AI Technology and Online Poker
Back in July, the latest AI developed by a team from Facebook and Carnegie Mellon University was tested against poker professionals. Out of more than 20,000 games of online poker six-player, No Limit Texas Hold’em (the most popular variety of the game), the AI – ‘Pluribus’ – successfully defeated 15 of the world’s top poker players, all of whom have winnings of more than $1 million from playing the game.
The report stated that Pluribus was created over eight days at a cost of $144. The authors of the report, Noam Brown and Tuomas Sandholm, commented: “Poker has in general been especially difficult for AI given the computer doesn’t know the cards their opponents have, and it involves a level of deceit or bluffing. Pluribus’ approach is a milestone in AI development and for poker, which could, be adapted “to basically any other form of poker” including, say, nine player variants or tournaments.”
How does it work?
Pluribus put in the hours, learning how to play from scratch. Its progression was considerably faster than human progression given it has the capacity to play a hand in 20 seconds – half the time of a human player. It uses a technique of ‘abstraction’, which involves slightly different hands or bets being treated as the same thing to reduce the complexity of the game.
What potential problems could this cause for the online poker industry?
The use of software to help players of online poker is nothing new. In fact, many professional players have commented that bots are easy to spot, and in such instances, can be an easy foe to defeat. However, the same cannot be said for the average amateur, playing online poker for enjoyment, rather than as a career choice. Should these amateur players stop seeing online poker as a game of skill, the number of players could drop off. Also, earnings of professional players will likely suffer as a result of bad players using more advanced software in order to win. However, there are also advantages for players willing to learn from the strategies of the bots. One example includes ‘donk betting’, which Pluribus embraced, and which involves using wildly different bet sizes with each hand.
What happens now?
Any business that is based around changing human behaviour on the internet has the potential to be impacted by software of this kind. Namely, software that replicates human behaviour. The effects of which are expected to be felt across all industries, if and when they are implemented on a mass scale. Given how advanced AI could shake up the online poker industry, it’s no surprise industry giants are already establishing ways to safeguard their communities against these bots, with specialist teams dedicated to identifying the bots and kicking them out.
Image credit: AI Technology via Phonlamai Photo/Shutterstock