The world's best professional poker players appear to have found their match: An artificial intelligence developed by researchers from the Carnegie Mellon University (CMU).
Humans Versus AI
The AI dubbed Libratus has already accumulated winnings of nearly $800,000 against human poker professionals at the Brain Vs. Artificial Intelligence competition at Rivers Casino in Pittsburgh.
The human players compete to win shares of the $200,000 prize while Liberatus aims to be the first computer program to win in a professional poker tournament.
One Of The Hardest Games For Computers
Many AI researchers consider poker to be among the hardest games for computers to beat humans at.
How AIs fare against human players when performing tasks has long been used as a measure of progress in the field of AI research. Any win for machine intelligence signifies a major milestone. Over the years, artificial intelligence managed to win in games such as chess with newer AIs even able to teach themselves how to play chess using strategic moves just like human players.
What Makes Poker Challenging For Artificial Intelligence
Poker, however, is trickier for AI to master since it involves complicated decisions that a player needs to make based on incomplete information in addition to poker strategies that include bluffs.
An earlier AI dubbed Claudico attempted to challenge human poker pros in 2015, but it was no match for its opponents. It appears though that the field of artificial intelligence has made significant developments since then as it looks like CMU's Libratus is doing well in the poker tournament.
Strides In The Field Of Artificial Intelligence
Libratus still has around 55,000 hands left for the competition but at more than halfway through the game, it already has a profit of $794,392.
The loss of a predecessor of the Liberatus program in two years ago may help explain why several professional poker players initially underestimated the ability of CMU's AI but the players eventually noticed how the AI has progressed.
"I didn't realize how good it was until today. I felt like I was playing against someone who was cheating, like it could see my cards," said Dong Kim, a high-stakes poker player who specializes in no-limit Texas Hold 'Em. "I'm not accusing it of cheating. It was just that good."
How Liberatus Plays Poker
The Carnegie Mellon team that developed Liberatus equipped the AI with algorithms to allow it to analyze the rules of poker and set its strategy. Using the powerful supercomputer Bridges, Liberatus refines its skills at playing poker by sifting through past games including those that were played at the current tournament. Bridges perform calculations in real time during the games helping the AI come up with computer end-game strategies.
"The bot gets better and better every day. It's like a tougher version of us," said poker player Jimmy Chou said. "The first couple of days, we had high hopes. But every time we find a weakness, it learns from us and the weakness disappears the next day."