An artificial intelligence has managed to beat the professional poker champions: what other uses will it have?

Playing is one of the keys to learning, including for the machines. In fact, in the road evolutionary artificial intelligence There are many playful episodes. Skills from the Deep Blue supercomputer that beat the great chess champions to recent developments that have surpbaded the players in complex video games.

Now, researchers from Facebook and Unversidad Carnegie Mellon, from the United States, have jointly developed a artificial intelligence who wins poker games. And beyond the virtues of this game, experts say the knowledge gained will also be used in other areas, such as finance or cybersecurity.

Artificial bets

Called "Pluribus", This AI system boasts some feats, from one side, Beat the poker professionals, successful in their field. For the rest, the big step of this technology is overcome many humans in unison.

The antecedents of this system, called Claudico (created in 2015) and Libratus (2017), had not crossed the thresholds to recognize their respective victories over humans.

Usually, machines win in face-to-face clashes, although they are not as skilled at defeating many players who face each other at the same time. For that, Multiplayer variants pose additional challenges to these systems. In fact, in Latin "pluribus"It means" a lot ".

It should be noted that the "Pluribus" victories against humans took place according to the unlimited modality of Texas Hold & # 39; em, a variant common in the main poker competitions.

"No other popular recreational game captures the challenges of hidden information as effectively", Comment the researchers involved in the study.

Darren Elias, professional player and owner of record of World Poker Tour titles, said that the machine "goes from a mediocre player to compete with the best in the world in a few weeks" and that "its strength lies in its ability to use mixed strategies, in the same way that humans do" .

More precisely, Pluribus conceives its tactics according to the results of the virtual games that it plays against copies of itself. In 12 days, play around 10,000 hands. In the six-player tables (the system included), he defeated Elias and Chris Ferguson, six-time World Series of Poker event winners.

In the study, scientists noted that poker posed great challenges to the AI ​​system for which he had to understand and badyze multiple pieces of information, including cheating. And that's why progress can be applied in other areas, beyond this game.

The plan of the experts is that the learning provided by "Pluribus" will serve in complex situations involving many people, from finance to traffic jams. Also to improve computer security or create other robotic systems, according to experts.

"It was fascinating to play against the bot and see some of the strategies that he chose"said Michael Gagliano, another participating player. There have been several moves that humans simply do not do, especially with regard to the size of the bets, artificial intelligence is an important part of the evolution of poker, and that's all. was incredible to have concrete experience of this great step towards the future, "he concluded.