by Silicon Valley, the folks in Washington DC want their say about how it will work, too.

In two reports, the White House outlined its strategy for promoting artificial intelligence research and development in the US. While most of the bigger questions were punted to future legislators (“more research is needed” is a key phrase), the executive branch did draw some lines in the sand. And most importantly to the research community, the White House is not pushing for AI to be broadly regulated—instead, the use of the technology will be held to specific standards in the automotive, aviation, and finance industries.

Three key guiding philosophies were presented across the reports: AI needs to augment humanity instead of replacing it, AI needs to be ethical, and there must be an equal opportunity for everyone to develop these systems.

These are subjects on US president Barack Obama’s mind, too—as he says in a well-timed feature in Wired magazine. “Most people aren’t spending a lot of time right now worrying about singularity—they are worrying about ‘Well, is my job going to be replaced by a machine?’” Obama said in the interview.

Human-machine collaboration in its many forms is a major theme in the reports, titled “Preparing for the Future of Artificial Intelligence” and “National Artificial Intelligence Research and Development Strategic Plan.”

“The walls between humans and AI systems are slowly beginning to erode, with AI systems augmenting and enhancing human capabilities,” the Strategic Plan report says.

Credit: CB Insights

The White House imagines virtual personal assistants housed in smart glasses, automated factories that assist humans in complex building tasks, and systems that provide better data for farmers, all in the context that these could be job creators and not job stealers. When it comes to manufacturing, the White House believes that more efficient automation could incentivize companies to bring their production back to US soil. While these are appealing futures, the reports mainly rely on macroeconomic forces to make them come to fruition, rather than any set initiatives.

The reports do briefly acknowledge the possibility that AI goes the complete opposite direction, making low- and medium-skill jobs redundant. The White House plans to release another report later this year to delve into that prospect more fully.

National security is also on the list, with the reports suggesting AI has a role in cyber security, and can be used to detect and annul cyberattacks as they target US citizens or infrastructure. The White House also recommends algorithmic surveillance of individuals and crowds, while acknowledging more study is needed on the matter, especially given current attempts to implement “predictive policing” have been shown to be racially biased.

That brings ethics into the discussion. “The government needs to be involved a little bit more,” Obama told Wired. “Not always to force the new technology into the square peg that exists but to make sure the regulations reflect a broad base set of values. Otherwise, we may find that it’s disadvantaging certain people or certain groups.”

The government seemingly wants to tackle ethics with more transparency. “Researchers must learn how to design these systems so that their actions and decision-making are transparent and easily interpretable by humans, and thus can be examined for any bias they may contain, rather than just learning and repeating these biases,” the Strategic Plan says.

Right now, there’s no standard for transparency in AI research. Companies like Google and Facebook release open-source code and sometimes reveal the data they use, but no regulation exists. Recently, Facebook, Google, Amazon, IBM, and Microsoft joined to make the Partnership on AI, with the goal of setting best practices for transparency, although no specifics have been released.

One way to combat biased algorithms is to provide unbiased data, something the White House reports urge. Not only would carefully-curated open data sets allow everyone to see how algorithms are taught about the world, but it also means that anyone can use them to develop their own AI.

“The lack of vetted and openly available datasets with identified provenance to enable reproducibility is a critical factor to confident advancement in AI,” the report reads.

Currently, there are standard open data sets in computer science, like the MNIST database of hand-written numbers or ImageNet, but they’re not regulated and can contain their own bias. As an example, ImageNet contains an overwhelming amount of images of dogs and cats. When Google researchers used DeepDreamlast year to find out what ImageNet-trained algorithms naturally gravitated towards, the algorithms generated dog faces, showing a bias towards finding those patterns in images. Google recently announced their open-source answer to ImageNet, the Open Images dataset, which covers six times as many categories, and is expected to be an improvement.

The government makes clear in the reports that it had led the charge on tech development for decades. It claims the internet and deep learning (the basis of most modern AI) as examples of government funding that paid off over the long-term; federal government investment in both technologies in the 1960s. In 2015, the federal government invested more than $1 billion in AI research.

But at the same time, the White House acknowledges that to accomplish what it lays out in the two reports, it must work with researchers and innovators in the field. The reports suggest most of the funding will come in the form of traditional grants and Grand Challenges, where teams build experimental tech to compete for a prize.

“The gap between the talent in the federal government and the private sector is actually not wide at all,” Obama told Wired. “The technology gap, though, is massive.”

Article originally published on WEF.

If you liked this article, you may also want to read:

• White House launches public workshops on AI issues
• White House economic report looks to robotics for the future
• We the Geeks: Whitehouse hangouts celebrate robots, maker culture and more

See all the latest robotics news on Robohub, or sign up for our weekly newsletter.

From L-R: PhD Fadel Adib, PhD Mingmin Zhao and Professor Dina Katabi demonstrating different ’emotions’ like the picture. Credit: Jason Dorfman, MIT CSAIL

By measuring your heartbeat and breath, this device from MIT’s Computer Science and Artificial Intelligence Lab can tell if you’re excited, happy, angry or sad
.

As many a relationship book can tell you, understanding someone else’s emotions can be a difficult task. Facial expressions aren’t always reliable: a smile can conceal frustration, while a poker face might mask a winning hand.

But what if technology could tell us how someone is really feeling?

Researchers from MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) have developed “EQ-Radio,” a device that can detect a person’s emotions using wireless signals.

By measuring subtle changes in breathing and heart rhythms, EQ-Radio is 87 percent accurate at detecting if a person is excited, happy, angry or sad – and can do so without on-body sensors or facial-recognition software.

MIT professor and project lead Dina Katabi envisions the system being used in entertainment, consumer behavior and health-care. Film studios and ad agencies could test viewers’ reactions in real-time, while smart homes could use information about your mood to adjust the heating or suggest that you get some fresh air.

Professor Dina Katabi (middle) explains how PhD Fadel Adib’s face (right) is neutral, but that EQ-Radio’s analysis of his heartbeat and breathing show that he is sad. Credit: Jason Dorfman MIT CSAIL

“Our work shows that wireless signals can capture information about human behavior that is not always visible to the naked eye,” says Katabi, who co-wrote a paper on the topic with PhD students Mingmin Zhao and Fadel Adib. “We believe that our results could pave the way for future technologies that could help monitor and diagnose conditions like depression and anxiety.”

EQ-Radio builds on Katabi’s continued efforts to use wireless to measure human behavior like breathing and falling. She says that she will incorporate emotion-detection into her spin-off company Emerald, which makes a device that is aimed at detecting and predicting falls among the elderly.

Using wireless signals reflected off people’s bodies, the device measures heartbeats as accurately as an ECG monitor, with a margin of error of approximately 0.3 percent. It then studies the waveforms within each heartbeat to match a person’s behavior to how they previously acted in one of the four emotion-states.

The team will present the work next month at the Association of Computing Machinery’s International Conference on Mobile Computing and Networking (MobiCom).

How it works

Existing emotion-detection methods rely on audiovisual cues or on-body sensors, but there are downsides to both techniques. Cues like facial expressions are famously unreliable, while on-body sensors like chest bands and ECG monitors are inconvenient to wear and become inaccurate if they change position over time.

EQ-Radio instead sends wireless signals that reflect off of a person’s body and back to the device. Its beat-extraction algorithms break the reflections into individual heartbeats and analyze the small variations in heartbeat intervals to determine their levels of arousal and positive affect.

These measurements are what allow EQ-Radio to detect emotion. For example, a person whose signals correlate to low arousal and negative affect is more likely to tagged as sad, while someone whose signals correlate to high arousal and positive affect would likely be tagged as excited.

The exact correlations vary from person to person, but are consistent enough that EQ-Radio could detect emotions with 70 percent accuracy even when it hadn’t previously measured the target person’s heartbeat.

“Just by generally knowing what human heartbeats look like in different emotional states, we can look at a random person’s heartbeat and reliably detect their emotions,” says Zhao.

For the experiments, subjects used videos or music to recall a series of memories that each evoked one the four emotions, as well as a no-emotion baseline. Trained just on those five sets of two-minute videos, EQ-Radio could then accurately classify the person’s behavior among the four emotions 87 percent of the time.

Compared to Microsoft’s vision-based “Emotion API”, which focuses on facial expressions,  EQ-Radio was found to be significantly more accurate in detecting joy, sadness and anger. The two systems performed similarly with neutral emotions, since a face’s absence of emotion is generally easier to detect than its presence.

One of the CSAIL team’s toughest challenges was to tune out irrelevant data. In order to get the heart-rate, for example, the team had to dampen the breathing, since the distance that a person’s chest moves from breathing is much greater than the distance that their heart moves to beat.

To do so, the team focused on wireless signals that are based on acceleration rather than distance traveled, since the rise and fall of the chest with each breath tends to be much more consistent –  and therefore have a lower acceleration – than the motion of the heartbeat.

Although the focus on emotion-detection meant analyzing the time between heartbeats, the team says that the algorithm’s ability to captured the heartbeat’s entire waveform means that in the future it could be used for non-invasive health monitoring and diagnostic settings.

“By recovering measurements of the heart valves actually opening and closing at a millisecond time-scale, this system can literally detect if someone’s heart skips a beat,” says Adib. “This opens up the possibility of learning more about conditions like arrhythmia, and potentially exploring other medical applications that we haven’t even thought of yet.”

Read more about the research here.

Artificial intelligence (AI) has already transformed our lives — from the autonomous cars on the roads to the robotic vacuums and smart thermostats in our homes. Over the next 15 years, AI technologies will continue to make inroads in nearly every area of our lives, from education to entertainment, health care to security.

The question is, are we ready? Do we have the answers to the legal and ethical quandaries that will certainly arise from the increasing integration of AI into our daily lives? Are we even asking the right questions?

Now, a panel of academics and industry thinkers has looked ahead to 2030 to forecast how advances in AI might affect life in a typical North American city and spark discussion about how to ensure the safe, fair, and beneficial development of these rapidly developing technologies.

“Artificial Intelligence and Life in 2030” is the first product of the One Hundred Year Study on Artificial Intelligence (AI100), an ongoing project hosted by Stanford University to inform debate and provide guidance on the ethical development of smart software, sensors, and machines. Every five years for the next 100 years, the AI100 project will release a report that evaluates the status of AI technologies and their potential impact on the world.

Credit: CB Insights

Barbara Grosz, Higgins Professor of Natural Sciences at the Harvard John A. Paulson School of Engineering and Applied Sciences, chairs the AI100 Standing Committee.

“Now is the time to consider the design, ethical, and policy challenges that AI technologies raise,” said Grosz. “If we tackle these issues now and take them seriously, we will have systems that are better designed in the future and more appropriate policies to guide their use.”

“We believe specialized AI applications will become both increasingly common and more useful by 2030, improving our economy and quality of life,” said Peter Stone, a computer scientist at the University of Texas, Austin, and chair of the report. “But this technology will also create profound challenges, affecting jobs and incomes and other issues that we should begin addressing now to ensure that the benefits of AI are broadly shared.”

The report investigates eight areas of human activity in which AI technologies are already affecting urban life and will be even more pervasive by 2030: transportation, home/service robots, health care, education, entertainment, low-resource communities, public safety and security, employment, and the workplace.

Some of the biggest challenges in the next 15 years will be creating safe and reliable hardware for autonomous cars and health care robots; gaining public trust for AI systems, especially in low-resource communities; and overcoming fears that the technology will marginalize humans in the workplace.

Issues of liability and accountability also arise with questions such as: Who is responsible when a self-driven car crashes or an intelligent medical device fails? How can we prevent AI applications from being used for racial discrimination or financial cheating?

The report doesn’t offer solutions but rather is intended to start a conversation between scientists, ethicists, policymakers, industry leaders, and the general public.

Grosz said she hopes the AI 100 report “initiates a century-long conversation about ways AI-enhanced technologies might be shaped to improve life and societies.

This article is published in collaboration with Harvard Gazette and World Economic Forum.

If you liked this article, you may also want to read:

• Taking measure of artificial intelligence and the Turing Test
• Artificial intelligence expedites breast cancer risk prediction
• How cooperative behaviour could make artificial intelligence more human
• Artificial intelligence could transform healthcare, but we need to accept it first
• Why football, not chess, is the true final frontier for robotic artificial intelligence

See all the latest robotics news on Robohub, or sign up for our weekly newsletter.

How will artificial intelligence change the world of work? The Economist is holding a Facebook Live Q&A with their Deputy Editor on Tuesday, September 13, 4pm London time. You can watch, listen, and tune in on their Facebook page, here.

Be sure to also read The Economist: AI and Future of Work, an interactive and downloadable PDF.

You may also want to read:

• NPR podcasts discuss automation and the future of work
• ‘Robot kindergarten’ trains droids of the future
• The future of work: Robots, unions and globalization
• White House launches public workshops on AI issues
• What coachmen from the 1920s can tell us about robots and jobs
• Robohub roundtable: Job loss through automation, Foxconn controversy

See all the latest robotics news on Robohub, or sign up for our weekly newsletter.

As far as party games go, the Imitation Game is a pretty clever and fairly entertaining one. A man and a women answer questions as if they were the other, trying to fool party guests into guessing the wrong gender when identifying the answers. The idea of impersonation is too intriguing to pass up in a fun context.

This game, with its basic premise and mystery, is the basis of the infamous Turing Test. British mathematician Alan Turing used the game to describe and develop a test that measures a machine’s ability to imitate a human. The question long asked by early robotics pioneers was “Can machines think?” But Turing reframed the question to instead ask, “Can machines show they can think like a human?”

His test revolutionized how machine intelligence was viewed in relation to human thinking and functionality. The philosophy behind his question and test has guided the progress made in advancing modern artificial intelligence (AI).

While the Turing Test has historical significance and contributed a great deal to the study and growth of technology, it shouldn’t be the sole standard or measure of artificial intelligence. As the goals of artificial intelligence have made gains, the Turing Test is becoming a relic of the past and an irrelevant measure of the effectiveness of AI with a few caveats.

Complete human behavior

The Turing Test was developed to measure if a machine could act like a human, meaning it would imitate both intelligent and unintelligent behavior. Turing’s test was not a way to perfect a machine to solve a complicated problem, but to emulate the human brain—imperfections and all. In the context of the test, a machine would imitate human typing errors to try and prove it was human and not machine.

It’s important to remember that the Imitation Game—both the original party favorite and the technical version—is a game of deception not accuracy. The machine is trying to be identified as the human and the woman is trying to be identified as the man. With honest answers and the element of imitation removed, it becomes a question and answer game with very little point.

To be the ultimate player of the Turing game, machines are not programmed to produce the best answers, but the most human ones. When just answering Turing’s reframed question, “Can machines behave like humans?” the answer can most certainly be yes.

The Loebner competition, founded in 1991 as a glorified show of the Turing Test, proved that. Declared “artificial stupidity,” the Loebner competition is reviled by AI experts because instead of focusing on improving AI to do what it is intended, people are competing to prove a bot can replicate human fallibility.

Updated standards

Now, as scientists and engineers try to craft artificial intelligence for practical purposes and daily use, the importance of the Turing Test is shrinking, but not disappearing. The goal of AI now is to, in fact, solve complicated problems and equations that humans would normally struggle with. Machine learning and the advancement of how technology understands and fits into the world is becoming the standard. Machines are here to help, not simply create more human-like behavior.

On the other hand, including human-like elements in robotics and AI keeps the technology grounded in human reality. As Professor Sethu Vijayakumar FRSE, who holds a Chair in Robotics within the School of Informatics at the University of Edinburgh told Sputnik International, human forms and human behavior in robots, like the one Vijayakumar is working on to man a mission to Mars, help the people using it understand its functions in a relatable context.

His point is that we should use artificial intelligence as a way to enhance the human experience. Using technology to accomplish tasks risky to human life and using the technology to refine discoveries is an ideal application of AI and robots.

Developing AI should still involve the Turing Test to ensure there is a human element included in its design. It shouldn’t be the end all be all of AI’s effectiveness or ability to succeed in the world, but it should be a small part of determining how well it will fit into a society’s reality.

Cancer cells. Credit: CC0

Researchers have developed an artificial intelligence (AI) software that reliably interprets mammograms, assisting doctors with a quick and accurate prediction of breast cancer risk. The AI computer software intuitively translates patient charts into diagnostic information at 30 times human speed and with 99 percent accuracy.

Teaching social cues to robots could better integrate them into human society. Source: Bigstockphoto/Michal Bednarek

Cooperation is one of the hallmarks of being human. We are extremely social compared to other species. On a regular basis, we all enter into helping others in small but important ways, whether it be letting someone out in traffic or giving a tip for good service.

We do this without any guarantee of payback. Donations are made at a small personal cost but with a bigger benefit to the recipient. This form of cooperation, or donation to others, is called indirect reciprocity and helps human society to thrive.

Group-based behaviour in humans originally evolved to overcome the threat of larger predators. This has led to us having a sophisticated brain with social abilities, which is disproportionately larger in size than those of other species. The social brain hypothesis captures this idea: it proposes that the large human brain is a consequence of humans evolving in complex social groups where cooperation is a distinctive component.

Indirect reciprocity is important because we see donations happening in society despite the threat of “free riders”. These are participants who readily receive but don’t donate. This idea presents a complex interdisciplinary puzzle: what are the conditions in nature that promote donation over free-riding?

Economists, biologists, mathematicians, sociologists, psychologists and others have all contributed to examining donation behaviour. Investigation is challenging, however, because it involves observing evolution, but computer science can make an important contribution.

Using software, we can simulate simplified groups of humans in which individuals choose to help each other with different donation strategies. This allows us to study the evolution of donation behaviour by creating subsequent generations of the simplified group. Evolution can be observed by allowing the more successful donation strategies to have a greater chance of existing in the next generation of the group.

In modern times, cooperation is becoming increasingly important for engineering and technology. Many intelligent and autonomous devices, like driverless cars, drones and smartphones, are emerging and as these “robots” become more sophisticated we will need to address cooperative decision making for when they come into contact with other devices or humans.

How should these devices choose to help each other? How can exploitation by free-riders be prevented? By crossing the boundaries of traditional academic disciplines, our findings can provide helpful new insights for emerging technologies. This can allow the development of intelligence which can help autonomous technology decide how generous to be in any given situation.

Modelling evolution

To understand how cooperation may evolve in social groups, we ran hundreds of thousands of computer-simulated “donation games” between randomly paired virtual players. The first player in each pair made a decision on whether or not to donate to the other player. This was based on how they judged their reputation. If the player chose to donate, they incurred a cost and the receiver gained a benefit. Each player’s reputation was then updated in light of their action, and another game was initiated. This allowed us to observe which social comparison decisions yield a better payoff.

Social comparison is another key feature of human behaviour that we sought to include. From evolving in groups, we have become adept at comparing ourselves with others and this is highly relevant for making informed donation decisions. This is a considerable cognitive challenge when social groups are large, so sizing up others in this way could have helped to promote the evolution of larger human brains.

The particular donation behaviour we used in our research was based on players making self comparisons of reputation. This leads to a small number of possible outcomes, for example, relative to myself, your reputation could be considered either broadly similar, higher, or lower. The major element of thinking comes from estimating someone’s reputation in a meaningful way.

Being human is about more than just looking the part.

Our results showed that evolution favours the strategy to donate to those who are at least as reputable as oneself. We call this “aspirational homophily”. This involves two main elements: first, being generous maintains a high reputation; second, not donating to lower reputation players helps to prevent free-riders.

It is important to remember that our results are from a simplified model: the donation decisions involved no exceptions that may occur in real life, and economic resources are assumed to drive behaviour rather than emotional or cultural factors. Nevertheless, such simplification allows us to gain useful clarity.

Most importantly, the social brain hypothesis is supported by our findings: the large human brain is a consequence of humans evolving in complex social groups where cooperation is a distinctive component. Understanding this through computing opens up a new line of thought for the development of sophisticated social intelligence for autonomous systems.

This article was originally published on The Conversation. Read the original article.

Scientists in Japan reportedly saved a woman’s life by applying artificial intelligence to help them diagnose a rare form of cancer. Faced with a 60-year-old woman whose cancer diagnosis was unresponsive to treatment, they supplied an AI system with huge amounts of clinical cancer case data, and it diagnosed the rare leukemia that had stumped the clinicians in just ten minutes.

The Watson AI system from IBM matched the patient’s symptoms against 20m clinical oncology studies uploaded by a team headed by Arinobu Tojo at the University of Tokyo’s Institute of Medical Science that included symptoms, treatment and response. The Memorial Sloan Kettering Cancer Center in New York has carried out similar work, where teams of clinicians and data analysts trained Watson’s machine learning capabilities with oncological data in order to focus its predictive and analytic capabilities on diagnosing cancers.

IBM Watson first became famous when it won the US television game show Jeopardy in 2011. And IBM’s previous generation AI, Deep Blue, became the first AI to best a world champion at chess when it beat Garry Kasparov in a game in 1996 and the entire match when they met again the following year. From a perspective of technological determinism, it may seem inevitable that AI has moved from chess to cancer in 20 years. Of course, it has taken a lot of hard work to get it there.

But efforts to use artificial intelligence, machine learning and big data in healthcare contexts have not been uncontroversial. On the one hand, there is wild enthusiasm – lives saved by data, new medical breakthroughs, and a world of personalised medicine tailored to meet our needs by deep learning algorithms fed by smartphones and FitBit wearables. On the other there’s considerable scepticism – a lack of trust in machines, the importance of individuals over statistics, privacy concerns over patient records and medical confidentiality, and generalised fears of a Brave New World. Too often the debate dissolves into anecdote rather than science, or focuses on the breakthrough rather than the hard slog that led to it. Of course, the reality will be somewhere in the middle.

There’s not just a technical battle to win

In fact, it may surprise you to learn that the world’s first computerised clinical decision-support system, AAPhelp, was developed in the UK way back in 1972 by Tim De Dombal and one of my colleagues, Susan Clamp.

This early precursor to the genius AI of today used a naive Bayesian algorithm to compute the likely cause of acute abdominal pain based on patient symptoms. Feeding the system with more symptoms and diagnosis helped it to become more accurate over time and, by 1974, De Dombal’s team had trained the system to the point where it was more accurate at diagnosis than junior doctors, and almost as accurate as the most senior consultants. It took AAPhelp overnight to give a diagnosis, but this was on 1970s computer hardware.

The bad news is that 40 years on, AAPhelp is still not in routine use.

This is the reality check for the most ardent advocates of applying technology to healthcare: to get technology such as predictive AIs into clinical settings where they can save lives means tackling all those negative connotations and fears. AI challenges people and their attitudes: the professionals that the machine can outperform, and the patients that are reduced to statistical probabilities to be fed into complex algorithms. Innovation in healthcare can take decades.

Nevertheless, while decades apart both AAPHelp and IBM Watson’s achievements demonstrate that computers can save lives. But the use of big data in healthcare implies that patient records, healthcare statistics, and all manner of other personal details might be used by researchers to train the AIs to make diagnoses. People are increasingly sensitive to the way personal data is used and, quite rightly, expect the highest standards of ethics, governance, privacy and security to be applied. The revelations that one NHS trust had given access to 1.6m identifiable patient records to Google’s DeepMind AI laboratory didn’t go down well when reported a few months ago.

The hard slog is not creating the algorithms, but the patience and determination required to conduct careful work within the restrictions of applying the highest standards of data protection and scientific rigour. At the University of Leed’s Institute for Data Analytics we recently used IBM Watson Content Analytics software to analyse 50m pathology and radiology reports from the UK. Recognising the sensitivities, we brought IBM Watson to the data rather than passing the data to IBM.

Using natural language processing of the text reports we double-checked diagnoses such as brain metastases, HER-2-positive breast cancers and renal hydronephrosis (swollen kidneys) with accuracy rates already over 90%. Over the next two years we’ll be developing these methods in order to embed these machine learning techniques into routine clinical care, at a scale that benefits the whole of the NHS.

While we’ve had £12m investment for our facilities and the work we’re doing, we’re not claiming to have saved lives yet. The hard battle is first to win hearts and minds – and on that front there’s still a lot more work to be done.

This article was originally published on The Conversation. Read the original article.

Disclosure statement

Owen A Johnson receives research funding from MRC, EPSRC, NIHR, the NHS and InnovateUK. He is a director of X-Lab Ltd., an e-health software company focused on disruptive innovation in healthcare.

If you liked this article you may also want to read:

• Talking Machines: Machine learning in healthcare and the AlphaGo matches
• The state of telepresence: Healthcare and telemedicine
• Healthcare robotics for therapy, with Ayanna Howard
• Living with a prosthesis that learns: A case-study in translational medicine
• Artificial intelligence is a tool, not a threat

See all the latest robotics news on Robohub, or sign up for our weekly newsletter.

Baxter throwing shade? Source: YouTube

Stories about racist Twitter accounts and crashing self-driving cars can make us think that artificial intelligence (AI) is a work in progress. But while these headline-grabbing mistakes reveal the frontiers of AI, versions of this technology are already invisibly embedded in many systems that we use everyday.

These everyday uses include everything from fraud detection systems that monitor credit card transactions to email filters that learn not to swamp your inbox with spam. You’ve probably already interacted with an AI system today without even knowing it and probably enjoyed the experience.

One increasingly common form of AI can be found in chatbots, a type of software that lets you interact with it by having a conversation. The iPhone assistant technology, Siri, is an obvious example. Microsoft’s experimental Twitter account that learned how to speak from other users and ended up spouting racist phrases is another. But many websites and apps are now using chatbots to let people order services or locate specific information – without descending into bigotry.

For example, Amy is an AI assistant that schedules meetings for you via email exchanges with your contacts. Very few of these chatbots could pass themselves off completely as a human, however, so their designers need to think carefully about how people react to AI if they want their creations to be accepted. Otherwise it ends up feeling like you’re talking to a really bad PA.

Teaching a machine

There are many different approaches to make these digital machines behave in an intelligent way that mimics human behaviour. But what all of them have in common is that they base what they are doing on huge amounts of data that they have gathered from their environment.

Chatbots are often “trained” by being given months of Twitter traffic as examples which is then analysed using complex statistical methods to find frequent patterns of behaviour. For example “fine, thank you” is a frequent response to a question such as “how are you?”. Quite often, AI will not truly understand what it is saying, it will simply repeat what it has seen.

Having a conversation with another human is actually quite complex. You need to first recognise the words in a sentence, know when it is your turn to answer, then generate your own appropriate response that relates to the point of the conversation. Several things can go wrong, from simply not knowing a word to getting the intent of the conversation wrong. Obviously, the more errors there are, the less you think the conversation is going well, and in the worst case, you might stop interacting.

We already know that people will interact differently with a human than a machine. They trust AI less, they do not engage as deeply with it, and they will talk to it in a simpler way than with real humans. In fact, there is evidence that the more the machine tries to mimic a real human conversation, the more off-putting it is, similar to the “uncanny valley” effect that happens the more humanoid robots look.

So how can we design an AI system that is more acceptable to people? First, better and more examples of correct behaviour are needed so that it makes fewer errors. People need to start working hand-in-hand with machines to shape the behaviour of AI systems.

What also seems to matter is how much a user understands how a system works. For example, a recent study on conversational agents found that people wanted to know what the system could do, what is was doing, how it was doing it and whether it was changing due to how the user was interacting with it in the past. This point seems to apply to all kinds of AI, as transparency of an AI system seems to have a positive impact on user satisfaction.

Make it less human

Obviously, people are less likely to trust error-prone systems. But they also don’t want AI to act by itself without any confirmation. For example, if you know a system often misunderstands you then you would not want it to dial a phone number without first checking it is correct. The system also needs to make clear to the user that it’s a robot. It won’t be like talking to another human, and that’s quite ok.

We can expect to see AI systems become more accurate and more integrated into everyday life, but there will also be spectacular failures. Mostly, these systems work fine but what do we do when they don’t? Since the dawn of science fiction, there have been questions about the ethics and laws of AI and how we can control it, which continue to this day. These are still open research questions that have to be answered, along with where AI should and shouldn’t be used, and who is responsible for making decisions and ultimately answerable for mistakes.

In the meantime, more and more companies are starting to integrate AI into their systems and products, with some success. Google’s Nest Learning Thermostat – which memorises your schedule and changes depending on how you use it – is one obvious example but there are scores of start-ups that now leverage the power of AI to provide a personalised experience for consumers. And thanks to the rise in data science that provides the information that will teach these systems, there has never been a better time for firms to turn to the power of AI.

This article was originally published on The Conversation. Read the original article.

Screenshot from presentation about deep learning we may use everyday. Source: Frank Chen, with Andreessen Horowitz

Frank Chen, a partner at Andreessen Horowitz, the Silicon Valley venture capital and private equity firm, said, “It is absolutely non-controversial that deep learning is the most fundamental advance in AI research since the start [of A.I.] in 1956.”

In the 45-minute podcast primer shown below – a worthwhile investment of your time – he went on to say:

“We [Andreessen Horowitz] think AI and deep learning can be a fundamental and technology platform shift as mobile and cloud have been in the last 5-10 years … All the serious applications from here on out need to have deep learning and AI inside in exactly the same way that all serious computing systems needed to have Intel chips inside of them. I think now about the startups that we see, that deep learning needs to be inside these new systems as a fundamental technique that we expect to see in all serious applications moving forward.”

Enjoy. It’s an informative primer.

Julie Wosk is the author of a provocative book that gives the history of female robots in movies, television, art, literature, and includes a chapter on robotics. My Fair Ladies: Female Robots, Androids, and Other Artificial Eves (Rutgers University Press) features the way men have used science and technology to create their idea of “The Perfect Woman” — women like the beautiful robots in The Stepford Wives that are always sexually available and love to cook and clean. The book also highlights today’s women in robotics who are challenging the old stereotypes by using their skills and expertise to develop personal robots for the future. Here is a short statement from the author and excerpt from her chapter, “Dancing With Robots and Women in Robotics Design.”

In my book, I see multiple tracks in robotics today. There are humanoid female robots being developed by roboticists like Professor Hiroshi Ishiguro of Osaka University in Japan, and David Hanson of Hanson Robotics based in Hong Kong. These attractive robots which usually look like young women in their twenties have expressive faces and rudimentary interactive speaking capabilities. So far, Professor Ishiguro has used his robots for roles like a receptionist and even an actor (in the play Sayonara by Oriza Hirata, Ishiguro’s female robot Geminoid F plays the role of a caretaker for a terminally ill young woman in a near-future world. The 2015 film version by noted director Kôji Fukada will be showing at Manhattan’s Japan Society on July 17, 2016).

There are also sex robots like Douglas Hines’ True Companion Roxxxy sex robots that are interactive and can have electronic “conversations.” For users, these silicone robots have special appeal: they are totally compliant and unlike real women, don’t have any needs or minds of their own.

And there are more gender neutral, socially-aware robots with the capacity for empathy  being developed by American roboticists, such as Professor Cynthia Breazeal at MIT and Professor Andrea Thomaz at Georgia Institute of Technology who are working with colleagues to create  robots for social use in areas such eldercare and childcare.

The female robots described in My Fair Ladies (including robots in the old vintage television Twilight Zone and Star Trek series) helped set the stage for today’s movies like Ex Machina where the robot Ava is intelligent, alluring, empathetic, and seductive. These  earlier robots often reflect men’s fantasies, and are forerunners of the latest Roxxxy sex robots advertised as using technology to “provide a perfect partner.” On its website, Hanson Robotics pictures its current prototype robot named Sophia and asks coyly, “Could You Fall in Love With This Robot?” Even here, it seems hard for men not to picture female robots as love objects.

We are left wondering: Should women worry? Will these “perfect females” replace them sometime in the future?

Chapter adapted from My Fair Ladies: Female Robots, Androids, and Other Artificial Eves. New Brunswick: Rutgers University Press, 2015. Copyright © 2015 by Julie Wosk.  Reprinted by permission of Rutgers University Press.

Download (PDF)

Robot from Computer Science and Artificial Intelligence Lab suggests where to move patients and who should do C-sections.

By: Adam Conner-Simons

Today’s robots are awkward co-workers because they are often unable to predict what humans need. In hospitals, robots are employed to perform simple tasks like delivering supplies and medications but they have to be explicitly told what to do.

A team from MIT’s Computer Science and Artificial Intelligence Lab (CSAIL) thinks that this will soon change, and that robots might be most effective by helping humans perform one of the most complex tasks of all: scheduling.

In a pair of new papers, CSAIL researchers demonstrate a robot that, by learning from human workers, can help assign and schedule tasks in fields ranging from medicine to the military.

In one paper, the team demonstrated a robot that assisted nurses in a labor ward, where it made recommendations on everything from where to move a patient to which nurse to assign to a C-section.

“The aim of the work was to develop artificial intelligence that can learn from people about how the labor and delivery unit works, so that robots can better anticipate how to be helpful or when to stay out of the way – and maybe even help by collaborating in making challenging decisions,” says MIT professor Julie Shah, the senior author on both papers.

In a second paper, the same system was put to the test in a videogame that simulates missile-defense scenarios. In the game, which was developed by Lincoln Laboratory researchers and involves using decoy missiles to ward off enemy attacks, the system even occasionally outperformed human experts at reducing both the number of missile attacks and the overall cost of decoys.

The labor-ward paper was presented at the recent Robotics: Science and Systems (RSS) Conference and was co-written by PhD student Matthew Gombolay, CSAIL postdocs Xi Jessie Yang and Brad Hayes, and Dr. Neel Shah and Toni Golen from Beth Israel Deaconess Medical Center, which is where the study took place.

The Navy-simulation paper is being presented at this week’s International Joint Conference on Artificial Intelligence (IJCAI), and was co-written by Gombolay and the  Lincoln Lab’s Reed Jensen, Jessica Stigile and Sung-Hyun Son.

Right on schedule

From visiting hospitals and factories, Shah and Gombolay found that a subset of workers are extremely strong schedulers, but can’t easily transfer that knowledge to colleagues.

“Figuring out what makes certain people good at this often seems like a mystery,” Gombolay says. “Being able to automate the task of learning from experts – and to then generalize it across industries – could help make many businesses run more efficiently.”

A particularly tough place for scheduling are hospitals. Labor wards’ head nurses have to try to predict when a woman will arrive in labor, how long labor will take, and which patients will become sick enough to require C-sections or other procedures.

They are deluged with an endless stream of challenging split-second decisions that include assigning nurses to patients, patients to beds, and technicians to surgeries. At Beth Israel, the head nurse has to coordinate 10 nurses, 20 patients and 20 rooms at the same time, meaning that the number of distinct scheduling possibilities adds up to a staggering 2 to the one millionth power, which is more than the number of atoms in the universe.

“We thought a complex environment like a labor ward would be a good place to try to automate scheduling and take this significant burden off of workers,” says Gombolay.

How it works

Like many AI systems, the team’s robot was trained via “learning from demonstration,” which involves observing humans’ performances of tasks. But Gombolay says that researchers have never been able to apply this technique to scheduling, because of the complexity of coordinating multiple actions that can be very dependent on each other.

To overcome this, the team trained its system to look at several actions that human schedulers make, and compare them to all the possible actions that are not made at each of those moments in time. From there, it developed a scheduling policy that can respond dynamically to new situations that it has not seen before.

“Rather than considering actions in isolation of each other, we crafted a model that understands why one action is better than the alternatives,” says Shah. “By considering all such comparisons, you can learn to recommend which action will be most helpful.”

The policy is “model-free,” meaning that the nurses do not have to train the robot by painstakingly ranking each possible action in each possible scenario by hand.

“You can put the robot on the labor floor, and, just by watching humans doing the different tasks, it will understand how to coordinate an efficient schedule,” says Gombolay.

The results

With this framework, the system – which the team has dubbed “apprenticeship scheduling” – can anticipate room assignments and suggest which nurses to assign to patients for C-sections and other procedures.

The approach was evaluated through experiments in which a robot provided decision-support to nurses and doctors as they made decisions on a labor floor.

Using the system on a Nao robot, nurses accepted the robot’s recommendations 90 percent of the time. The team also demonstrated that human subjects weren’t just blindly accepting advice – the robot delivered consciously bad feedback that was rejected at the same rate of 90 percent, showing that the system was trained to distinguish between good and bad recommendations.

Nurses had almost uniformly positive feedback about the robot. One said that it would “allow for a more even dispersion of workload,” while another said that it would be particularly helpful for “new nurses [who] may not understand the constraints and complexities of the role.”

“A great potential of this technology is that good solutions can be spread more quickly to many hospitals and workplaces,” says Dana Kulic, an associate professor of computer engineering at the University of Waterloo. “ For example, innovative improvements can be distributed rapidly from research hospitals to regional health centres.”

Shah says that the new techniques have many uses, from turning robots into better collaborators to helping train new nurses, but the goal is not to develop robots that fully make decisions on their own.

“These initial results show there is tremendous potential for machines to collaborate with us in rich ways that will enhance many sectors of the economy,” says Shah. “The awkward robots of the past will be replaced by valued team members.”

The RSS paper was supported by the National Science Foundation, CRICO Harvard Risk Management Foundation, and Aldebaran Robotics Inc. The IJCAI paper was supported the National Science Foundation and the U.S. Navy.

Read the paper here. 

Photo: Xi Jessie Yang

MIT-SUTD researchers are creating improved interfaces to help machines and humans work together to complete tasks.

As machines become more intelligent, they become embedded in countless facets of life. In some ways, they can act almost as full-fledged members in human-machine teams. In such cases, as with any team, trust is a necessary ingredient for good performance.

But the dynamics of trust between people and machines are not yet well-understood. With a two-year project funded through the SUTD-MIT Postdoctoral Fellows Program, a collaboration between MIT and the Singapore University of Technology and Design (SUTD), postdoc Xi Jessie Yang, MIT Department of Aeronautics and Astronautics Assistant Professor Julie Shah, and SUTD Engineering Product Development Professor Katja Hölttä-Otto aim to develop greater knowledge in this area. During her fellowship, Yang spent one year as a postdoc at SUTD, and is now finishing her research during a second year at MIT. She and her faculty mentors hope their efforts will contribute to better user interface design and better human-machine team performance.

Some tasks can simply be accomplished more effectively and efficiently by a machine or with assistance from a machine than by a human alone. But machines can make mistakes, especially in scenarios characterized by uncertainty and ambiguity. As such, a proper calibration of trust towards machines is a prerequisite of good human-machine team performance.

A variety of factors intersect to determine the extent to which the human trusts the machine, and in turn, how well they can perform on a given task. These factors include how reliable the machine is, how self-confident the human is, and the concurrent workload (i.e. how many tasks the human is responsible for carrying out simultaneously). Teasing out the effect of any one factor can be challenging, but through intricate experimental design, Yang, Shah and Hölttä-Otto have been able to approach the topic in a more realistic way, dealing with the imperfection of machines and isolating various factors in the human-machine trust relationship to study their impact.

At SUTD, Yang and Hölttä-Otto designed an experiment involving a human participant performing a memory and recognition task with the assistance from an automated decision aid. In the experiment, participants memorized a series of images and later recognized them in a pool of similar images. The decision aid provided recommendations for the images to select, very reliably for some people and not so for others. Results revealed that participants tended to undertrust highly reliable aids and overtrust highly unreliable ones. The trust-reliability calibration also improved as humans gained more experience using the decision aid.

At MIT, Yang and Shah explored how interface design could facilitate trust-reliability calibration. Specifically, of interest to Yang and Shah is the potential benefits of likelihood alarm displays. In contrast to traditional binary alarms indicating only “threat” or “no threat”, likelihood alarms provide additional information about the confidence level and the urgency of an alerted event. Yang and Shah believe that likelihood alarm would help mitigate the “cry wolf” effect, a phenomenon commonly observed in high-risk industries. In such industries, the threshold to trigger an alarm are oftentimes set very low in order to capture every critical event. The low threshold, however, inevitably results in false alarms which cause the human users to lose trust in the alarm system and eventually harm task performance.

In another experiment, currently underway, Yang and Shah simulate a multi-task scenario in which participants act as a solider to perform a search and detection task. They have to maintain level flight of four unpiloted aerial vehicles (UAVs) and at the same time detect potential threats in the photo steams sent back from the UAVs. Either a binary alarm or a likelihood alarm helps the participants to complete their mission. The latter would presumably alleviate the “cry wolf” effect, as well as improve participants’ attention allocation, and enhance human-machine team performance.

Ultimately, this research could help inform robotics and artificial intelligence development across many domains, from military applications like the simulations in the experiments to medical diagnostics to airport security.

“Lack of trust is often cited as one of the critical barriers to more widespread use of AI and autonomous systems,” Shah says. “These studies make it clear that increasing a user’s trust in the system is not the right goal. Instead we need new approaches that help people to appropriately calibrate their trust in the system, especially considering these systems will always be imperfect. ”

Yang will be presenting her research at the 7th International Conference on Applied Human Factors and Ergonomics this July, as well as the 60th annual meeting of the Human Factors and Ergonomics Society in September.

The wrench 3D-printed by the Additive Manufacturing Facility (AMF) aboard the International Space Station has the name of toolmaker Kobalt emblazoned on its side — the result of a partnership between Made In Space, which built the AMF, and home-improvement store Lowe’s.
Credit: NASA/Made In Space/Lowe’s

Investing in robots; China’s robotics acquisitions; robot designed to hurt; exoskeleton for kids; Russian bot escapes testing ground; 3-DP in space and more. Find out what’s happening in our robotics universe this week.

Want to invest in the robot uprising? Here’s how (WIRED)

An interesting new option for those thinking about investing in robotics was revealed this week.

Dubbed Motif, it’s “a concept-driven trading platform that creates intelligently weighted baskets of up to 30 stocks or exchange-traded funds — called ‘motifs’ — that you can buy into for $9.95.”

Motif’s major innovation is letting individuals own fractions of shares in company, something that’s typically not possible when buying stocks on the open market. Users can invest, say, $300 in a motif containing a weighted basket of 30 different company stocks, each of whose share prices could be tens or hundreds of dollars, and constantly fluctuating. Motif’s technology makes it possible for individuals to immediately own the underlying fractions of shares contained in the basket at the point of purchase.

Motif also gives individual investors fractional ownership on single shares, allowing a user with only $300 to still buy a piece of Alphabet (formerly known as Google) even though a single share is currently trading at more than $700.

Walia wouldn’t go into details about how exactly this works as he says there are several other startups interested in cloning his technology. What he will say is that Motif doesn’t aggregate trades. “We have a lot of patents”, he adds.

China continues to invest in European Industrial Automation (Robotics Business Review)

With so many Chinese firms investing in Western robotics companies, this article offers a timely breakdown of the most important recent acquisitions and offers.

AGIC Capital is spending €100 million ($113 million) to acquire a majority stake in Gimatic SRL, which makes end-of-arm tools for robotic arms.

Italy-based Gimatic supplies actuators, grippers, sensors, and plastic injection-molding equipment to the automotive, electronics, food, and pharmaceuticals industries. It has more than 100 patents and facilities worldwide, and its revenue has grown by more than 20 percent over the past three years.

This robot intentionally hurts people—and makes them bleed (Fast Company)

Alexander Reben, a Berkeley, California-based roboticists and artist, has created a tabletop robot that’s designed to hurt people –and generate a conversation among roboticists, ethicists, lawyers, and other stakeholders about the challenges robotics could present in the future.

Before you bar your doors and windows, let’s define the terms: The harm caused by Reben’s robot is nothing more than a pinprick, albeit one delivered at high speed, causing the maximum amount of pain a small needle can inflict on a fingertip.

And interestingly, he designed the machine so that injury is inflicted randomly. Sometimes the robot strikes. Sometimes it doesn’t. Even Reben, when he exposes his fingertip to danger, has no idea if he’ll end up shedding blood or not.

“No one has actually made a robot that was built to intentionally hurt and injure someone,” Reben claims in a giant room on the top floor of the beautiful Victorian mansion where he lives and works as a member of Stochastic Labs, a Berkeley arts, technology, and science events incubator. “I wanted to make a robot that does this that actually exists… It was important to take it out of the thought experiment realm into reality, because once something exists in the world, you have to confront it. It becomes more urgent. You can’t just pontificate about it.”

The headline on this story is slightly misleading, inasmuch as robots do not possess ‘intentionality’ in any human sense. Robots act in accordance with their design and they follow instructions.

If you’ve been following our ISO/TS 15066 coverage, you’ll have noticed that the new ISO document incorporates the results of a University of Mainz study that developed force and pressure limits for 29 body areas. Designed to minimize harm should any contact occur between a human and a robot in a collaborative environment, the study helps designers and integrators to perform risk assessments for collaborative robot systems.

ISO/TS 15066 was six years in the making and received input from leading robotics experts from all over the world. So, while the artist’s project is interesting and cool, the conversation around safety issues has been going on for quite some time among roboticists.

Spanish researchers design child exoskeleton (R&D)

Spinal muscular atrophy (SMA) is a genetic condition defined by muscle weakness, which often makes locomotion difficult for sufferers.

The good news is that the Spanish National Research Council and spin-off company Marsi Bionics have created an exoskeleton for children with SMA aged between ages three and 14. Marsi’s intelligent exoskeleton is currently in the preclinical phase, but if it’s successful, researchers hope that as well as helping with locomotion, the device could stave off secondary effects from the condition, such as scoliosis.

The exoskeleton fits around the child’s legs and torso, and consists of five motors in each leg that act as muscles, providing strength for the child to stand upright and walk.

“Our model includes intelligent joints which alter the brace’s rigidity automatically and adapt to the symptoms of each individual child … whenever required,” said Elena Garcia, part of the research council’s Automatics and Robotics Centre, in a statement.

Additionally, the apparatus is highly sensitive to a child’s intended movement and acts accordingly. It’s powered by a battery with a five-hour lifespan.

“The number one drawback in developing this type of pediatric exoskeleton is that the symptoms of neuromuscular illnesses—such as spinal muscular atrophy—change over time, as much in the articulation as in the body,” Garcia said in a statement. “That’s why it’s fundamental to have an exoskeleton capable of independently adapting to these changes.”

Robot escapes testing grounds in Russia (Sputnik)

Oh boy.

Robots be rolling. I think somebody’s trolling.

Reports emerged this week of a Russian robot ‘escaping’ testing grounds and finding its way onto the street, where its batteries went flat and the robot held up traffic until the police arrived. What do you think? Real or a publicity stunt?

The escape occurred due to a gate left open by an engineer, local media said. The robot had moved some 50 meters (160 feet) by the time its charge ran out, according to reports.

The loss was noticed almost an hour later, by which time the robot caused a minor traffic jam and gathered a crowd, with the police eventually arriving.

Space Station’s commercial 3D printer makes its 1st tool (Space.com)

The machine, known as the Additive Manufacturing Facility (AMF), was installed aboard the International Space Station (ISS) in late April. And last week, it printed out its first tool — a wrench that astronauts can use to do maintenance work aboard the orbiting lab.

The wrench was made with a special off-Earth feature — a fastening clip that should help astronauts keep track of the tool in zero g.

The AMF was built by California-based company Made In Space, which also constructed the only other 3D printer ever sent to space.

And Finally…

https://www.youtube.com/watch?v=VcMoDHbWYbk

https://www.youtube.com/watch?v=imvL5bsk87M

https://www.youtube.com/watch?v=QMzVJ8g6fMI

Infographic: The making of a collaborative robot (Robohub via National Science Foundation)

How will bots make money? Here are 7 business models (Venture Beat)
How To Buy A Drone [Infographic] (Drone360)
First Surgical Robot from Secretive Startup Auris Cleared for Use (IEEE Spectrum)
Jibo “social” robot to ship in October (Liliputing)
Pepper the robot begins work in Belgian hospitals (Daily Mail)
British, Indian scientists to use underwater robots in study of subcontinent’s monsoon (Japan Times)