Poker AI player dominates poker pros

The future takeover of the human race by Skynet may end up starting at the poker table.

A new paper released last week in the academic journal Science reported on a new poker playing AI bot that has taken AI to another level. The AI poker player dubbed Pluribus played and won at a six-player No-Limit Texas Hold’em table against some of the best poker pros in the world.

Pluribius didn’t just win, it dominated.

Created through a partnership between Facebook and Carnegie Mellon University, Pluribus went beyond what previous AI poker players have done in the past. AI poker players have easily defeated humans before but those were in heads-up play against one opponent.

By dominating against a six-player table Pluribus has gone well beyond what its predecessors have done.

Pluribus took part in two different poker scenarios. In the first scenario five bots played against poker pro Darren Elias, a four-time WPT champion. In the other scenario it was just Pluribus alone against five poker pros.

A few of those pros included six-time World Series of Poker bracelet winner Chris Ferguson, and high rollers Michael Gagliano and Tony Gregg.

Over 12 days and 10,000 hands Pluribus cashed in, while showing expert skill at bluffing and betting. If each chip Pluribus won was worth $1 it won on average $5 per hand or $1,000 per hour while playing against the pros.

One of the amazing things about Pluribus was how quickly it became a dominant poker player. Pluribus’ initial poker training was only eight days long and it learned to play the game by playing against copies of itself.

There are no plans on what the next step is for Pluribus or poker playing AI. However, since poker playing AI has now stepped up from heads-up play to six-table play could an AI bot taking on a massive tournament field be in the cards for the future?