PokerWorks

The second Man vs. Machine Poker Championship was played out at the 2008 Gaming Life Expo in Las Vegas. After four rounds of play, the Polaris artificial intelligence program beat the live professional poker players to claim victory.

Polaris came computer-to-face with Nick Grudzien, founder of Stox Poker, and several other pros who are members of or contributors to the site. Each round consisted of 500 hands wherein Polaris played against two humans whose points were then averaged to create the humans’ score. In the end, the four rounds of play resulted in Polaris claiming two wins, one draw, and only one loss.

Michael Bowling, who currently heads up the University of Alberta Computer Poker Research Group that developed and continue to improve Polaris, noted that Polaris was able to shift its strategy midway through the rounds to keep players from detecting its weaknesses. In addition, Polaris was then able to detect the type of strategy that the humans were using and select a counter-strategy.

The first Man vs. Machine Poker Championship was held in 2007 with Polaris versus two well-known poker pros – Phil Laak and Ali Eslami. The match was won by the Laak/Eslami team via two wins, one loss, and one statistical tie, but the humans noted that the play they experienced at the hands of Polaris was incredibly challenging. And as Polaris programmers continued to improve the artificial intelligence program (AI), they were confident that it could win the next match-up, which it did in Las Vegas.

In a setting that was not entirely coincidental – the 2008 World Series of Poker in Las Vegas playing out just down the hall in the Rio Convention Center – Polaris and its developers and creators were highly pleased that victory was achieved. But Bowling said that there must be several rematches and more competition in which Polaris is victorious before people are convinced that machines can best humans in the game of poker.

The University of Alberta Computer Poker Research Group is also planning to upgrade Polaris by enabling it to play more complicated poker games, expanding it beyond its current heads-up hold’em constraint. Those involved with the program also intend to take some of the intricacies of Polaris and begin applying them to other applications, such as sensors in buildings.

Further matches have yet to be announced, but if the AI programmers have their way, the poker world has not seen the last of Polaris.