Room 4 · The Awakening · 2016

AlphaGo Defeats Lee Sedol

AlphaGo Defeats Lee Sedol (DeepMind team, 2016): A machine masters a game thought to be out of its reach.

In March 2016, in Seoul, AlphaGo beat Lee Sedol, one of the best Go players in the world, 4 to 1. In the second game it played the famous move 37, a move so unexpected that commentators thought it was a mistake.

Go is played on a 19 by 19 board, and the number of possible positions is far larger than the number of atoms in the observable universe. After Deep Blue beat Garry Kasparov at chess in 1997, many experts thought a program that could beat top Go professionals was still a decade or more away.

DeepMind's AlphaGo combined two neural networks with a search method. A policy network suggested promising moves, and a value network estimated who was winning from a given position. Both guided a Monte Carlo tree search that explored possible continuations. The networks first learned from millions of moves by strong human players, and then improved by playing against versions of themselves. The method was published in Nature in January 2016, along with the news that AlphaGo had beaten the European champion Fan Hui 5 games to 0 the previous October.

In March 2016, in Seoul, AlphaGo played a five-game match against Lee Sedol, one of the strongest players of the previous decade, watched by millions of people. In game two it played move 37, a stone on the fifth line that professional commentators first took for a mistake. AlphaGo's own estimate was that a human would play it about one time in ten thousand. It turned out to be decisive.

AlphaGo won the match 4 to 1. In game four, Lee Sedol answered with his own unexpected move, move 78, and won. It remains the only game a human won against that version of AlphaGo in an official match.

In 2017 DeepMind presented AlphaGo Zero, which learned Go entirely by playing against itself, with no human games, and surpassed the earlier versions.

Ask your AI: what do Go players say they learned from AlphaGo's style of play?

Did you know? Lee Sedol won the fourth game with his move 78. AlphaGo had put the chance of a person playing it at about 1 in 10,000.

Source: doi.org/10.1038/nature16961

See it in the museum →

← ResNet: 152-Layer Networks · The Translator Learns with Neural Networks →

All of Room 4: The Awakening

Cite this: Nazareno, I. (2026). AlphaGo Defeats Lee Sedol (2016). New Mind Museum. Retrieved November 17, 2026, from https://newmindmuseum.com/museum/alphago-defeats-lee-sedol-2016/