The 20-day study will begin on Wednesday 11th January at Rivers Casino in Pittsburgh and will see each player participate in 120,000 hands for a cut of a $200,000 pot. Playing in pairs, the poker pros will pit their wits against ‘Libratus’, a brand new computer created by the university. The creation of a fresh AI unit follows a similar experiment in 2015 when ‘Claudico’, also a creation of experts at Carnegie Mellon, failed to collect more chips than three out of four of its human counterparts. However, with just 80,000 hands played, not enough statistical information was gathered in order to draw a definitive conclusion.
Tuomas Sandholm, professor of computer science at Carnegie Mellon and co-creator of ‘Libratus’, said in a university press release: “Since the earliest days of AI research, beating top human players has been a powerful measure of progress in the field.” The 48-year-old Finn continued: “That was achieved with chess in 1997, with Jeopardy! in 2009 and with the board game Go just last year. Poker poses a far more difficult challenge than these games, as it requires a machine to make extremely complicated decisions based on incomplete information while contending with bluffs, slow play and other ploys.”
Built from scratch, Libratus, meaning balanced and powerful in Latin, is a significant improvement on the previous Claudico model and is designed with an algorithm developed to compute and calculate potential winning strategies throughout games using the Pittsburgh Supercomputing Center’s Bridge supercomputer. “Extending AI to real-world decision-making, where details are unknown and adversaries are actively revising their strategies, is fundamentally harder than games with perfect information or question-answering systems,” explained Pittsburgh Supercomputing Center’s senior director of research Nick Nystrom.