Brains vs Artificial Intelligence: Libratus defeats human brain in poker

Libratus the undefeated AI has a "total speed is 1.35 petaflops, about 7,250 times as fast as a high-end laptop.

ai poker

We have all seen those sci-fi movies and in the end we have all been left stupefied by the question. Will Artificial intelligence (Ai) one day become so smart that it will overrun humanity itself? Stephen Hawking, the world renowned English theoretical physicist and cosmologist, rang the warning bell when he said that 'Artificial intelligence has the potential to evolve faster than the human race' and 'Once machines reach a critical stage of being able to evolve themselves we cannot predict whether their goals will be the same as ours.'

Maybe the days of a complete robot uprising still is more of a probability than a certainty. But in what is perhaps an indication of things, an AI (Artificial intelligence) built by two Carnegie Mellon researchers thrashed four of the world's top players in a nail biting 20-day poker tournament. Many will argue that we have seen this all before, computers have defeated the best minds out there at chess.

Well, this is different. Poker as a game is hugely different from chess. In poker unlike chess you can't see the cards that your opponent is holding. So there is a lot of intuition involved to counter and implement bluffing, slow play, and other tactics.

The tournament dubbed Brains vs Artificial Intelligence had four human players – Dong Kim, Jason Les, Jimmy Chou and Daniel McAulay – who devoted 11 hours every day in the Rivers Casino in Pittsburgh to pit it all out against Libratus – a piece of software developed by researchers from Carnegie Mellon University (CMU) at no-limit Texas Hold'em, a two-player unlimited form of poker.

Libratus was powered by the Pittsburgh Supercomputing Center's Bridges computer and has a "total speed of 1.35 petaflops, about 7,250 times as fast as a high-end laptop and its memory is 274 Terabytes, about 17,500 as much as you'd get in that laptop."

It was computed in such a way that it could learn on his own, "We didn't tell Libratus how to play poker. We gave it the rules of poker and said 'learn on your own'," said Carnegie Mellon University professor of computer science Tuomas Sandholm who, along with his PhD student Noam Brown, built Libratus. The bot might have started playing without any clear plan of action but over the course of playing trillions of hands it was able to refine its approach and arrive at a winning strategy.

"They put up the best fight they could," said Brown referring to the human players. But the Ai ultimately won and took home the prize.

This article was first published on February 1, 2017
Related topics : Artificial intelligence