AlphaGo has taken the mantle of the best Go player in the world after winning its second game against world number one Ke Jie. The best of three series was a hard fought victory for the new ‘Master’ version of the Deepmind AI, but brings an end to the question of whether a computer can play the ancient Japanese game better than any human.
According to AlphaGo’s analysis, Ke Jie played a perfect game for the first 50 moves. Forcing the computer to branch out its strategy and look for new ways around it. In the end, it was the computer managed to overcome Ke Jie’s aggressive strategy and force the human player to retire from the match after 156 turns.
It was the closest match between man and machine yet, at least where Go is concerned. Despite the difference in final score being much higher than the first game. Ke Jie even believed that he could win the match somewhere in the middle of the game.
To be fair, this new version of AlphaGo is far more powerful and streamlined than the one that beat Lee Sedol last year. The engineers at Google’s Deepmind project call this one ‘Master’, because it was designed to be unbeatable at Go. An anonymous trial run over the internet allowed the AI to go on a 60 game winning streak.
Ke Jie may have lost the series against AlphaGo, but he still has one last chance to defeat the AI in the third match. If anything, he is playing for the pride of the human race at this point. AlphaGo will later go on to play two more exhibition matches where it will face off against five expert Go players working as a team, and then a doubles mirror match where it will be partnered with a human player.
[Source: Future Of Go Summit]