AlphaGo is the reinforcement-learning system that defeated professional Go player Lee Sedol in a five-game match in 2016.
Google DeepMind developed AlphaGo by combining neural networks with tree-search methods. One network estimated promising moves, while another estimated the likely winner of a position. The system first learned from examples of human games and then improved through games played against itself.
In March 2016, AlphaGo beat Lee Sedol four games to one in Seoul. The match attracted worldwide attention because Go has a vastly larger space of possible positions than chess. Its success demonstrated that combining learned pattern recognition with search could handle a difficult strategic domain.
AlphaGo should not be confused with AlphaZero. AlphaZero was a later, more general system that learned chess, shogi, and Go from the rules and self-play rather than relying on human game examples. AlphaGo’s famous match also produced the celebrated Move 37 in game two.