Computers have bested human professional poker players. Whether the first victor was DeepStack at the University of Alberta or Libratus, developed at Carnegie Mellon University and deployed at the Pittsburgh Supercomputing Center, is up for debate. But nevertheless, machines have triumphed over humans in a game in which—unlike chess, where all the pieces are visible to both players—each player has limited knowledge.
Devin Powell at The Washington Post quotes Noam Brown, a graduate student at Carnegie Mellon University, as saying, “AIs have a lot of trouble with poker. It’s the holy grail of imperfect information games.” Brown developed Libratus along with CMU computer scientist Tuomas Sandholm.
Joshua Brustein at Bloomberg has details on the Libratus competition. “Four of the best professional poker players in the world spent most of January holed up at the Rivers Casino in Pittsburgh, losing,” he writes. “They’d show up before 11 am, wearing sweatpants and stylish sneakers, and sit down in front of computer screens. Each of them was supposed to play 1,500 hands of heads-up no limit Texas Hold ‘Em online before they could go back to the hotel for the night.”
Oliver Roeder at FiveThirtyEight explains that the four humans and the machine began each of the 120,000 hands with $20,000 in play money. At the end of the 20-day tournament, ,the humans had lost $1.8 million.
The good news for skilled human players is that they are unlikely to be playing Libratus for money soon in the real world. Powell quotes Brown as saying that running Libratus would cost millions of dollars per year, which potential winnings wouldn’t cover. Nevertheless, Brustein warns that online bots might already be able to exploit mediocre players.
Of course, the poker-winning software may have other applications. Roeder quotes Sandholm as saying, “The AI’s algorithms are not for poker: they are game-independent”—and may be applied to “negotiation, cybersecurity, military setting, auctions, finance, strategic pricing, as well as steering evolution and biological adaptation.”
