Card-Playing Machine Takes Early Lead In Historic Match Against Poker ProsPoker Bot Up About 750 Big Blinds |
|
Round 1 goes to the machine.
On Wednesday at Rivers Casino in Pittsburgh, Pennsylvania, four elite heads-up no-limit poker pros began the latest man vs. machine poker match against Carnegie Mellon University’s world-class bot. Before it was Claudico, this time around it’s Libratus.
The poker pros played a combined 2,800 hands (of the scheduled 120,000) over the course of about eight hours and finished down 75,000 chips (blinds 50-100). The chips have no cash value, but the players—Jason “PremiumWhey” Les, Dong “Donger Kim” Kim, Daniel “Dougiedan678” McAulay and Jimmy “ForTheSwaRMm” Chou—are on a $200,000 freeroll.
Here’s a look at how each player did on Wednesday:
Though they can share strategies during the 19-day contest, the poker pros play their matches separately. Each battle is live streamed on Twitch.
The computer program comes from computer scientist Tuomas Sandholm, who with his team has developed a series of high-level poker bots over the years.
“Libratus’s strategy is not based on the experience of expert human players, so its game play could differ markedly from the pros,” Sandholm and his team wrote.
“It uses algorithms to analyze the rules of poker and set its own strategy, based on approximately 15 million core hours of computation at the Pittsburgh Supercomputing Center (PSC). Libratus will continuously sharpen its strategy during the Brains vs. AI competition, performing computations with the PSC’s Bridges computer each night while the pros get some shuteye. During games, Bridges will perform live computations to aid Libratus with its end-game play.”
According to Sandholm, international betting sites consider Libratus a definitive underdog in the contest, with odds varying between 4-to-1 and 5-to-1 against the AI.
Got a pretty sweet setup here at @WinBigRivers for the #BrainsVsAI challenge. The heads up sweatshop starts now pic.twitter.com/f86ZNtRLC4
— Jason Les (@heyitscheet) January 11, 2017