AlphaGo’s move 37: the play no human would make
In the second game of its March 2016 match against Lee Sedol in Seoul, DeepMind's AlphaGo played a stone that, by its makers' own estimate, a human would have chosen about once in 10,000 times. Google DeepMind says move 37 "upended centuries of traditional wisdom." AlphaGo won the match 4-1.
What was AlphaGo?
AlphaGo was a Go-playing program built by DeepMind, an AI lab that is now Google DeepMind. DeepMind co-founder Demis Hassabis introduced it on Jan. 27, 2016, in a Google blog post, the day its paper was "published in Nature."
Go is hard for computers because of its size, Hassabis explained: the number of possible positions is "more than the number of atoms in the universe, and more than a googol times larger than chess," Hassabis wrote. AlphaGo had already beaten the player Fan Hui: "In a closed-doors match last October, AlphaGo won by 5 games to 0." Next was "a five-game challenge match in Seoul against the legendary Lee Sedol."
What happened with move 37?
Lee Sedol was a "9-dan professional with 18 world titles," as Hassabis described him.
"Based on our data, AlphaGo's bold move 37 in Game 2 had a 1 in 10,000 chance of being played by a human," Hassabis wrote in his March 16, 2016, account of the match. "Commentators noted that AlphaGo played many unprecedented, creative, and even 'beautiful' moves."
Google DeepMind's AlphaGo page describes it simply: "In game two, it played Move 37 — a move that had a 1 in 10,000 chance of being used."
“Based on our data, AlphaGo's bold move 37 in Game 2 had a 1 in 10,000 chance of being played by a human.”
Did humans answer back?
Yes, once. "Lee countered with innovative moves of his own, such as his move 78 against AlphaGo in Game 4," Hassabis wrote. He said that move had the same estimated one-in-10,000 chance and that it led Lee to win that game, his only victory of the match. "The final score was 4-1."
How many people watched?
"AlphaGo's 4-1 victory in Seoul, South Korea, in March 2016 was watched by over 200 million people worldwide," according to Google DeepMind. AlphaGo received a 9-dan professional ranking, which DeepMind calls "the first time a computer Go player had received the highest possible certification."
Why move 37 belongs in the museum
Hassabis framed the match as more than a game: "Our public test of AlphaGo, however, was about more than winning at Go." The $1 million in prize money went to "organizations that support science, technology, engineering and math (STEM) education and Go, as well as UNICEF," he wrote.
Move 37 is the moment of the match that DeepMind itself singles out: a play its own data said a human would almost never make, in a match where commentators called AlphaGo's moves creative.
FAQ
What was move 37?
A move AlphaGo played in Game 2 against Lee Sedol in March 2016. DeepMind estimated it had a 1 in 10,000 chance of being played by a human.
Who won AlphaGo vs. Lee Sedol?
AlphaGo won the five-game match in Seoul in March 2016 by 4-1. Lee Sedol won Game 4.
What was Lee Sedol's move 78?
A move Lee played in Game 4 that DeepMind also rated as a 1 in 10,000 choice. It led to his only win of the match.
Sources
Cite this
AlphaGo’s move 37: the play no human would make, New Mind Museum, November 6, 2026. https://newmindmuseum.com/history/2026/11/alphago-move-37-lee-sedol-2016/
New Mind Museum. (2026, November 6). AlphaGo’s move 37: the play no human would make. https://newmindmuseum.com/history/2026/11/alphago-move-37-lee-sedol-2016/
“AlphaGo’s move 37: the play no human would make.” New Mind Museum, 6 Nov. 2026, https://newmindmuseum.com/history/2026/11/alphago-move-37-lee-sedol-2016/.
@misc{new-mind-museum-2026-alphagos-move-37-the-play-no-human-would,
author = {{New Mind Museum}},
title = {AlphaGo’s move 37: the play no human would make},
year = {2026},
url = {https://newmindmuseum.com/history/2026/11/alphago-move-37-lee-sedol-2016/}
}Free to quote and cite with a link to this page.