DeepMind's Player of Games AI Tackles Both Perfect and Imperfect Information Games

DeepMind, backed by Google, has developed an AI system called Player of Games that excels in both complete and incomplete information games. This advancement is seen as a significant step towards solving real-world problems, such as efficient routing, customer relations, and contract negotiations.

Player of Games is notable for being the first algorithm to achieve strong empirical performance in large perfect and imperfect information games. This capability sets it apart from other AI systems that are typically designed for specific goals.

In recent tests, Player of Games comfortably defeated Slumbot, a strong NL Hold’em AI program. It also held its own against Stockfish, one of the top chess engines, proving stronger in many configurations. However, against AlphaZero in a game of go, Player of Games won only 0.5% of the games but was noted to perform almost at the level of a professional human player.