The AI solution has already proven that it can beat some of the best poker players in the industry
Artificial intelligence (AI) may have seemed something more futuristic than realistic just a few short years ago, but it has evolved significantly since then. Teams of experts around the world have been diligently advancing the technology and it is now reaching levels that some could even equate to self-awareness. One of the most notable AI solutions developed, DeepStack, is so far ahead that, in 2016, it proved its worth by teaching itself how to play poker. Not only did it learn, but it beat some of the greatest poker players around.
DeepStack is capable of interpreting “imperfect” information to make better decisions. According to its developers, it “doesn’t need to reason about the entire remainder of the game because it substitutes computation beyond a certain depth with a fast approximate estimate, DeepStack’s ‘intuition’ – a gut feeling of the value of holding any possible private cards in any possible poker situation.”
The results showed how the AI can be used to avoid reasoning and automatically train itself to improve. After DeepStack had completed 44,852 games, its results were ten times greater than what most poker pros considerable a sizable winning margin. It was the first theoretically sound application of heuristic search methods in imperfect games, and revolutionized how the world viewed the capabilities of AI in real-world applications. This has already become the basis for new AI solutions and provided evidence for the technology’s ability to develop its own “neuron” networks to be able to think and solve problems autonomously.