Facebook Unveils Poker Ace AI

Facebook and CMU's poker AI beat five pros at once

Facebook and CMU's poker AI beat five pros at once

Apart from Pluribus, there are other AI as well that have gained mastery over zero-sum games such as checkers, chess, Go, two-player poker, StarCraft 2, and Dota 2.

"Now that we've achieved that", said Brown, "it's really about going beyond poker to other domains".

Over 10,000 hands of no-limit Texas hold'em, the most popular form of the game, Pluribus won a virtual $48,000 (£38,000), beating five elite players who were selected each day from a pool who agreed to take on the program. "Pluribus is a very hard opponent to play against".

TWO YEARS AGO, a poker-playing robot called Libratus whooped professional card sharks at their own game, winning $1.7m in the process. Then, when playing against humanopponents, Pluribus improves the blueprint strategy by searching in real time for a strategy that better suits the circumstances of the current game. The AI was able to bluff and switch up its strategy; it also bet boldly where human players might have shirked from the stress, Gizmodo's George Dvorsky reports.

What Libratus did was to go back four moves for every player move and try to reason why they would play the way they did.

Noam Brown, a research scientist at Facebook AI Research, and Tuomas Sandholm, a computer scientist at Carnegie Mellon, describe their results in a brand new paper titled "Superhuman AI for multiplayer poker", revealed today within the journal Science.

Sean Ruane, one of the players who took on Pluribus, found the program a tough opponent.

To reduce the role of luck, the researchers developed a version of the AIVAT variance reduction algorithm, which applies a baseline estimate of the value of each situation to reduce variance while still keeping the samples unbiased. For instance, betting would make sense if the AI held the best possible hand, but if the AI bets only when it has the best hand, opponents will quickly catch on. The program was created using a 64-core server packed with less than 512GB of RAM, but it runs on two Haswell processors and uses 128GB during games.

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

Another significant achievement of this bot is that it was trained using little power and memory resources.

In addition to that, Pluribus also has "faster self-play algorithms for games with hidden information", which means that it is more efficient when it comes to dealing with hidden information in games it played against copes of itself. The first involved five human players and one AI, and the second one had five AIs and one human. The Pittsburgh Supercomputing Center provided computing resources through a peer-reviewed XSEDE allocation.

Multi-player interactions pose serious theoretical and practical challenges to AI techniques.

Artificial intelligence has already surpassed humans at games such as chess and Go, but in these games players can see the positions of all of the pieces; there is no information that is hidden from them.

Few applications in the real world scenario can be taking action on harmful content and dealing with cybersecurity challenges, as well as managing an online auction or navigating traffic.

The poker bots have been at it again, crushing human players on the virtual felt. With Pluribus, AI has entered the coveted arena of AGI and it looks poised to make more breakthroughs.