Contact us

Next challenge for AI after winning with top human chess, go and poker players

In January 2017 the Libratus AI system won over $1.7 million in chips with four of the world’s best professional players in 20-days No-Limit Texas Hold ‘Em poker tournament at a Pennsylvania casino.

Go represents the ultimate in games where all the information is available to the players. In poker a player doesn’t know what card is coming next. “A poker-proficient AI (especially no-limit poker that is much more complicated version of the game) is remarkable because poker is a game of “imperfect information”. Players don’t know what cards their opponents have, so never have a full view of the state of play, says Peter Stone at the University of Texas at Austin. This means the AI has to take into account how its opponent is playing and rework its approach so it doesn’t give away when it has a good hand or is bluffing. “This is like reality. The real world is a game of imperfect information.” says Georgios Yannakakis at the University of Malta.

Libratus was taught only rules of poker and no game strategy. AI even needed to learn how to bluff so its moves do not become predictable for professional human players. The AI has not been taught any strategies. Part of Libratus called the “endgame solver” takes into account “mistakes” the AI’s opponents made – instances where they left themselves open to exploitation – to predict the result of each hand. The part of the AI looks also for its own strategic weaknesses so it could change how it played before the next session. This sought to identify things its opponents were exploiting, such as a giveaway “tell” that another player had noticed. “Its strategy seems to be improving as time goes on and it is tougher and tougher every day.” said Jason Les, one of the professional players in the tournament.

Libratus could be applied to any situation that requires a response based on imperfect information like cybersecurity, negotiations, military settings, auctions or even fighting infections,” says Tuomas Sandholm, scientist at Pittsburgh Supercomputing Center where Libratus AI was developed. PSC is joint effort of Carnegie Mellon University, the University of Pittsburgh and Westinghouse Electric nuclear power company. „There is still a long way to go before AI can take on the real world, says Simon Lucas at the University of Essex, UK. “In the real world, you often have a lot more choices than in a card game. The possibilities are more open-ended,” he says.


Sources and media:

You might also like

Could you train your vacuum cleaner like your dog?

Author: Tomasz Tajmajer

Artificial intelligence methods are often perceived as extremely complicated and understandable only by people that graduated computer science. The idea of an electronic device that, by itself,  learns how to behave still seems to be very futuristic. In fact, we have many “smart” devices around us, but many of them are not smarter than a motion-based light switch…

Read more
Nature-Inspired Algorithms

Author: Maciej Świechowski Ph.D.

“Look deep into nature, and then you will understand everything better” – Albert Einstein

Since ancient times, nature has been serving as inspiration for humans in the shaping of our civilization and culture. A plethora of brilliant inventions happened just through careful observation of nature…

Read more
How Machine Learning can improve AI in Hearthstone: Heroes of Warcraft

Author: Andrzej Janusz

This paper reviews different applications of machine learning techniques in a development of AI player for a collectible card video game Hearthstone: Heroes of Warcraft.

Read more
Decision making algorithms for board or card games

Author: Piotr Beling

Decision making algorithm is an algorithm which tries to calculate the best move possible to make in a given position in a game. In other words, for given input position, it points the move which is possible the best one.

Read more