The scientifically monitored poker competition between four respected professional poker players and Carnegie Mellon University’s Libratus artificial intelligence computer program continued to illustrate this week how far the technology has progressed over previous man vs. computer contests.
Poker pros Jimmy Chou, Dong Kim, Jason Les and Daniel McAulay are struggling to contain the program, which is apparently able to learn as it plays and indulge in a form of bluffing. After 68,000 hands of the scheduled 120,000 planned for the contest, Libratus is crushing the pros and is ahead by $701,300 in chips.
The pros won’t have to pay that, but if they win they could walk away with $200,000 in real money. That prospect is fading as the competition continues toward the two-thirds complete mark, and some of the remarks made by members of the human team appear to acknowledge that their chances of victory are decreasing.
Chou said this week that Libratus seems to get better as each day of play passes. “It’s like a tougher version of us,” he observed.
Kim agreed, saying: “I didn’t realize how good it was. I felt like I was playing against someone who was cheating, like it could see my cards…I’m not accusing it of cheating. It was just that good.”
In 2015 a contest between the Carnegie Mellon AI program Claudico and four poker pros – Dong Kim, Jason Les, Bjorn Li, and Doug Polk – showed that the human mind remained in charge; the four pros took $732,713 off Claudico over a total of 80,000 hands.
Libratus was built with more than 15 million core hours of computation as compared to 2-3 million for Claudico, and judging by the players’ reactions its capability is superior.
The competition ends on January 31.