PSC Helps Power Human-AI Poker Match-Up

Friday, April 24, 2015

First, Deep Blue came for Kasparov, besting the world’s top-ranked chess player in 1997. Then Watson beat Jeopardy!’s best and brightest in 2011. Now Claudico, a program developed at Carnegie Mellon University using high performance computing thanks to an XSEDE allocation on Pittsburgh Supercomputing Center’s Blacklight system, will take on the next great challenge in machine learning: facing off against four of the world’s top 10 players at no-limit, heads-up Texas hold ’em poker. Read the CMU press release

The two-week “Brains vs. Artificial Intelligence” tournament begins today at Rivers Casino in Pittsburgh, with poker pros Doug Polk, Dong Kim, Bjorn Li and Jason Les facing off against Claudico. Unlike earlier human-machine poker matches involving bid-limited Texas hold ’em, this competition will involve the vastly more complex unlimited version, which offers 10161 (1 followed by 161 zeroes) possible game situations. Not only a program for playing Poker, Claudico is the product of a set of algorithms developed by CMU’s Tuomas Sandholm and students that can automatically derive optimum strategies for virtually any situation in which opponents possess incomplete information, given only the rules of the “game.”

View continuous play:

You can also follow the tournament at @psc_live (Twitter) and

Rivers Casino is a gambling establishment; no one under age 21 is permitted on casino property.