Human Delays Arrival of Complete Computer Dominance for Just a Bit Longer

Go champion Lee Sedol's first victory proves Google's DeepMind AI is still learning.

|
Mar 13 2016, 3:09pm

Image: DeepMind

Take heart, humans: We're not completely obsolete yet. Earlier today, 18-time world Go champion Lee Sedol won a game against Google's AlphaGo AI program at the DeepMind Challenge Match in Seoul after enduring three straight defeats. There's only one game left in the five-match challenge, which means artificial intelligence has technically won, but Lee's victory seems to show that humans can still bring something to the table that programmed intelligences cannot.

At least for now. Tweets from Demis Hassabis, the founder of Google's DeepMind project, which is responsible for creating AlphaGo, revealed that AlphaGo made a misstep on move 79 and didn't realize it until move 87. Hassabis heartily congratulated Lee, but also emphasized in a statement to The Verge that he intends to use the defeat as a learning experience for improving AlphaGo, which continuously alters its strategies as a match progresses.

"Lee Sedol is an incredible player and he was too strong for AlphaGo today," said Hassabis. "For us this loss is very valuable. We're not sure what happened yet."

It's been an incredible few days for champions of artificial intelligence, particularly since Lee had predicted before the tournament that he'd either sweep all five matches or win all of them save one. That all changed, though, after Lee lost the first round. And then the second. And then the third. Reporting on location, The Vergequoted Lee as saying AlphaGo's playstyle "surprised" him and that he never thought the AI "would play the game in such a perfect manner."

These are significant words coming from Lee, who holds a 9-dan rank by the standards of the ancient Chinese board game, which boasts a complexity that poses far greater challenges for AI programs than chess. Lee's rank proves he has a considerable edge over Fan Hui, the 2-dan European player AlphaGo beat with a 5-0 sweep last October, and commentators widely expected that AlphaGo wasn't quite ready for an opponent as accomplished as Lee.

Lee said in the post-match press conference that securing a win after enduring three losses makes his recent victory even more valuable. It seems those in attendance thought so as well, as the applause for Lee was "rapturous."

"I've never been congratulated so much just because I won one game," Lee said. Lee added that he intends to try to win with the black Go pieces for the final match, which he says is more "valuable" than winning with the white pieces.

Lee was originally competing for a $1 million cash prize, but now that AlphaGo has effectively won, that money will now go to charity.