Artificial Intelligence In Online Poker Games

Artificial Intelligence developers are constantly making improvements and developing the AI technology that they already have. AI is well known for mastering a huge range of games, from chess to Go, and more recently, even poker! Every single year, more and more records are broken using AI systems against humans, and it wasn’t until recently that the incredible happened and an AI system beat not just one, but four of the world’s best poker players.

This may not seem like such a big achievement when so many other types of games have been won by AI systems before, but poker is different. When you deal with poker – probably the greatest of the real money skill games – you deal with lots of imperfect information, making it very complicated for a machine to understand.

So in poker, the machine not only has to think strategically about how to win the game, but it also has to deal with risk management, deception on the part of the human, unreliable information, and maybe even bluffing itself.

All interesting stuff, so let’s take a look at some information provided by Pokersites with their timeline and a few facts on the incredible development of poker AI:

Humans are slowly starting to trust AI, especially that of millennials, 62% of people between the ages of 17-24 trust certain AI systems, compared to only 34% of people over the age of 55. That doesn’t mean, however, that those that don’t trust AI systems don’t believe that they could benefit the human race in the future.

The very first poker AI system was developed back in 1984. Mike Caro, the system’s creator named the software Orac and he used it to compete in WSOP. Compared to the systems that are available today, Orac was relatively basic, but it was the system that set the wheels in motions for the development of further poker.

In 1991, the University of Alberta team began their work on their poker AI research. It was not until later however that they realized that their work would be one of the biggest developments in poker AI.

Loki was developed in 1997 and has had a huge impact on the poker world. The AI bot was designed to play a full table, and still to this day, a win of this level is yet to be achieved.

2003 was a big year in terms of the development of poker AI, as developers started to steer away from the Chess model. It was this shift that leads to the development of some very successful AI.

This year marked the most dominant performance of an AI system versus a human in poker.

Polaris was launched in 2007 and is to this day one of the most famous poker bots, mainly thanks to its match against Phil Laak. Although the bot lost, the match was certainly very close. Polaris is a different kind of bot to what was formally developed as it consists of lots of bots all working together. In 2008 the system was up against six humans and finished with a record of three wins, two losses, and a tie.

After many years of development, the University of Alberta released Capheus, a bot that has now solved heads-up Limit Hold’em. The bot mastered the game by playing itself for two months and came from a long line of other famous AI bots.

In 2017, The University of Alberta released Deepstack AI which used a new algorithm that learns from self-play. This is a particularly notable AI bot due to the fact that it was able to become a winning poker player with no training from experts.

The AI system names Libratus managed to beat a human team of poker champions, a benchmark that has arguably been a long time coming. The bot not only won, but it simply destroyed its opponents. Libratus would analyze its own play every day in order to correct the mistakes that it formally made.

 

As the timeline shows, almost every year has seen some huge developments in AI technology. Researchers are still working on developing the technology today in order to improve poker bots. This artificial intelligence could have a big impact on other any other industries, so Poker AI has an exciting future ahead!