Poker-Playing AI Will Not Be Released For Fear of Threatening the Online Poker Industry

Poker has always been one of those games that requires more skill than luck.

To triumph over his opponents, a player must be good at reading people, find a way to trick them into handing over their chips.

This set of skills is something that comes naturally to human beings. Now, for the first time, an AI program has been developed that has proven to be capable of outwitting a whole table of poker professionals using similar skills.

A team from Carnegie Mellon University (CMU) and Facebook combined techniques to create an AI program to out-bet and out-bluff human players in a game of six-player, no-limit Texas Hold’em.

And these human players were not amateurs. The table included poker pros Darren Elias, who holds the record for most World Poker Tour titles, Chris “Jesus” Ferguson, a winner of six World Series of Poker titles, as well as one inferior poker bot.

The new AI, named Pluribus, demonstrated its skill while playing 5,000 hands against the poker players and consistently won more than its opponents.

It also emerged victorious in another test involving 13 players and 10,000 hands. Pluribus seemed to adopt some unexpected strategies, such as “donk betting”, or ending one round with a call but then starting the next with a bet. It also showed incredible bluffing skills worthy of a seasoned pro.

This achievement represents a major landmark for artificial intelligence. A computer program must use more than just brute computation to prevail at poker, as well as an ability to negotiate under unpredictable circumstances, using the principles of game theory.

The algorithm was so successful that the researchers have decided not to release its code because it could lead to online poker companies losing a lot of money.

“It could be very dangerous for the poker community,” says Noam Brown, a Facebook researcher and former student at CMU who helped in developing the algorithm.