At today’s fourth game in the Google DeepMind Challenge Match, the best Go player of the last decade, Lee Sedol, came back from three consecutive losses to beat the Go-playing computer program, AlphaGo. Playing as white, Lee won when AlphaGo resigned after 180 moves.
Reporters gave Lee Sedol a standing ovation as he entered the press conference. Lee commented, “I’ve never been congratulated so much just for winning a single game. I would not exchange today’s win with anything else in the world. I’d like to thank you all for your encouragement and support — the driver behind the win I achieved today.”
Demis Hassabis, CEO of Google DeepMind said, “Lee Sedol has once again showed what an incredible player he is. He was too good for AlphaGo today. AlphaGo started off strong but was pressured by Lee’s brilliant play into making some mistakes, of which he took advantage. We came to Korea to test the limits of AlphaGo, and we’ve certainly done that today. Huge congratulations to Lee Sedol, and we’re excited to see what the final game on Tuesday brings.”
David Silver, Lead Researcher on AlphaGo at Google DeepMind, explained, “AlphaGo learns for itself and improves from self-play, so it's hard to say when and where holes in its knowledge will occur. But it's possible that a strong player might be able to push AlphaGo into one of these holes, and I'd like to congratulate Lee on finding a sequence in the centre that caused AlphaGo such difficulties. This is why we came here — it’s hard to find weaknesses from internal evaluations alone — and we needed a creative genius to help us learn more.”
Michael Redmond, professional 9-dan Go player, said “Today’s game was another example of AlphaGo playing a very interesting, good game. However, move 78 by Lee Sedol was really brilliant — and enabled him to win.“
Song Taegon, 9-dan, Korean commentator, said “It seems Lee Sedol can now read AlphaGo better and has a better understanding of how AlphaGo moves. The fifth game will be a far closer battle than before. Professional Go players said that they became even more interested in Go after witnessing AlphaGo’s innovative moves. People started to rethink moves that were previously regarded as undesirable or bad moves. AlphaGo can help us think outside of the box in Go games.“
AlphaGo’s first three wins already secured overall victory in the Google DeepMind Challenge Match, but today’s loss heightens the drama going into the final game, Game Five, which will be played on Tuesday, March 15 at 1pm KST.