Humans Prove They Can Compete With AI Poker Software

Humans Prove They Can Compete With AI Poker Software May 11, 2015 May 11, 2015 Tim Glocks
Posted on  May 11, 2015 | Updated on  May 11, 2015 by Tim Glocks

Tuomas SandholmAn expert team at Carnegie Mellon University created an artificial intelligence software that was designed to play Heads-Up No-Limit Texas Hold ‘Em.

The AI named Claudico was built after studying thousands of hours of poker and then formulating an algorithm based on the best poker strategy the machine had mapped out after its analysis.

Claudico had the challenge of competing with four of the top poker players in the world and testing its poker playing skills against these experts.

The four players had an exceptional record playing Heads-Up No-Limit Texas Hold ‘Em and accepted the challenge to play Claudico over a two week span that involved over 80,000 hands.

After two weeks of intense competition, there was no outright winner and the final result was registered as a statistical tie. It was Bjorn Li who finished the competition with the most prize money in the form of $529,000 in chip; Dough Polk took second place with $213,671 in; third was Dong Kim with $70,491 in chips and Claudico beat Jason Les took take the fourth place.

All four professional poker players received a total of $100,000 to play the event. The money was raised by the Rivers Casino and Microsoft Research who wanted to further the research of artificial intelligence in gaming and see how well Claudico could compete. The end result proved that while Claudio was the best AI software in the poker industry, biological professional poker players still had the upper hand.

In a statement, Tuomas Sandholm, Computer Science Professor from Carnegie Mellon University and the man behind the development of Claudico said

We knew Claudico was the strongest computer poker program in the world, but we had no idea before this competition how it would fare against four Top 10 poker players. It would have been no shame for Claudico to lose to a set of such talented pros, so even pulling off a statistical tie with them is a tremendous achievement.

While there are a number of poker variants out there, the game of No-limit Texas Hold ‘Em is the best game for an AI to learn and then play. The game is based on bluffing as all the cards cannot be seen at once and this specific model is extremely important to developing and advancing all forms of artificial intelligence. However, the end result shows that Claudico still has a long way to go before becoming proficient enough to play and beat the top poker players in the world.

Tim GlocksAuthor

Tim Glocks is a retired professor, he currently contributes to Tim enjoys playing poker and has taken it up as a hobby since his retirement. He has taken part in many online tournaments and has become a veteran in a short space of time. Visit Tim’s google + page here