South Korean 'Go' grandmaster Lee Se-dol has won his first victory in a match against Google DeepMind's Artificial Intelligence supercomputer, AlphaGo.

Lee, one of the world's greatest players of the ancient Chinese boardgame, said that his win over the AI was 'invaluable,' reports BBC. This is his first win after AlphaGo won the first three games.

Like Us on Facebook

"I've never been congratulated so much because I've won one game," said the human champion.

The Grandmaster's Observations

In a press conference after the game, Lee revealed that he had found two weaknesses that the London-made AI possessed, reports CRIEnglish.

First, he said that AlphaGo appeared to have difficulty in playing when it grabs black stones. Second, he noted that the AI had difficulty in responding when Lee makes an unexpected move.

During the middle of the game, Lee made some attacks that had AlphaGo making questionable moves. Commentators viewing the match said that the AI might have made mistakes and algorithm errors.

At about three hours into the game, Lee was able to turn the tides and dominate the match, winning in the end.

Developers' Reactions

While people congratulated the South Korean player, Google developers also considered the defeat a win for them, saying that it is a 'very valuable' indicator of a problem that they will fix.

AlphaGo played very well during the first games in the best-of-five tournament. During the first match, it only won against the human champion by a narrow margin.

After winning in the second match, AlphaGo's play was considered a 'nearly perfect game' by its opponent.

In the third game, Google's AI had won Lee's 'top game.” This defeat is surprising, but is still something that Google DeepMind wants to use to further develop the AlphaGo program.

"It learns from its mistakes," Demis Hassabis, DeepMind CEO, said,

TagsLee Se-dol, AlphaGo, Go, Chinese boardgame, Google, DeepMind, Artificial Intelligence