CMU Builds Poker-Playing AI

Researchers at Carnegie Mellon University have developed a Texas Hold ‘Em playing poker program, and they plan to enter it in the American Association for Artificial Intelligence Computer Poker Competition in July.  Unlike most poker algorithms, this program, called GS1, bases its strategy on hand analysis rather than taking its cues from human-inspired poker heuristics.

Texas Hold ‘Em Poker is believed to be a better test of artificial intelligence than chess, in which computer players have become nearly unbeatable.  In Hold ‘Em, the player must make decisions based upon guesses or assumptions about the hidden cards in others’ hands, whereas the entire board is a known quantity in a chess match.  For this reason, a computer program that can win reasonably often at Hold ‘Em would be displaying a different type of intelligence than one that consistently wins at chess.  The intelligence gained would be much more appropriate for analysis of real-world situations, in which unknown factors can play a large part.

Read More | Science Daily