Excerpted and adapted from a report in The New York Times
“Last year, (AlphaGo) was still quite humanlike when it played,” said Ke Jie 9P after the first match against the go-playing AI Tuesday. “But this year, it became like a god of Go.”
“AlphaGo is improving too fast,” Ke said in a news conference after the game. “AlphaGo is like a different player this year compared to last 
year.”
Mr. Ke, who smiled and shook his head as AlphaGo finished out the game, said afterward that his was a “bitter smile.” After he finishes this week’s match, he said, he would focus more on playing against human opponents, noting that the gap between humans and computers was becoming too great. He would treat the software more as a teacher, he said, to get inspiration and new ideas about moves.
Chinese officials perhaps unwittingly demonstrated their conflicted feelings at the victory by software backed by a company from the United States, as they cut off live streams of the contest within the mainland even as the official news media promoted the promise of artificial intelligence.
Excerpted from Wired
This week’s match is AlphaGo’s first public appearance with its new architecture, which allows the machine to learn the game almost entirely from play against itself, relying less on data generated by humans. In theory, this means DeepMind’s technology can more easily learn any task.
Underpinned by machine learning techniques that are already reinventing everything from internet services to healthcare to robotics, AlphaGo is a proxy for the future of artificial intelligence.
This was underlined as the first game began and (DeepMind CEO Demis) Hassabis (in photo) revealed that AlphaGo’s new architecture was better suited to tasks outside the world of games. Among other things, he said, the system could help accelerate the progress of scientific research and significantly improve the efficiency of national power grids.
DeepMind Match 1 wrap up
“There was a cut that quite shocked me,” said Ke Jie, “because it was a move that would never happen in a human-to-human Go match. But, afterwards I analyzed the move and I found that it was very good. It is one move with two or even more purposes. We call it one stone, two birds.”
“Ke Jie started with moves that he had learned from the Master series of games earlier this year, adding those new moves to his repertoire,” said Michael Redmond 9P. “Ke Jie used the lower board invasion point similar to AlphaGo in the Masters games, and this was a move that was unheard of before then. Although this was one of the most difficult moves for us to understand, in the last month or players have been making their own translations and interpretations of it.”
“Every move AlphaGo plays is surprising and is out of our imagination,” said Stephanie Yin 1P. “Those moves completely overthrow the basic knowledge of Go. AlphaGo is now a teacher for all of us.”
photos: (top) courtesy China Stringer Network, via Reuters (middle) Noah Sheldon/Wired (bottom) DeepMind
New version of AlphaGo self-trained and much more efficient
Wednesday May 24, 2017
by Andy Okun, reporting from the ‘Future of Go’ summit in Wuzhen, China
play games of previous versions of AlphaGo, a Google DeepMind engineer told an audience in China. David Silver (at right), lead researcher on the AlphaGo project, told the Future of AI Forum in Wuzhen that because AlphaGo had become so strong, its own games constituted the best available data to use.
In addition, Silver revealed that DeepMind had measured the handicap needed between different versions of the software. AlphaGo Fan could give four stones to the previous best software, such as Zen or CrazyStone, which had reached 6d in strength. AlphaGo Lee, in turn, could give AlphaGo Fan three stones, and AlphaGo Master, which at the new year achieved a 60-game undefeated streak against top pros before coming to this challenge, is three stones stronger than AlphaGo Lee. Silver delivered this with the caveat that these handicap stones are not necessarily directly convertible to human handicaps. Professional players suggested that this may be due to AlphaGo’s tendency to play slowly when ahead — i.e., an AlphaGo receiving a three stone handicap may give its opponent ample opportunities to catch up, just as yesterday’s AlphaGo let Ke Jie get to a 0.5 point margin. This also reveals that AlphaGo is able to play with a handicap, previously a matter of speculation in the go community.
The version of AlphaGo that defeated Ke Jie 9p in the first round of the three game challenge match yesterday was trained entirely on the self-
The version of AlphaGo that beat Fan Hui 2p in 2015 (AlphaGo Fan) and the one that defeated Lee Sedol 9p last year in Seoul (AlphaGo Lee) each included a “value network,” designed to evaluate a position and give the probability of winning, and a “policy network,” designed to suggest the best next move, that were trained using hundreds of thousands of skilled human games. The most recent version, AlphaGo Master, trained both networks on a database of its self-play games generated by its predecessors.
This was not the only new information Silver revealed about system. The version playing Ke Jie is so much more efficient that it uses one tenth the quantity of computation that Alphago Lee used, and runs on a single machine on Google’s cloud, powered by one tensor processing unit (TPU). AlphaGo Lee would probe 50 moves deep and study 100,000 moves per second. While that sounds like a lot, by comparison, the tree search powering the Deep Blue chess system that defeated Gary Kasparov in the 1990s looked at 100 million moves per second.
“AlphaGo is actually thinking much more smartly than Deep Blue,” Silver said.
Silver’s talk came after DeepMind chief Demis Hassabis gave a passionate account of how go and AI research have fed each other. Go is so combinatorially large that playing it well is intuitive as well as a matter of calculation. The methods that have worked so well with AlphaGO have generated moves and strategies that seem high level, intuitive, even creative. These same methods have applications in medicine, energy and many other areas. He quoted Kasparov: “Deep Blue was the end. AlphaGo is the beginning.”
photos by Dan Maas