Lee Sedol vs AlphaGo, Game 4
AlphaGo (Black) vs Lee Sedol (White)
Google DeepMind Challenge Match, Four Seasons Hotel, Seoul · 13 March 2016
In March 2016 a computer program played the Korean professional Lee Sedol over five games. AlphaGo won the first three. In the fourth, Lee was behind for much of the game when he found move 78, a wedge in the centre that the commentators had not considered. The program misjudged its answer, and Lee went on to win. It was the only game he won in the match, and it is remembered as a human high point in Go's encounter with machines.
Moments to look for
List the stones on the board
Keyboard: with the stepper focused, the left and right arrow keys move one step.
Result: Lee Sedol (White) won by resignation. White received 7.5 points for moving second.
Where the records differ:
- The match was played under Chinese rules with 7½ points of komi for White.
Sources checked (each has the same 180 moves):
- Andries Brouwer's professional game archive (CWI), AlphaGo/LeeSedol/4.sgf
- Go Game Guru live record, 13 March 2016 (Internet Archive copy)
Every move is replayed on our own Go engine, so the board is always a legal position. The introduction and notes are our own words.