Computing | The Knowledge Dynasty

Computing

‘It’s able to create knowledge itself’: Google unveils AI that learns on its own

In a major breakthrough for artificial intelligence, AlphaGo Zero took just three days to master the ancient Chinese board game of Go … with no human help.

Chinese Go board

 

Google’s artificial intelligence group, DeepMind, has unveiled the latest incarnation of its Go-playing program, AlphaGo an AI so powerful that it derived thousands of years of human knowledge of the game before inventing better moves of its own, all in the space of three days.

Named AlphaGo Zero, the AI program has been hailed as a major advance because it mastered the ancient Chinese board game from scratch, and with no human help beyond being told the rules. In games against the 2015 version, which famously beat Lee Sedol, the South Korean grandmaster, in the following year, AlphaGo Zero won 100 to 0.

The feat marks a milestone on the road to general-purpose AIs that can do more than thrash humans at board games. Because AlphaGo Zero learns on its own from a blank slate, its talents can now be turned to a host of real-world problems.

At DeepMind, which is based in London, AlphaGo Zero is working out how proteins fold, a massive scientific challenge that could give drug discovery a sorely needed shot in the arm.

AlphaGo vs Lee Sedol

Match 3 of AlphaGo vs Lee Sedol in March 2016. Photograph: Erikbenson

“For us, AlphaGo wasn’t just about winning the game of Go,” said Demis Hassabis, CEO of DeepMind and a researcher on the team. “It was also a big step for us towards building these general-purpose algorithms. Most AIs are described as narrow because they perform only a single task, such as translating languages or recognising faces, but general-purpose AIs could potentially outperform humans at many different tasks.” In the next decade, Hassabis believes that AlphaGo’s descendants will work alongside humans as scientific and medical experts.

Previous versions of AlphaGo learned their moves by training on thousands of games played by strong human amateurs and professionals. AlphaGo Zero had no such help. Instead, it learned purely by playing itself millions of times over. It began by placing stones on the Go board at random but swiftly improved as it discovered winning strategies.

David Silver describes how the Go playing AI program, AlphaGo Zero, discovers new knowledge from scratch. Credit: DeepMind

“It’s more powerful than previous approaches because by not using human data, or human expertise in any fashion, we’ve removed the constraints of human knowledge and it is able to create knowledge itself,” said David Silver, AlphaGo’s lead researcher.

The program amasses its skill through a procedure called reinforcement learning. It is the same method by which balance on the one hand, and scuffed knees on the other, help humans master the art of bike riding. When AlphaGo Zero plays a good move, it is more likely to be rewarded with a win. When it makes a bad move, it edges closer to a loss.

Demis Hassabis
Demis Hassabis, CEO of DeepMind: “For us, AlphaGo wasn’t just about winning the game of Go.” Photograph: DeepMind/Nature

At the heart of the program is a group of software neurons that are connected together to form an artificial neural network. For each turn of the game, the network looks at the positions of the pieces on the Go board and calculates which moves might be made next and probability of them leading to a win. After each game, it updates its neural network, making it stronger player for the next bout. Though far better than previous versions, AlphaGo Zero is a simpler program and mastered the game faster despite training on less data and running on a smaller computer. “Given more time, it could have learned the rules for itself too, ” Silver said.

Q&A

What is AI?

Artificial Intelligence has various definitions, but in general it means a program that uses data to build a model of some aspect of the world. This model is then used to make informed decisions and predictions about future events. The technology is used widely, to provide speech and face recognition, language translation, and personal recommendations on music, film and shopping sites. In the future, it could deliver driverless cars, smart personal assistants, and intelligent energy grids. AI has the potential to make organisations more effective and efficient, but the technology raises serious issues of ethics, governance, privacy and law.

Writing in the journal Nature, the researchers describe how AlphaGo Zero started off terribly, progressed to the level of a naive amateur, and ultimately deployed highly strategic moves used by grandmasters, all in a matter of days. It discovered one common play, called a joseki, in the first 10 hours. Other moves, with names such as small avalanche and knights move pincer soon followed. After three days, the program had discovered brand new moves that human experts are now studying. Intriguingly, the program grasped some advanced moves long before it discovered simpler ones, such as a pattern called a ladder that human Go players tend to grasp early on.

AlphaGo Zero starts with no knowledge, but progressively gets stronger and stronger as it learns the game of Go. Credit: DeepMind

“It discovers some best plays, josekis, and then it goes beyond those plays and finds something even better,” said Hassabis. “You can see it rediscovering thousands of years of human knowledge.”

Eleni Vasilaki, professor of computational neuroscience at Sheffield University, said it was an impressive feat. “This may very well imply that by not involving a human expert in its training, AlphaGo discovers better moves that surpass human intelligence on this specific game,” she said. But she pointed out that, while computers are beating humans at games that involve complex calculations and precision, they are far from even matching humans at other tasks. “AI fails in tasks that are surprisingly easy for humans,” she said. Just look at the performance of a humanoid robot in everyday tasks such as walking, running and kicking a ball.

Tom Mitchell, a computer scientist at Carnegie Mellon University in Pittsburgh called AlphaGo Zero an outstanding engineering accomplishment. He added: “It closes the book on whether humans are ever going to catch up with computers at Go. I guess the answer is no. But it opens a new book, which is where computers teach humans how to play Go better than they used to.”

David Silver describes how the AI program AlphaGo Zero learns to play Go. Credit: DeepMind

The idea was welcomed by Andy Okun, president of the American Go Association: “I don’t know if morale will suffer from computers being strong, but it actually may be kind of fun to explore the game with neural-network software, since it’s not winning by out-reading us, but by seeing patterns and shapes more deeply.”

While AlphaGo Zero is a step towards a general-purpose AI, it can only work on problems that can be perfectly simulated in a computer, making tasks such as driving a car out of the question. “AIs that match humans at a huge range of tasks are still a long way off,” Hassabis said. More realistic in the next decade is the use of AI to help humans discover new drugs and materials, and crack mysteries in particle physics. “I hope that these kinds of algorithms and future versions of AlphaGo-inspired things will be routinely working with us as scientific experts and medical experts on advancing the frontier of science and medicine,” Hassabis said.

Read more: https://www.theguardian.com/science/2017/oct/18/its-able-to-create-knowledge-itself-google-unveils-ai-learns-all-on-its-own

Amazon to open visually focused AI research hub in Germany

Ecommerce giant Amazon has announced a new research center in Germany focused on developing AI to improve the customer experience — especially in visual systems.

Amazon said research conducted at the hub will also aim to benefit users of Amazon Web Services and its voice driven AI assistant tech, Alexa. 

The center will be based in Tübingen, near the Max Planck Institute for Intelligent Systems‘ campus, and will be staffed with more than 100 machine learning engineers.

The new 100+ “highly qualified” jobs will be created over the next five years, it said today. The site is Amazon’s fourth Research Center in Germany — after Berlin, Dresden and Aachen

For the Tübingen hub, the company is collaborating with the Max Planck Society on an earlier regional research collaboration that kicked off in December 2016 and is also focused on AI, as well as on bolstering a local startup ecosystem.

Robotics, machine learning and machine vision are key areas of focus for the so-called ‘Cyber Valley’ initiative. Existing partner companies in that effort include BMW, Bosch, Daimler, IAV, Porsche and ZF Friedrichshafen — and now Amazon.

As with other research partners, Amazon will be contributing €1.25 million to set up research groups in the Stuttgart and Tübingen regions, the Society said today.

“We appreciate Amazon’s commitment in the Cyber Valley and to research on artificial intelligence,” said Max Planck president Martin Stratmann in a statement. “We gain another strong cooperation partner who will further increase the international significance of research in the area of machine learning and computer vision in the Stuttgart and Tübingen region.”

“With our Amazon Research center in Tübingen, we will become part of one of the largest research initiatives in Europe in the area of artificial intelligence. This underlines our commitment to create high-skilled jobs in breakthrough technologies,” added Ralf Herbrich, director of machine learning at Amazon and MD of the Amazon Development Center Germany, in another supporting statement.

Earlier this month TechCrunch broke the news that Amazon had acquired 3D body model startup, Body Labs, whose scientific advisor and co-founder — Dr Michael J Black — is a director at the Max Planck Institute for Intelligent Systems’ Department of Perceptive Systems.

The Institute generally describes its goal being “to understand the principles of perception, learning and action in autonomous systems that successfully interact with complex environments and to use this understanding to design future systems”.

Amazon said today that Dr Black will support its the new research hub as an Amazon Scholar, along with another Max Planck director, Dr Bernhard Schölkopf, who is based in the Department of Empirical Inference.

Both will also continue to manage their respective departments at the Institute, it added.

Schölkopf is a leading expert in machine learning in Europe and co-inventor of computer-aided photography. He has also developed pioneering technologies through which computer causality can be learned. With causality, AI systems predict customer behavior in response to automated decisions, such as the order of the search results, to optimize the search experience,” said Amazon. “Black is a leading expert in the field of machine vision and co-founder of the Body Labs company, which markets AI body procedures for capturing human body movements and shapes from 3D images for use in various industries.”

As we suggested at the time, Amazon’s purchase of the 3D body model startup looks primarily like a talent-based acquihire — to bring Black’s visual systems’ expertise into the fold.

Although the Max Planck Institute also manages and licenses thousands of patents — so smoother access, via Black’s connections, to key technologies for licensing purposes may also be part of its thinking as it spends a few euros to forge closer ties with the German research network.

Investing in business critical research and the next generation of AI researchers is also clearly on the slate here for Amazon: As part of the collaboration it says it will be providing the Society with research awards worth €420,000 per year.

A spokesperson confirmed this funding will be provided for five years, although it’s not clear exactly how many PhD candidates and Post-Doc research students will get funded from out of Amazon’s pot of money each year.

The Society said it will use the funding to finance the research activities of doctoral and postdoctoral students at the Max Planck Institute for Intelligent Systems.

“The support from Amazon and the other Cyber Valley partners enables us to further improve the training of highly qualified junior researchers in the field of artificial intelligence,” said Schölkopf in a statement. “This will help to ensure that we continue to provide both science and industry with creative minds to consolidate our pioneering position in intelligent systems.”

Computer vision has become a hugely important AI research area over the past decade — yielding powerful visual systems that can, for example, quickly and accurately detect and recognize objects, individual faces and body postures, which in turn can be used to feed and enhance the utility and intelligence of AI assistant systems.

And while CV research has already been fairly widely commercially applied by tech giants, there’s plenty of challenges remaining and academics continue to work on enhancing and expanding the power of visual AI systems — with tech giants like Amazon in close pursuit of any gains.

The basic rule of thumb is: The bigger the platforms, the bigger the potential rewards if smarter visual systems can shave operating costs and user friction from products and services at scale.

The Tübingen R&D hub is Amazon’s first German center focused on visual AI research. Though it’s just the latest extension of already extensive Amazon R&D efforts on this front (a quick LinkedIn job search currently lists ~470 Amazon jobs involving computer vision in various locations worldwide).

Amazon’s Berlin research hub started as a customer service center but since 2013 has also included dev work for the cloud business of Amazon Web Services (including hypervisors, operating systems, management tools and self-learning technologies).

While its Dresden hub houses the kernel and OS team that works on the core of EC2, the actual virtual compute instance definitions and Amazon Linux, the operating system for its cloud.

In Aachen its R&D hub houses engineers working on Alexa and architecting cloud AWS services.

Read more: https://techcrunch.com/2017/10/23/amazon-to-open-visually-focused-ai-research-hub-in-germany/

Should we ban sex robots while we have the chance? | Jenny Kleeman

“AI sex dolls are on their way, with potentially sinister social consequences. So before they hit the market, we must ask whether they should,” writes robotics expert Jenny Kleeman.

sex robots

 

People are blowing a fuse about sex robots or rather, rape robots. Journalists from the New Statesman and the New York Times among others have all reported on the sex robot Roxxxy TrueCompanions controversial Frigid Farrah setting: a mode in which she has been programmed to resist sexual advances and which will allow men to act out rape fantasies.

Women’s rights activists have lined up to condemn Roxxxy. Everyday Sexism’s Laura Bates describes her as the sex robot that’s yours to rape for just $9,995. Writing in the Times on Thursday, the barrister Kate Parker called for sex robots like Roxxxy to be criminalised. “The sophistication of the technology behind Roxxxy marks a step forward for robotics. For human society, it’s an unquestionable regression,” she says.

Rise of the sex robots
Theres a problem with this story: the robot doesnt exist. Douglas Hines, the man behind Roxxxy TrueCompanion, has been drumming up publicity for his creation ever since he unveiled her to the public at the 2010 AVN Adult Entertainment Expo in Las Vegas. Even though his website pulsates with throbbing Order Her Now! buttons, no journalist has seen or photographed Roxxxy since 2010, and no one in the surprisingly extensive robot enthusiast community has ever reported owning one.

I tried to meet Hines in person many times over the past year while researching a documentary and article on sex robots, and although he was happy to talk over the phone he avoided meeting me when I asked to see Roxxxy in the flesh. Roxxxy, much like the replicants and Stepford Wives of science fiction, seems to be nothing more than fantasy.

But while Roxxxy may not be available to buy, models like her will be very soon. Abyss Creations are due to ship the first talking, animatronic, AI-enabled heads for their hyper-realistic silicone sex dolls by the end of the year. And while the sex robots on offer from China and Japan may currently have more in common with push-button talking baby dolls than Ava from Ex Machina, theres commercial pressure to get sophisticated models with AI on sale as soon as possible.

The sex tech industry is worth $30bn a year, and with two thirds of heterosexual men in a recent survey saying they could imagine buying a sex robot for themselves, the race is on to make the fantasy a reality. But before sex robots hit the market, we have the space to ask whether they should.

The issue with sex robots in general not just hypothetical ones programmed to have a resist function is how their existence will affect how human beings interact with each other. Sex robots are different from sex dolls and sex toys because they have AI. More than just a mechanism for giving you an orgasm, a sex robot is designed to be a substitute partner: a vibrator doesnt laugh at your jokes and remember your birthday, but Abyss Creations Harmony model can.

If men (and it will be men even the few male sex dolls produced by Abyss Creations every year are generally shipped to male customers) become used to having sex with synthetic companions that are programmed to meet their most precise specifications, how will they then interact with real women who have the inconvenience of having their own idiosyncrasies and free will? If you are used to having sex with ultra-life-like humanoids whenever and however you want, will you be more likely to expect complete dominance in your relationships with other humans?

Young people who have grown up in the age of online porn might consider shaved pubic hair and double penetration to be completely normal. Similarly, the generation growing up when sex robots are commonplace might see brutally selfish sex as both desirable and achievable.

Sex robots exist purely to satisfy their owners. Is any sexual relationship healthy if its only ever about one persons pleasure? Can sex with a robot ever be consensual? This isnt about robot rights its about the kind of sex that will become normal within human societies if we start having sex with robots.

Child sex dolls have been banned in the UK because of fears they will encourage the desire to abuse among paedophiles, rather than simply sate it. Parker is calling for a similar ban for all sex robots. But while we might be able to stop them being imported or manufactured here, we cant stop them being developed overseas.

Perhaps the most important question to ask is why there is a market for sex robots in the first place. Why do some people find the idea of a partner without autonomy so attractive? Until we have the answer to that, well need to prepare ourselves for the inevitable rise of the sex robots.

Jenny Kleeman is a freelance journalist

 

Read more: https://www.theguardian.com/commentisfree/2017/sep/25/ban-sex-robots-dolls-market

Subscribe to Blog via Email

Enter your email address to subscribe to this blog and receive notifications of new posts by email.

Categories

Coursera

New Skills, New You: Transform your career in 2016 with Coursera

Follow us on Twitter