Computer Go

Computer Go is the study and development of programs that play the board game Go, combining search, position evaluation, and learned strategies to select moves.

Connect

Monte Carlo tree search: A search method that estimates action values by sampling outcomes from a tree of possible decisions. It made selective search practical in Go’s enormous branching game tree.
Computer chess: The development of computer programs that analyze and play chess. Chess engines advanced through deep search, while Go resisted that route for decades.
Go: A two-player strategy board game in which players surround territory by placing stones on a grid. The game’s rules define the positions and legal moves that programs must model.
KataGo: An open-source Go engine that combines neural networks with search and supports analysis features. It makes modern engine analysis available for study, review, and play.
Lee Sedol: A South Korean professional Go player who competed against AlphaGo in 2016. His match exposed professional players to moves generated by a nonhuman opponent.
Neural network: A computational model that learns patterns by adjusting weighted connections between artificial units. Go programs use neural networks to estimate positions and guide move selection.
John McCarthy: An American computer scientist who helped establish artificial intelligence as a research field. McCarthy proposed using Go to study machine intelligence and game-playing programs.
Go opening theory: The study of established strategies and patterns for the early stage of a Go game. Opening knowledge once supplied programs with guidance beyond brute-force search.
Leela Zero: An open-source Go engine trained through self-play using an approach inspired by AlphaGo Zero. Its public self-play project demonstrated community-driven development of strong Go software.
Move 37: A move played by AlphaGo in its second game against Lee Sedol that surprised expert commentators. It became a prominent example of an engine move challenging human strategic intuition.