In the world of poker, players bet on odds based on cards, reading the other players at the table. With the recent growth of AI technology and computers adapted to play the game, the potential of reading an opponent is not possible. AI, short for Artificial Intelligence, has been making huge strides of late to catch up to the human mind specifically in the poker world. These new supercomputers that can think, reason and apply strategy are now some of the best poker players in the world. According to the extensive data on poker AI, found at PokerSites
, these three AI's have become the leaders in the AI poker revolution.
Created by scientists at the Carnegie Mellon University, Claudico is a Texas hold-em poker playing AI. The project chaired by chief scientist, Tuomas Sandholm, was released in 2015, after years of research and testing. In its big casino debut, Claudico was tested against four of the world's best poker players. Doug Polk, Bjorn Li, Dong Kim and Jason Les took on the AI, at the Rivers Casino, for a friendly testing of the AI's capabilities.
During the course of the game, 170 million dollars in virtual money was bet over a 13-day period. In total, 80,000 hands were played, 20,000 for each human player. The four poker pros were able to defeat the AI, for total combined winnings of $732,713. However, they had to work in tandem to defeat the AI. Though they were able to defeat the AI, onlookers agree that, in a single head to head match-up, the AI would have likely beat all the pros.
In 2017, the University of Alberta released their No-Limit Texas hold'em playing AI, named â??Deepstack." Deepstack, the brainchild of Micheal Bowling, was tested against dozens of other players and played a total of 44,000 hands of poker. Over the course of this testing, Deepstack averaged a win rate approximately nine times higher than the average professional poker player. Deepstack employs deep neural networks to emulate human intuition and learn on the go. The AI was able to reduce game situations to 10 to the 14th scenarios, where 2 human players would have 10 to the 160th possible scenarios, proving Deepstack could simplify the game of poker much easier than a human player.
In 2017, Libratus, a head to head poker-playing AI, delivered the most decisive blow to human competition it faced. Liberatus, created at Carnigie Mellon University in Pittsburg by scientist Noam Brown, is the most advanced AI to date. Against four poker pros, Libratus came out as the clear winner, defeating all four at a rate of 14.71 USD per hand over 120,000 hands played. All four human players were defeated, and the total winning of Liberatus in real cash would have equated to over 1.7 million dollars. Liberatus is a self-learning AI and spends its nights going over gameplay and adjusting its strategies to continually become better, making it a true poker AI and likely the best poker player on the planet.
These are just three of the poker AI's under development. There are a few notable others, but these three have been vigorously tested, as opposed to their counterparts. All poker and gambling sites worldwide have banned the use of AI after witnessing the results of these early tests. The truth is, who can blame them? The AIs are capable of out-classing the human players with ease and look to continue to become even better in the years to come!
Source: HOB Team