OpenAI takes the robotic imitation of human behavior into a whole new level

OpenAI, anElon Musk-backed nonprofit artificial intelligence platform just announced a new milestone in training robots. They are working with a new algorithm known as one-shot imitation learning, which lets human being train a robot by demonstrating it first in virtual reality.

In the video below, a person is trying to teach a robotic arm how to stack a series of colored cube-shaped blocks by first performing it manually within a VR environment. The whole system is powered by two neural networks. The first one determines the objects spatial position to the robot by taking a camera image. However, the neural network was trained only with a host of simulated images, which means it knew how to cooperate with the real world before it ever actually met it. The second one emulates any task the demonstrator shows it by scanning through recorded actions and observing frames telling it what to do next.

Our robot has now learned to perform the task even though its movements have to be different than the ones in the demonstration, explains Josh Tobin, a member of OpenAIs technical staff. With a single demonstration of a task, we can replicate it in a number of different initial conditions. Teaching the robot how to build a different block arrangement requires only a single additional demonstration.

The model is currently a prototype, but this concept could help researchers in the long run. They could use this concept to teach the robots more complex tasks in future without using any physical elements at all. OpenAIs long term plan is to give the AI the ability to learn to adapt to unpredictable changes in the environment.

Read more:
OpenAI takes the robotic imitation of human behavior into a whole new level - The TechNews

When people work together on a project, they often come to think theyve figured out the problems in their own respective spheres. If trouble persists, its somebody elseengineering, say, or the marketing departmentwho is screwing up. That local focus means finding the best way forward for the overall project is often a struggle. But what if adding artificial intelligence to the conversation, in the form of a computer program called a bot, could actually make people in groups more productive?

This is the tantalizing implication of a study published Wednesday in Nature. Hirokazu Shirado and Nicholas Christakis, researchers at Yale Universitys Institute for Network Science, were wondering what would happen if they looked at artificial intelligence (AI) not in the usual wayas a potential replacement for peoplebut instead as a useful companion and helper, particularly for altering human social behavior in groups.

First the researchers asked paid volunteers arranged in online networks, each occupying one of 20 connected positions, or nodes, to solve a simple problem: Choose one of three colors (green, orange or purple) with the individual, or local, goal of having a different color from immediate neighbors, and the collective goal of ensuring that every node in the network was a different color from all of its neighbors. Subjects pay improved if they solved the problem quickly. Two thirds of the groups reached a solution in the allotted five minutes and the average time to a solution was just under four minutes. But a third of the groups were still stymied at the deadline.

The researchers then put a botbasically a computer program that can execute simple commandsin three of the 20 nodes in each network. When the bots were programmed to act like humans and focused logically on resolving conflicts with their immediate neighbors, they didnt make much difference. But when the researchers gave the bots just enough AI to behave in a slightly noisy fashion, randomly choosing a color regardless of neighboring choices, the groups they were in solved the problem 85 percent of the timeand in 1.7 minutes on average, 55.6 percent faster than humans alone.

Being just noisy enoughmaking random color choices about 10 percent of the timemade all the difference, the study suggests. When a bot got much noisier than that, the benefit soon vanished. A bots influence also varied depending on whether it was positioned at the center of a network with lots of neighbors or on the periphery.

So why would making what looks like the wrong choicein other words, a mistakeimprove a groups performance? The immediate result, predictably, was short-term conflict, with the bots neighbors in effect muttering, Why are you suddenly disagreeing with me? But that conflict served to nudge neighboring humans to change their behavior in ways that appear to have further facilitated a global solution, the co-authors wrote. The humans began to play the game differently.

Errors, it seems, do not entirely deserve their bad reputation. There are many, many natural processes where noise is paradoxically beneficial, Christakis says. The best example is mutation. If you had a species in which every individual was perfectly adapted to its environment, then when the environment changed, it would die. Instead, random mutations can help a species sidestep extinction.

Were beginning to find that errorand noisy individuals that we would previously assume add nothingactually improve collective decision-making, says Iain Couzin, who studies group behavior in humans and other species at the Max Planck Institute for Ornithology and was not involved in the new work. He praises the deliberately simplified model used in the Nature study for enabling the co-authors to study group decision-making in great detail, because they have control over the connectivity. The resulting ability to minutely track how humans and algorithms collectively make decisions, Couzin says, is really going to be the future of quantitative social science.

But how realistic is it to think human groups will want to collaborate with algorithms or botsespecially slightly noisy onesin making decisions? Shirado and Christakis informed some of their test groups that they would be partnering with bots. Perhaps surprisingly, it made no difference. The attitude was, I don't care that youre a bot if youre helping me do my job, Christakis says. Many people are already accustomed to talking with a computer when they call an airline or a bank, he adds, and the machine often does a pretty good job. Such collaborations are almost certain to become more common amid the increasing integration of the internet with physical devices, from automobiles to coffee makers.

Real-world, bot-assisted company meetings might not be too far behind. Business conferences already tout blended digital and in-person events, featuring what one conference planner describes as integrated online and offline catalysts that use virtual reality, augmented reality and artificial intelligence. Shirado and Christakis suggest slightly noisy bots are also likely to turn up in crowdsourcing applicationsfor instance, to speed up citizen science assessment of archaeological or astronomical images. They say such bots could also be useful in social mediato discourage racist remarks, for example.

But last year when Microsoft introduced a twitter bot with simple AI, other users quickly turned it into epithetspouting bigot. And the opposite concern is that mixing humans and machines to improve group decision-making could enable businessesor botsto manipulate people. Ive thought a lot about this, Christakis says. You can invent a gun to hunt for food or to kill people. You can develop nuclear energy to generate electric power or make the atomic bomb. All scientific advances have this Janus-like potential for evil or good.

The important thing is to understand the behavior involved, so we can use it to good ends and also be aware of the potential for manipulation, Couzin says. Hopefully this new research will encourage other researchers to pick up on this idea and apply it to their own scenarios. I dont think it can be just thrown out there and used willy-nilly.

Go here to see the original:
Pushy AI Bots Nudge Humans to Change Behavior - Scientific ... - Scientific American

There are few people in the design world more familiar with field research, and the extensive travel that goes with it, than Jan Chipchase. On May 6th, Jan launched what has turned out to be a very successful Kickstarter campaign to publish his most recent book,The Field Study Handbook. There's still time to pre-order a copy of what looks to be a fascinating read, even if field research in the Hindu Kush is not on your immediate list of things to do.

Jan is a researcher, writer and photographer whose work focuses on the intersection of design, tech, human behavior and culture. Over the years, he's led research teams investigating both mainstream and emerging markets for Nokia and frog design. In 2014, he founded Studio D, a research, design and innovation consultancy, and later used his extensive travel experiences to create an ultra-light luggage brand, SDR Traveller.

I had a chance to catch up with Jan to talk about the book launch and other recent adventures and activities, as well as ask for some travel advice.

This interview has been edited for length and clarity.

Core77: In addition to Studio D and SDR Traveller, I see also that you run as service called The Fixer List. What is that all about? It seems mysterious.

JC: Fixers are a staple of field research, at least with the way I run projects. Each project includes a local crew that is usually led by someone I call a "fixer". They come from very diverse backgrounds, often speak multiple languages, have a very good sense of their home locale and know how to hustle. Over the years, many of these connections have stayed in touch.

The Fixer List is the Studio D list of unusual talent that we can draw on to run projects around the world. We receive a lot of applicants.

You spend a large percentage of your time on the road, traveling extensively across the world. What are some notable recent field trips you've taken?

JC: Saudi Arabia was interesting and challenging. We were there gathering insights to understand the value proposition of a new brand offering. The new service, Jawwy, went live last year. The team had to achieve a high level of understanding of the local culture and how it maps to a mobile service, in only a month. Many foreigners would struggle to achieve this level of understanding in a lifetime. All credit to our local team of ten people for getting us that far.

Last year I took a tough, 7,000km overland expedition through Kyrgyzstan, Tajikistan's GBAO region and China's western provinces. The trip was full of remote communities, dubious border crossings, permeable borders and lots of checkpoints. I learned a lot on that trip and wrote up my thoughts in a Medium post called 61 Glimpses of the Future.

Another interesting trip involved training a client's team on field research methods, including setting up a mountain retreat at the edge of the Tibetan Plateau to process the data. On every project we get to ask "Where is the best place to figure out x?" and "Where do we want to be today?"

Naturally, the Kickstarter includes expeditions as rewards, if you're up for A Short Walk in the Hindu Kush.

In the Kickstarter video, you mention that the book is part of a mission to reframe the relationship between those that make things and those that consume them. Can you talk about this a bit?

JC: Increasingly, the impetus for creating new products will be based on the data analytics of mass consumption, feedback on marketing strategies, and optimized value engineering. The product creators are removed from their audience by several steps, and it is easy to lose touch with that actual audience. The ability to gather an over abundance of data only compounds the problem. As more data comes on stream, revealing what people are doing and how, there is a growing danger of people being treated as little more than lines in a database, stripped of personality and context, there solely to be mined and monetized.

Field research is defined by closeness and empathy. You get close to those you are studying, and in doing so, develop a deeper empathy for their lives and ways of living. You then take that empathy andin the best casesreflect it back on them through your work. In the right hands, it generates very rich, and very nuanced, data that is capable of answering why people do what they do. Understanding the motivations behind people's actions can lead to very different outcomes, if all you knew before was what people did and how.

I try (and usually fail) to not pack too much when I travel. As a seasoned traveller what are some of your suggestions and tips for packing?

JC: Everybody over packs. It's human nature! By packing less, you're actually more flexible and better able to alter your planned itinerary to react to interesting opportunities. My own preference is to use hand-held luggage (no wheels allowed!), make sure the bag can fit under an economy-class seat or a business class foot-well, and even with that small size, still leave room in your bag for things you might pick up along the way. I wrote about how the psychology of packing impacts the experience of the journey, and it still holds true.

Check out the book, and other rewards available for pre-order on the Kickstarter page forThe Field Study Handbook.

Article illustration, and book illustrations, by Lee Phillips (@leejohnphillips)

Join over 300,000 designers who stay up-to-date with the Core77 newsletter...

Subscribe

Test it out; it only takes a single click to unsubscribe

Stuart Constantine is a co-founder of Core77. He lives in Connecticut with his wife, three children, a collection of bikes and guitars.

Read more here:
How Understanding Human Behavior Can Open Up New Design Opportunities - Core77.com

Robots that occasionally act randomly can help groups of humans solve collective-action problems faster, new research has shown.

Playing a game with someone unpredictable can be annoying, particularly when you're on the same team. But in an online game designed to test group decision-making, adding computer-controlled players that sometimes behave randomly more than halved the time it took to solve the problem, according to the new study.

That shouldn't come as too much of a surprise, said study leader Nicholas Christakis, director of the Human Nature Lab at Yale University. Random mutations make evolution possible; random movements by animals in flocks and schools enhances group survival; and computer scientists often introduce noise a statistical term for random or meaningless information to improve search algorithms, he said. [Super-Intelligent Machines: 7 Robotic Futures]

But the discovery that these effects are mirrored in combined groups of humans and machines could have wide-ranging implications, Christakis told Live Science. To start, self-driving cars will soon share roads with human drivers, and more people may soon find themselves working alongside robots or with "smart" software.

In the study, published online today (May 17) in the journal Nature, the researchers describe how they recruited 4,000 human workers from Amazon's Mechanical Turk online crowdsourcing platform to play an online game.

Each participant was assigned at random to one of 20 locations, or "nodes," in an interconnected network. Players can select from three colors and the goal is for every node to have a different color from the neighbors they are connected to.

Players can see only their neighbors' colors, which means that while the problem may seem to have been solved from their perspective, the entire game may still be unsolved.

While highly simplified, this game mimics a number of real-world problems, such as climate change or coordinating between different departments of a company, Christakis said, where from a local perspective, a solution has been reached but globally it has not.

In some games, the researchers introduced software bots instead of human players that simply seek to minimize color conflicts with neighbors. Some of these bots were then programmed to be "noisy," with some having a 10 percent chance of making a random color choice and others a 30 percent chance.

The researchers also experimented with putting these bots in different areas of the network. Sometimes they were placed in central locations that have more connections to other players, and other times they were just placed at random or on the periphery where there are fewer links.

What the researchers found was that games in which bots exhibiting 10 percent noise were placed in the center of the network were typically solved 55.6 percent times faster than sessions involving just humans.

"[The bots] got the humans to change how they interacted with other humans," Christakis said. "They created these kinds of positive ripple effects to more distant parts of the network. So the bots in a way served a kind of teaching function." [The 6 Strangest Robots Ever Created]

There's a fine balance, though. The researchers found that the bots that had a 30 percent change of making a random color choice introduced too much noise and increased the number of conflicts in the group-decision-making process. Similarly, bots that exhibited no randomness actually reduced the randomness of human players, resulting in more of them becoming stuck in unresolvable conflicts, the scientists said.

Iain Couzin, director of the Max Planck Institute for Ornithology in Germany and an expert in collective behavior, said the study's findings mimic what he has seen in animals, where uninformed individuals can actually improve collective decision-making.

He said it is a very important first step toward a scientific understanding of how similar processes impact human behavior, particularly in the context of interactions between humans and machines.

"Already we are making our decisions in the context of algorithms and that's only going to expand as technology advances," he told Live Science. "We have to be prepared for that and understand these types of processes. And we almost have a moral obligation to improve our collective decision-making in terms of climate change and other decisions we need to make at a collective level for humanity."

The new research also points to an alternative paradigm for the widespread introduction of artificial intelligence into society, Christakis said. "Dumb AI" (bots that follow simple rules compared to sophisticated AI) could act as a catalyst rather than a replacement for humans in various kinds of cooperative networks, ranging from the so-called sharing economy (which encompasses services like ride-sharing, home-lending and coworking) to citizen science.

"We're not trying to build AlphaGo or [IBM's] Watson to replace a person we are trying to build technology that helps supplement groups of people, and in a way, I think that might be a little less frightening," Christakis said. "The bots don't need to be very smart because they're interacting with smart humans. They don't need to be able to do stuff by themselves; they just need to help the humans help themselves," he added.

Original article on Live Science.

See the rest here:
How Robots Acting Randomly Can Help Speed Human Problem-Solving - Live Science