Artificial intelligence
Get more with myNEWS
A personalised news feed of stories that matter to you
Learn more
Researchers at Carnegie Mellon University announced Thursday that their program defeated a group of top pros in six-player no-limit Texas hold ‘em. Photo: Reuters

Artificial intelligence program ‘Pluribus’ beats professionals in six-player Texas hold ‘em poker

  • Researchers at Carnegie Mellon University, with Facebook’s AI initiative, announced their program had beatthe world’s best poker players
  • AI programs have already beat us humans at draughts, chess, Go and two-player poker. But never multiplayer
Artificial intelligence programs have bested humans in draughts, chess, Go and two-player poker, but multiplayer poker was always believed to be a bigger ask. Mission: accomplished.

Researchers at Carnegie Mellon University, working with Facebook’s AI initiative, announced Thursday that their program defeated a group of top pros in six-player no-limit Texas hold ‘em.

The program, Pluribus, and its big wins were described in the US journal Science.

“Pluribus achieved superhuman performance at multiplayer poker, which is a recognised milestone in artificial intelligence and in game theory,” said Tuomas Sandholm, a computer science professor at Carnegie Mellon.

Sandholm worked with Noam Brown, who is working at Facebook AI while completing his doctorate at the Pittsburgh-based university.

Bots to dominate financial sector communications within two years in automation wave, messaging firm Symphony says

China's 19-year-old Go player Ke Jie reacts during the first match against Google's artificial intelligence programme AlphaGo in Wuzhen. Photo: AFP

“Thus far, superhuman AI milestones in strategic reasoning have been limited to two-party competition,” Sandholm said in a statement released by the school.

According to the creators of Pluribus, the technology could be used to solve a “wide variety of real-world problems” that, like in poker, involve actors who bluff, or hide key information.

The program first defeated two major poker champions, Darren Elias and Chris Ferguson, who each played 5,000 hands against it.

A replica of a game between 'Go' player Lee Se-Dol and a Google-developed super-computer, in Seoul, which beat the best human players. Photo: AFP

Pluribus then took on 13 pros in a separate experiment, five at a time. In a total of 10,000 hands, the program “emerged victorious,” researchers said.

First, the program practised against itself, learning little by little how to use poker moves to best advantage. Surprises cropped up.

Would you pay US$432,500 for this blurry painting created by an AI algorithm?

“Its major strength is its ability to use mixed strategies,” said Elias.

“That is the same thing that humans try to do. It’s a matter of execution for humans – to do this in a perfectly random way and to do so consistently. Most people just cannot.”

One surprise was that Pluribus used “donk betting” – ending one round with a call and starting the next with a bet – far more than would the pros, who traditionally see the move as a weak one.

Brown even ventured so far as to say that some of the program’s strategies “might even change the way pros play the game.”

This article appeared in the South China Morning Post print edition as: AI program calls bluff on major poker pros