Earlier this month, Carnegie Mellon University decided to put its Artificial Intelligence (AI) computer called Libratus to the test by challenging four poker pros to a 20 day marathon poker tournament that was promoted as “Brains vs. Artificial Intelligence: Upping the Ante”. The tournament ended on January 30 and in the end it was the AI who completely outclassed the four poker pros.
The tournament was based on a game of heads-up no limit Texas hold’em and over 120,000 hands were played. In the end, it was Libratus who had a combined amount of chips that totaled $1,766,250 and finally proved that AIs do not have ability to outclass and outperform humans. The victory at poker was also a triumph for Carnegie Mellon University professor Tuomas Sandholm who developed Libratus along with Noam Brown, who is pursuing his Ph.D in computer science.
The AI victory over the human brain will also encourage companies to venture further into AI as it can now be used in other fields such as cybersecurity, military strategy, medical treatment and business planning. Frank Pfenning who heads the Carnegie Mellon computer science department stated that when a computer can start successfully communicating with humans, its reach will spread across numerous fields.
In a statement, Pfenning said “The computer can’t win at poker if it can’t bluff. 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.”
The four poker pros who played against Libratus were Jason Les, Daniel McAulay, Jimmy Chou and Dong Kim who will share a $200,000 reward based on how they each performed at the event. Les and McAulay stated that their battle against the AI was a lot tougher than the expected.
Libratus’s gaming strategy was formulated based on the Pittsburgh Supercomputing Center Bridge’s computer which is at the Rivers Casino. The AI was able to base its poker strategy and play all four poker players at the same time and fill in the gaps that were caused due to imperfect information and put together a better strategy than what the players were able to come up with. The biggest difference with the Libratus AI was the fact that it did not focus on the player’s weaknesses but concentrated on finding its own weaknesses and then fixing it.