AlphaGo defeated Go champion Lee Sedol by four games to one in a 2016 match.
Developed by Google DeepMind, AlphaGo combined neural networks with tree-search methods. Its policy network estimated promising moves, while its value network evaluated positions. The system was trained first on human games and then improved through self-play, allowing it to handle Go’s vast number of possible positions.
The five-game match took place in Seoul, South Korea, in March 2016. Lee Sedol won the fourth game, producing a celebrated move that challenged AlphaGo, but the program won the other four. The result was especially notable because Go has a much larger search space than chess, making straightforward brute-force calculation impractical. Later systems, including AlphaGo Zero, learned through self-play with less reliance on human game records.