NEWS

No Game Is Safe: AI Program Thrashes World-Class Poker Pros

Libratus, an artifical intelligence robot, collected $1.5m in chips from the world's best poker players. Thankfully for the humans, it wasn't real money.

02/02/2017 17:23 SAST | Updated 02/02/2017 17:29 SAST
Carnegie Mellon University
Tuomas Sandholm (centre) and PhD student Noam Brown developed Libratus.

After Google's artificial intelligence system beat grandmasters of the world's hardest board game last year, it was only a matter of time till computers started thrashing humans at poker too.

Now, an AI developed at Carnegie Mellon University (CMU) has defeated four world-class players in a game of heads-up, no-limit Texas Hold'em.

By the end of the 20 day match, Libratus had collected $1,766,250 (£1.4m) in chips, a victory researchers said would pave the way for the introduction of AI into a range of business negotiations.

Fortunately for the pros, the chips only represented virtual money. They were really playing for a share of $200,000 (£158,000), which was divided up between the quartet depending on ranking.

Reflecting on his AI's victory, Tuomas Sandholm, a professor of computer science at CMU, said: "The best AI's ability to do strategic reasoning with imperfect information has now surpassed that of the best humans."

In a similar tournament in 2015, the pros claimed victory. But Libratus took a different approach.

"Typically researchers develop algorithms that try to exploit the opponent's weaknesses," Sandholm said. "In contrast, here the daily improvement is about algorithmically fixing holes in our own strategy."

After play finished each day, a meta-algorithm would analyse the holes the players had identified in its own strategy.

"It then prioritized the holes and algorithmically patched the top three using the supercomputer each night," Sandholm added.

Frank Pfenning, head of the Computer Science Department at CMU said the victory could have consequences for business negotiation, military strategy and medical treatment planning.

"The computer can't win at poker if it can't bluff," Pfenning said. "Developing an AI that can do that successfully is a tremendous step forward scientifically and has numerous applications. Imagine that your smartphone will someday be able to negotiate the best price on a new car for you. That's just the beginning."