After its latest major competition with human poker professionals, computerized artificial intelligence is 3-0.
“Lengpudashi,” an updated version of the Libratus artificial intelligence program developed at Pittsburgh’s Carnegie Mellon University, handily whipped Chinese venture capitalist and World Series of Poker veteran Alan Du in a contest staged at a resort conference center on China’s Hainan Island.
After Du and his five team members played 36,000 hands of Heads-Up, No-Limit Texas Hold ‘em over five days, the computer slammed the door with nearly $800,000 in chips and $290,000.
The name, incidentally, is Mandarin for “cold poker master.”
The poker-playing AI computer is housed in a supercomputing center near the university in Pittsburgh. In 2015, the program, developed by Professor Tuomas Sandholm and Noam Brown, a graduate student at CMU, won the first “Brains vs. AI” contest at Pittsburgh’s Rivers Casino, in which the Libratus program beat professional players led by California pro Jason Les. In January of this year, there was a rematch held at Rivers between Libratus and a group of four pros—Les, Dong Kim, Daniel McAuley and Jimmy Chou. The computer again took the pot.
The recent competition with Du’s team was organized by Sinovation Ventures and Hainan’s local government. In an interview with Bloomberg, program co-developer Brown said AI is capable of making strategic decisions in a controlled environment. “People have a misunderstanding of what computers and people are each good at,” he said. “People think that bluffing is very human—it turns out that’s not true. A computer can learn from experience that if it has a weak hand and it bluffs, it can make more money.”
