- Alphabet's AlphaGo AI program has beaten the world's top Go player 4-1.
- AlphaGo is able to discover new optimal strategies by analyzing games and playing against itself.
- The technologies at the heart of AlphaGo are the future and will likely revolutionize many important sectors.
In February Amigobulls reported that Alphabet Inc-A (NASDAQ:GOOGL) claimed a major Artificial Intelligence (AI) breakthrough with a software program, dubbed AlphaGo, which taught itself to beat a top human player of the board game Go.
Go is much more complex than chess and, until recently, no machine had beaten a human Go champion. In fact, experts predicted it would be at least another 10 years until a computer could beat one of the world’s elite group of Go professionals. But AlphaGo achieved this milestone AI challenge much sooner, beating European Go champion Fan Hui in October 2015. The announcement was delayed until the publication of a research article titled "Mastering the game of Go with deep neural networks and tree search" in Nature.
AlphaGo combines Monte-Carlo tree search with deep neural networks that have been trained by supervised learning, from human expert games, and further improves its performance by learning from games played against itself.
AlphaGo was developed by Google DeepMind, a British AI company founded in 2010 as DeepMind Technologies and acquired by Google in 2014. Google DeepMind has created neural networks that learn how to play games in a similar fashion to humans and appears to mimic key cognitive aspects of the human brain.
Google announced that "AlphaGo's next challenge will be to play the top Go player in the world over the last decade, Lee Sedol." The match eventually took place at the Four Seasons Hotel in Seoul, South Korea, between 9 and 15 March, has been video-streamed live and watched by tens of thousands of viewers worldwide. The result: AlphaGo won four games to one.
Go and AI experts have started analyzing the match. In particular, AlphaGo's move 37 in the second game, which is attracting a lot of attention, is considered beautiful, effective - and inhuman. "It’s not a human move. I’ve never seen a human play this move,” said Fan Hui. The move was initially questioned but then acclaimed by experts. Based on its knowledge base derived from both games played by human experts and games played against itself, AlphaGo "knew" that a human wouldn't play that particular move, but also that the move is ultimately effective.
"While games are the perfect platform for developing and testing AI algorithms quickly and efficiently, ultimately we want to apply these techniques to important real-world problems," said AI researcher and co-founder of DeepMind Demis Hassabis in January." Because the methods we’ve used are general-purpose, our hope is that one day they could be extended to help us address some of society’s toughest and most pressing problems, from climate modeling to complex disease analysis. We’re excited to see what we can use this technology to tackle next!"
The validation of DeepMind's neural network and machine learning techniques will open the way to tackling more complex AI challenges. After AlphaGo's victory, the focus will turn to real-world targets for AI research.
"The victory is notable because the technologies at the heart of AlphaGo are the future," notes Wired. "They’re already changing Google and Facebook and Microsoft and Twitter, and they’re poised to reinvent everything from robotics to scientific research." The Wall Street Journal mentions important applications to health care as well.
In February, the company announced the launch of DeepMind Health, a division within DeepMind that collaborates directly with frontline clinicians to develop technology that helps improve patient care. While AI is not part of the first pilot projects in collaboration with the UK National Health Service, it could be used in future projects. "It’s certainly something we are excited about for the future," notes the DeepMind Health site.
In fact, health care is emerging as a first promising application for next-generation AI technology. Earlier this month, IBM (NYSE:IBM) and the New York Genome Center announced a partnership to accelerate cancer research and scale access to precision medicine using cognitive insights from IBM's AI system Watson, an open "cognitive computing" technology platform, where "cognitive computing" indicates computer systems that understand the world in the way that humans do.
Like AlphaGo, Watson continuously learns and adds to its knowledge base, gaining in value and knowledge over time. A difference is that Watson has already demonstrated more general-purpose abilities than AlphaGo, which at this moment only knows how to play Go, but Hassabis' remarks above indicate that Alphabet is working at widening the application scope of its AI research.
Another competitor in the race to AI is Facebook, which is becoming one of the most advanced AI research centers in the world, with an ambitious research program aimed at revolutionizing human-computer interaction. Head of Facebook AI research Yann LeCun wants to build the best AI research lab in the world, and machines that think and learn. Gizmodo reports that Facebook is also working on a Go software, considered as an ideal AI testbed, and quotes LeCun's positive and not-so-positive reactions, which show how the race to AI is heating up.
Alphabet's breakthrough indicates that AI systems are becoming adept at more and more high-level cognitive tasks that used to be considered as too complex for automation. The surprising Move 37 indicates that next-generation AI systems will be able to devise their own strategies to optimally perform their tasks, independently of human input. It seems likely that AI will play a more and more important role in computing, revolutionize many important sectors like robotics, health care, drug discovery, user interface design, and social networking, and boost the stocks of the visionary companies involved in early development. This could be a big boost to Alphabet's efforts to dominate the AI space.