AI To Develop Computational Skill For Poker Games

There are tournament rounds going on in Pittsburgh currently where several hands of poker are being played every day for a certain period in order to test the mind power of humans against AI.

There was a similar contest held last year, which included 80, 0000 rounds being played. In that contest the AI lost, but by a very narrow margin. The team CMU that is pitting the AI machine against human players hopes to see a different outcome this year.

The contest is a marathon one that is an attempt by the university brain trust to showcase that AI can win against humans in complicated games like No Limit Hold ‘em poker. There have already been contests held for simpler games like tic tac toe, chess and Go where the machine has been able to beat human players. However, certain complicated games of poker provide a more difficult scenario.

For instance, players do not showcase their hands in such a game as well as purposely mislead their opponents by bluffing. These are strategies that make the game complicated and difficult for a machine to replicate and match up to.

The imperfect amount of information makes it hard to make the machine match up to human players, especially in the more complicated games of poker. For instance, the no limit Texas Hold’em game includes complex challenges as there can be different combination of moves at different stages of such a game. With such statistics, the randomness of moves becomes difficult to predict and for a machine to match the same as found in human players. The contest talks about pitching humans against machines, but there is much more to it. If the machines can be showcased to be able to defeat human players it would show advances in game theory and computation skills. Game theory is particularly applicable as two or more players are involved here.