The AlphaGo versus Ke Jie match was a three-game Go competition held as part of the Future of Go Summit in Wuzhen, China, on 23, 25, and 27 May 2017. The match pitted the computer program AlphaGo Master, developed by Google DeepMind, against Ke Jie, who was then the world's number one ranked Go player. AlphaGo won all three games, marking a significant milestone in Artificial intelligence and Machine learning research. Following the match, AlphaGo was retired by DeepMind, and the program was subsequently awarded an honorary professional 9-dan title by the Chinese Weiqi Association.
Background
At the time of the match, Ke Jie held the top ranking among human players under multiple systems, including the Rémi Coulom ranking and those of the Korea Baduk Association, Japan Go Association, and Chinese Weiqi Association, a position he had maintained since late 2014. The version of AlphaGo used in this match, AlphaGo Master, had previously won 60 consecutive online games against top professionals in late 2016 and early 2017, including a 3-0 victory over Ke Jie during that streak. AlphaGo Master ran on a single machine with four TPUs and had an estimated Elo rating of 4,858. DeepMind claimed this version was three stones stronger than the version that had defeated Lee Sedol in 2016. Notably, AlphaGo Master was not DeepMind's strongest version at the time; the company had already developed AlphaGo Zero, a more powerful iteration, but did not reveal its existence until October 2017, when a paper on it was published in Nature.
Games
Game 1
The first game took place on 23 May 2017. AlphaGo, playing white, won by a margin of 0.5 points, a narrow victory that highlighted the program's precision.
Game 2
The second game was played on 25 May 2017. About one hour into the game, Demis Hassabis, co-founder of DeepMind, tweeted that AlphaGo's evaluations indicated Ke Jie was playing perfectly. However, Ke Jie later lost ground on the lower part of the board. Approximately four hours into the game, AlphaGo simplified the position, making it clear that Ke Jie was losing. AlphaGo, playing black, won by resignation after move 155.
Game 3
The third game occurred on 27 May 2017. Ke Jie, playing white, resigned, giving AlphaGo a 3-0 series victory. At the time of resignation, AlphaGo had about an hour and a half of its time remaining, while Ke Jie had roughly 32 minutes left on the clock.
Prize and Aftermath
Google DeepMind offered a winner's prize of $1.5 million for the match, with the losing side receiving $300,000 for participating in the three games. After the match, AlphaGo was retired, and DeepMind shifted its focus to other areas of Artificial intelligence research. The Chinese Weiqi Association later awarded AlphaGo a professional 9-dan title, recognizing its exceptional performance.
Coverage
The match was barred from being live-streamed in China, but it was covered extensively on Chinese online platforms and national television via Zhejiang TV. The games have since been analyzed by Go enthusiasts and researchers, contributing to discussions about the capabilities of neural networks and deep learning in complex strategy games.
See Also
- AlphaGo versus Lee Sedol