DeepMind's AlphaGo AI has defeated Ke Jie in the first round of a best-of-three Go match in China. A video of the match is embedded below. Ke Jie was defeated by just a half a point—the closest margin possible—but scoring versus AlphaGo is a little bit disingenuous: DeepMind's AI doesn't try to win by a large margin; it just plots the surest route to victory, even if it's only by half a point.
Ke Jie is generally considered to be the world's best human Go player, but he wasn't expected to win; AlphaGo defeated the Chinese 19-year-old earlier in the year during an unbeaten online 60-match victory streak.
Today's real-life match was a little different, though. According to DeepMind cofounder Demis Hassabis, Ke Jie "used the ideas AlphaGo used in the online games in January"; in other words, Ke Jie tried to use AlphaGo's own moves against itself. Clearly it didn't quite work out, but "some wonderful moves were played," says Hassabis.
The next game will be on Thursday, and the finale on Saturday. On Friday there'll be a number of exhibition matches, including AlphaGo versus the combined mental muscle of five Chinese pros. Check out the event website for full details of all the matches being played and their timings.
reader comments
99