Genius AI Steals $290,000 In A Chinese Poker Competition, Defeats World Series Winner
5 months ago 0
Short Bytes: Lengpudashi, an AI system developed by a research duo at the Carnegie Mellon University in Pittsburg managed to defeat a 6-membered ‘Team Dragon’ poker team included Yue Du, who won the World Series of Poker in January this year. The prize money of $290,000 was awarded to the winner.
Hearing stories about AIs challenging human minds isn’t a big deal anymore. According to a Bloomberg report, an AI system named Lengpudashi – housed at the Pittsburg Supercomputing Center – tasted success after defeating six poker players in a five-day competition held in China. It involved a total of 36,000 hands between the AI and the team over the span of five days.
A prize money of $290,000 was awarded to Lengpudashi. The opponent human team including engineers, computer scientists, and investors was led by a world series winner Yue Du. The team tried to leverage their know-how of machine learning and game theory during the gameplay but things didn’t turn out in their favor.
Lengpudashi is a brainchild of Tuomas Sandholm, a computer science professor at Carnegie Melon University, and a Ph.D. student Noam Brown. This research duo was also behind another artificial intelligence system called Libratus ( a predecessor to Lengpudashi) which won a 20-day poker competition earlier this year.
The price money earned from the poker competition be used to fund their firm Strategic Machine.
Artificial intelligence systems have proved their excellence in other games including Chess and Go. However, the case of poker is a little bit different, it’s often tagged by computer scientists as an “imperfect information game”. Most of the game is based on assumptions. Contrary to Chess where all the pieces are visible on the board.
Poker involves complex betting strategies. Also, one has to depend on their bluffing abilities and should be able to sense if the opponent is bluffing. “People think bluffing is very human – It turn’s out that’s not true,” Noam Brown told Bloomberg.
Making AI systems defeat humans isn’t something that should be presumed as a demonstration of superiority. According to Sandholm, it’s a way of sharpening strategic decision-making abilities which can be beneficial in other areas such as business, cyber security, etc.
If you have something to add, drop your thoughts and feedback.