Skip to content
BENIBIX Learn

AlphaGo

AI🧭 FoundationsAI History

Also known as AlphaGo vs Lee Sedol, DeepMind AlphaGo

DeepMind's Go program that beat top player Lee Sedol 4 to 1 in 2016.

Think of it like

It is like a kid who plays a game with themselves a million times. They get good.

Example

In October 2015, AlphaGo beat the European champion Fan Hui. Five months later, it beat Lee Sedol in Seoul.

Why it matters

Many experts thought Go was a decade away for AI. It showed that learning could beat search on very hard problems.

Definition

AlphaGo is a Go-playing program made by DeepMind. Go is a board game with far more possible moves than chess, so search alone was not enough. AlphaGo learned from human games and then from playing against itself. In March 2016, in Seoul, it beat Lee Sedol, one of the best players in the world, by 4 games to 1. Over 200 million people watched the match.
AlphaGo, 2016: a program that learned Go beat one of its best players.

Common mistake

Thinking AlphaGo was told how to play. It learned its moves from games, including millions against itself.

Key takeaways

  1. 1
    Learned, not coded. From human games and self-play.
  2. 2
    Beyond search. Go has too many moves to check.
  3. 3
    4 games to 1. Lee Sedol, Seoul, 2016.

Compare

Learn it properly