AlphaZero
AlphaZero is a general-purpose board game AI system developed by DeepMind. It combines deep reinforcement learning with Monte Carlo Tree Search and requires no human prior knowledge, learning entirely through self-play from scratch. The system has reached or surpassed top human performance in Go, chess, and shogi, demonstrating the broad potential of general-purpose game AI.
Core Facts
DeepMind's AlphaZero was released in 2017 and, through self-play alone, achieved a level surpassing the world's top humans in chess, shogi, and Go.
A world model refers to a robot's internal predictive capability of the physical environment—given the current state and an action, it can predict what will happen next.
DeepMind的AlphaGo通过自我对弈数百万局来习得围棋策略
Timeline (last 90 days)
In chess, the 'centaur' (human + AI) model was once regarded in the late 1990s to mid-2000s as the optimal solution for board game play, but with the emergence of self-learning systems such as AlphaZero, pure AI playing strength quickly opened a gap over any human combination, and by the 2010s this model had basically lost significance in top-level competition.
The mechanism of RL itself does not provide a source for goal generation; AlphaZero's game space is closed and its reward function is fixed.
AlphaZero proved that in domains with clear rules and definable rewards, RL can learn strategies surpassing humans from scratch.
PUCT（Predictive Upper Confidence bound applied to Trees）是 AlphaZero 系列算法中使用的一种树搜索策略，全称为 Predictor + UCT。
The AlphaGo series improved its playing strength through self-play.
METIS-Core implements a natively running AlphaZero self-play loop, in which Monte Carlo Tree Search (MCTS) is executed in a multi-threaded manner in C++.
AlphaZero extended the same framework to multiple games including chess and shogi.
AlphaZero used no human chess records and surpassed Stockfish through just 4 hours of self-play.
AlphaProof combines AlphaZero's reinforcement learning approach with the Lean formal language to generate proofs through self-play.
DeepMind's AlphaZero was released in 2017 and, through self-play alone, achieved a level surpassing the world's top humans in chess, shogi, and Go.
1 more timeline events
All Facts (13)
DeepMind's AlphaZero was released in 2017 and, through self-play alone, achieved a level surpassing the world's top humans in chess, shogi, and Go.
75%VerifiedA world model refers to a robot's internal predictive capability of the physical environment—given the current state and an action, it can predict what will happen next.
75%VerifiedDeepMind的AlphaGo通过自我对弈数百万局来习得围棋策略
65%UnverifiedAlphaProof is based on a combination of reinforcement learning and a formal proof system (the Lean proof assistant), exploring proof paths through algorithms such as Monte Carlo tree search.
70%UnverifiedThe AlphaGo series improved its playing strength through self-play.
60%UnverifiedIn chess, the 'centaur' (human + AI) model was once regarded in the late 1990s to mid-2000s as the optimal solution for board game play, but with the emergence of self-learning systems such as AlphaZero, pure AI playing strength quickly opened a gap over any human combination, and by the 2010s this model had basically lost significance in top-level competition.
50%UnverifiedAlphaZero proved that in domains with clear rules and definable rewards, RL can learn strategies surpassing humans from scratch.
50%UnverifiedThe mechanism of RL itself does not provide a source for goal generation; AlphaZero's game space is closed and its reward function is fixed.
50%UnverifiedPUCT（Predictive Upper Confidence bound applied to Trees）是 AlphaZero 系列算法中使用的一种树搜索策略，全称为 Predictor + UCT。
50%UnverifiedMETIS-Core implements a natively running AlphaZero self-play loop, in which Monte Carlo Tree Search (MCTS) is executed in a multi-threaded manner in C++.
50%UnverifiedAlphaZero extended the same framework to multiple games including chess and shogi.
50%UnverifiedAlphaZero used no human chess records and surpassed Stockfish through just 4 hours of self-play.
50%UnverifiedAlphaProof combines AlphaZero's reinforcement learning approach with the Lean formal language to generate proofs through self-play.
50%