Can “Claudico,” an AI poker supercomputer, use technological programming to top the world’s best professionals in the game? That’s what researchers at the Carnegie Mellon University in Pittsburgh are trying to discover, but halfway through the 80,000 hands of competition, it’s quickly becoming evident that brains tops artificial intelligence when it comes to poker.
The computer software is up against Doug Polk, Dong Kim, Bjorn Li, and Jason Les, with each pro playing 20,000 hands of Heads-Up No-Limit Texas Hold’em over the two-week marathon event. “Computing the world’s strongest strategies for this game was a major achievement, with the algorithms having future applications in business, military, cybersecurity and medical arenas,” Tuomas Sandholm, professor of computer science and Claudico lead said.
After the first week of play, the pros have opened up a substantial lead.
A large component of poker is reading the competition, identifying and distinguishing subtle clues to give you the upper hand. Of course, that aspect of the game is eliminated from the experiment as Claudico won’t be giving off any hints.
While that is a considerable difference from playing live poker, it’s actually quite similar to competing online as players typically can’t see one another. Instead, players monitor the habitual strategies of their opponents to try and predict their next move.
Polk, one of the world’s best Heads-Up No-Limit Texas Hold’em players, said his strategy would differ from his normal approach. “I think there will be less hand reading so to speak, and less mind games. In some ways I think it will be nice as I can focus on playing a more pure game, and not have to worry about if he thinks that I think, etc.”
After the first week of play, the pros have opened up a substantial lead. As of Friday morning, 42,100 hands into the epic event, the players are cumulatively up $568,662 in pretend money, with Les being the lone brain in the hole versus the computer.
Friday morning standings:
Though much could change during the second week, both players and computer programmers are already thinking about a rematch. “I’d like to do this again because the computer will eventually be able to beat the humans,” Polk said.
The humans versus computers concept is nothing new. In 1997, IBM’s “Deep Blue” artificial intelligence program outdueled Garry Kasparov in a chess match, arguably one of the greatest players ever. In 2011, “Watson” topped “Jeopardy” champs Ken Jennings and Brad Rutter in a three-episode contest.
But where “Jeopardy” is a game of facts and knowledge, poker is much more complicated, making the computer programming exponentially more complex.
“Poker is now a benchmark for artificial intelligence research, just as chess once was,” Sandholm explains. “It’s a game of exceeding complexity that requires a machine to make decisions based on incomplete and often misleading information, thanks to bluffing, slow play, and other decoys. And to win, the machine has to out-smart its human opponents.”
Polk’s prediction of the computer one day becoming better at heads-up poker remains to be seen, but for now, humans reign supreme.