Scientists from Carnegie Mellon University (CMU) developed an artificial intelligence computer program, Libratus. It managed to beat world’s best and famous poker players. The AI won $1.8 million after playing 120,000 hands at Heads up no-limit Texas hold’em. This has since been considered a groundbreaking achievement in AI game playing ability.
Poker Player Bowed Down To Libratus
Libratus competed with four human poker players over a span of 20 days. And every day, it thrashed them repeatedly at every hand of poker. Jason Les, the player who stood last, described the experience as “demoralizing”. Nevertheless, he did not hold back from giving Libratus its due credit.
“We make a living trying to find vulnerabilities and strategies – that’s what we do every day when we play heads up no-limit. So if the public had any doubt about the quality of this technology, I can tell you from our experience, we tried everything we could but it was too strong,” Business Insider quoted Les as saying. Even though Les managed to seize his beginner’s luck and managed to win $49,072 initially, it was not enough. Despite comparing notes with other players and analyzing the strategies of Libratus, Les ended up on the losing side.
No Luck, Yet No Luck!
Eerie as it might be, Libratus soon learned to adopt new means of attack as the game progressed. After eight days, it became unbeatable, much to the frustration of the poker pros. And one cannot attribute the huge win for the computer program on just a matter of luck. This was because the game was adjusted to minimize the factor of luck.
The 4 human players were eventually split up into two teams. One team was locked away in a room, without any communication to the outside world. The second team was allowed to operate out in the open and had the enjoyed the luxury of researching on the internet and talking to their peers. However, it was observed eventually, that added advantage did not make much of a difference for the second team. It also ended up losing big to the mighty Libratus.
“Heads up no-limit Texas hold’em remained elusive in that never before has it been possible to beat the absolute top no limit Texas hold’em professionals. And in this event, this actually happened. So this was a landmark really in AI game playing,” said Tuomas Sandholm, co-creator of Libratus and a machine-learning professor at Carnegie Mellon University, reported The Register. Nevertheless, it wasn’t a total loss for the poker pros. They got to walk away with a portion of $200,000, depending on how well they played.