Since the appearance of artificial intelligence, poker has been one of the interesting topics that experts wanted to solve. Unlike chess and go, poker is a zero-sum game with hidden information. This makes it more complicated to solve than chess and go. This can be confirmed by the time when AI first managed to beat professional human player. IBM developed supercomputer Deep Blue defeated chess world champion Garry Kasparov in 1997. AlphaGo, developed by Google subsidiary DeepMind, has beaten professional Go player in 2015. It was until 2017 that AI has beaten professional poker player in a 1-versus-1 format. In 2019, Facebook AI and researchers from Carnegie Melon University have made further development and created Pluribus. It managed to beat 5 other professional poker players in a 6-handed poker game, which is much more complex than the heads-up format.
This shows the advancement of artificial intelligence in history. It is able to surpass human ability in more and more complex game. The defeated poker players said that Pluribus was a tough opponent because of it was capable of putting its opponent in difficult spot due to its unconventional strategy. Here is a video of some interesting hands played by Pluribus. People with some poker experience and knowledge will find the actions of it unorthodox but effective.
Beating human players in a 6-handed poker game demonstrates the ability of AI to handle complicated situation with hidden information. To poker players, it provided new insights for their approaches towards the game. There are still undiscovered strategies in the game and the development of AI accelerates the evolution of poker.
Not only poker players are benefited from the insights brought by the AI, but they also mean a lot to the business world. Since chess and Go are games with one opponent and complete information, strategies for them are less applicable to the real world. However, the fact that AI can handle incomplete information multi-player game situations means that it can be applied to a lot of different fields. For example, it can be applied to study fraud prevention and cyber security issues.
From different games that AI has advance to solve, I think that the potential of AI is still to be harnessed. We are still a long way to reaching its limit. I am looking forward to seeing AI tackling some more difficult problems and bring positive impact to the world. What implications do you think this will bring? Where do you think the limit of AI will be?
References
https://www.pokernews.com/news/2017/10/artificial-intelligence-poker-history-implications-29117.htm
