New Student of Games AI Masters Both Chess and Poker
Researchers introduce Student of Games, a unified algorithm that combines search, self-play, and game-theoretic reasoning to excel in both perfect and imperfect information games.
Researchers introduce Student of Games, a general-purpose artificial intelligence algorithm that unifies previous approaches by combining guided search, self-play learning, and game-theoretic reasoning. This new method bridges the gap between perfect information games like chess and imperfect information games like poker, which traditionally require entirely different AI strategies.
The algorithm demonstrates strong empirical performance across a variety of complex board games. Student of Games achieves high-level play in chess and Go, defeats the strongest openly available agent in heads-up no-limit Texas hold'em poker, and overcomes the state-of-the-art agent in Scotland Yard.
The creators prove that Student of Games is mathematically sound, showing that it converges to perfect play as computational power and approximation capacity increase. This breakthrough represents an important step toward developing truly general algorithms capable of navigating arbitrary environments.