Google DeepMind’s AI beat Go world champion Lee Sedol for the third time on Saturday to take the Challenge Series tournament, which is being held in South Korea.
The AlphaGo AI is now 3-0 up in the five-game series so there is no way 18-time world champion Lee can come back from here. However, the remaining games on Sunday and Tuesday will still be played out. The only thing left to play for is to find out if humanity’s best Go player can win even a single match against AlphaGo.
It’s a major milestone for artificial-intelligence research: Go is a simple game but has been notoriously difficult for computers to master because of the sheer number of potential moves. Go players believe the game relies on intuition as a strategy. While AI programs began being able to beat humans at chess decades ago, the best Go players in the world have always been able to outsmart Go-playing software — until now.
Go is a two-player, turn-based strategy game. Each player puts down either black or white stones in an attempt to outmaneuver and surround the other player. It’s easy to pick up but takes years to master.
At a post-game press conference, Demis Hassabis, DeepMind cofounder and CEO, said: “This is a moment to celebrate human achievement — the genius that is Lee Sedol and the incredible work of so many people to make this scientific breakthrough. We’re stunned, and to be honest, I’m a bit speechless.”
“AlphaGo evaluates tens of thousands of possible moves per second, and what is incredible is that Lee Sedol can compete so closely with just the power of his mind. For three games, Lee pushed AlphaGo to its limit. We’re thrilled to have won the tournament — a lifelong dream of mine and a grand challenge of artificial intelligence. But we still have two games to play and many more years of research ahead to understand how we can apply these techniques beyond games and help society solve some of its biggest problems.”
Lee apologised for not being able to satisfy people’s expectations. “I kind of felt powerless,” he said. “When I look back on the three matches, even if I were to go back and redo the first match, I think I would not be able to win because I misjudged AlphaGo.”
Lee added: “At the beginning of the second game, the game flowed the way I intended but there were a number of missed opportunities. For the third match, I have extensive experience in playing the game of Go, but I never felt a case like this with so much pressure, which I was unable of overcoming. For Games Four and Five, I ask that you continue to be interested and follow what happens.”
Hassabis and fellow cofounder Shane Legg have publicly celebrated their victory on Twitter. Hassabis described it as a “historic moment.” Legg wrote: “They said this wouldn’t happen for a long time… but it just did!”
The tournament has been closely watched by the most senior people at Alphabet and Google. Alphabet president Sergey Brin attended the third game, while Alphabet chairman Eric Schmidt was there for the first.
At the press conference, Brin said: “I spent a lot of time playing Go at grad school. Larry thought Google might never happen because I spent so much time playing Go.”
“Go is a very beautiful game and I think it teaches a lot about life. Much more so than a game like chess. When you watch great Go players play, it is like a thing of beauty.”
“So I’m very excited that we have been able to instil that kind of beauty in computers.”
Google CEO Sundar Pichai was also tweeting about the the tournament earlier in the week.
DeepMind has pledged to donate the $1 million (£700,000) in prize money to UNICEF, STEM charities, and Go organisations.