In an era where artificial intelligence continues to dazzle with its capabilities, Google’s DeepMind lab has unveiled a new AI known as ‘Student of Games’ (SoG), with the power to outperform human intellect in games that range from the strategic depth of chess to the unpredictable nature of poker. This AI marks a clear leap towards the development of an artificial general intelligence, a technology that aspires to outpace human ability in a sweeping array of tasks.
Recounting SoG’s evolutionary roots, it is evident how the project draws on the heritage of two pivotal AI endeavors – DeepStack and AlphaZero. DeepStack, originating from a collaboration including Martin Schmid at the University of Alberta, was the first AI to triumph over professional poker players. AlphaZero, another DeepMind creation, shattered records by defeating top-tier human players in chess and Go. The distinction between two was their handling of games with different levels of accessible information to the player: poker, a game of hidden information, versus the complete transparency seen in chess.
Google DeepMind's Student of Games AI Masters Both Chess and Poker https://t.co/GIqojvSVwm
— Amb Antonio Garza (@aogarza) November 16, 2023