Starting on January 11, the Carnegie Mellon University School of Computer Science and four poker professionals – Jason Les, Dong Kim, Daniel McAulay and Jimmy Chou – will once again battle on the virtual felt in the “Brains vs. Artificial Intelligence: Upping the Ante” competition.
In the 2015 competition, Les and Kim were joined by fellow poker professionals Bjorn Li and World Series of Poker bracelet winner Doug ‘WCGRider’ Polk in taking on “Claudico.” Over the span of 80,000 hands, the foursome was able to defeat the computer to the tune of $732,713, with Li racking up $528,033 of that amount, Polk picking up $213,671 and Kim earning slightly more than $70,000. Les will be the one looking for revenge as, in 2015, he was the only human to “lose” to “Claudico” by the amount of $80,482.
For the 2015 battle, the Carnegie Mellon computer “Claudico” was pre-programmed with three million hours of computation to base its decisions on. “Libratus” will get five times that amount (15 million hours) and has been adjusted for some of the “tells” that humans noticed from “Claudico” in the 2015 competition. All of this has the Carnegie Mellon people very confident about their chances.
Over the next 20 days (starting at 11AM and concluding at 7PM), each human player will play heads-up no limit Texas Hold’em against a new opponent, the brainchild of the programmers at Carnegie Mellon. Named “Libratus” (perhaps a derivative of the name of the Roman goddess Libertas, or “liberty,” or perhaps a name indicating balance (Libra), this computer program has been in development since 2015, when the Carnegie Mellon/Poker Pro battle was last waged.
The guidelines of the action have a few tweaks over the previous competition’s brilliant format. For the 2017 version, two matches will be played simultaneously online – one human player will be on the floor of the Rivers Casino in Pittsburgh, one human player in an isolated room away from that action. Along with the extended time frame (the 2016 competition lasted only 14 days), this will allow the extra hands to be played and should allow for a more definitive answer as to who wins the match. At stake is a $200,000 payout (an improvement over the $100,000 offered in 2015).
Tuomas Sandholm, a professor at Carnegie Mellon who, along with Ph.D. student Noam Brown, created “Libratus,” explained in the school’s press release the importance of the challenge:
“Since the earliest days of AI research, beating top human players has been a powerful measure of progress in the field,” Sandholm said. “It 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,” Sandholm concluded. The development of such computers – and their ability to make complex decisions rapidly – has an application to the benefit of mankind in that the AI can be used in medical analysis, the military, cybersecurity, and other business applications."