Computer puts on its poker face for test vs. top pros at casino
It promises to be a man versus machine competition that could become as iconic as chess champion Gary Kasparov taking on IBM’s Deep Blue computer two decades ago: Today, a Carnegie Mellon University-designed computer program will kick off a two-week challenge to beat four of the world’s best poker players.
Officially dubbed by CMU as “Brains Vs. Artificial Intelligence,” the tournament begins at 11 a.m. at Rivers Casino on the North Shore and runs every day for eight hours through May 7, playing 80,000 hands of the most complex version of one-on-one poker: Heads-Up, No-Limit Texas
Hold ’em.
The four competitors — including the world’s No. 1ranked player, Doug Polk — will compete for their share of a $100,000 pot of winnings funded by the casino and fellow sponsor Microsoft.
Tuomas Sandholm, the renowned CMU computer science professor whose team designed the computer program, said he is really excited about the competition because the hand-picked human competitors “are really behaving like Kasparov and going about it in a really brave way, even though there has been some concern in the poker community about losing to a bot.”
The casino has erected stands for the public to watch the live action (spectators must be 21 or older), television screens to view the electronic hands during play and a tally board to see how each human is doing against their computer opponent, dubbed “Claudico.” (Claudico is the Latin word for “limp.” It’s a reference to a poker strategy called “limping,” in which a player, often with a bad hand, calls rather than folds or raises. It is not generally a popular tactic among humans but the computer program finds it to be advantageous, Mr. Sandholm said.)
Mr. Polk, 26, from Pasadena, Calif., said, like his three fellow competitors, he came to Pittsburgh for the first time because “it’s pretty cool to be at the cutting edge of where computer technology is.”
But there’s also a much more visceral desire. “I hope we can stand up for humanity and take this computer down,” he said with a laugh. “I know computers will eventually be able to beat humans. But I hope we can make them go a few more rounds after this before they do, like Kasparov did.”
There is a real fear in the online poker community that a computer program could consistently beat a human in Heads-Up, No-Limit Texas Hold ’em since it is the most complex version of the game. Mr. Sandholm’s computer program has previously triumphed in less-complicated versions of poker, including Limit Texas Hold ’em. But taking on No Limit Texas Hold ’em — because of being able to bet as much or little as you want on each hand — is significantly more challenging.
In fact, the No Limit game is so complicated it opens up more possible situations than there are atoms in the universe, Mr. Sandholm said.
Some worry that if Mr. Sandholm and his team succeed it will spell the end of online poker since it would be difficult to devise a way to ensure that someone has not simply hooked up a computer program to play the game for them and win most of the time.
“That’s one of the major talking points in online poker: the invasion of artificial intelligence,” said Bjorn Li, 25, another of the human competitors. “There’s just no way to enforce” trying to keep people from using computers to help online.
Some foresee that it will drive the top online poker players to play in person, so everyone knows who they are playing.
“That might be what it becomes,” said Mr. Li, who grew up in Hong Kong. “But hopefully not because [inperson poker] is so slow and tedious.”
Since the United States crackdown five years ago on online poker, top players like Mr. Polk, Mr. Li, and the other two competitors — Jason Les, 29, from Costa Mesa, Calif.; and Dong Kyu Kim, 26, originally from Honolulu — have to leave the country to play online. When they do, they regularly play multiple games at the same time.
While today’s competition is focused on the poker game, Mr. Sandholm and his CMU team are hoping these highly skilled players will help them learn how to improve their program for much broader concerns.
“We are using these algorithms [in the computer program] for poker,” Mr. Sandholm said. “But these algorithms we developed are really for use in any game of incomplete information in general, from cybersecurity, to negotiations, to medicine.”
That idea — that playing poker at the highest level will directly benefit society — was an added incentive for the four competitors.
“We’re here to play poker, and we see the immediate effect of this program. And we don’t want to lose,” said Mr. Les. “But the fact that we could be part of something that does make the world better? That’s pretty incredible.”