The psychologist-turned-author-turned poker player has a discussion with the creator of Libratus
In a piece written by Russian-American writer, psychologist and poker player Maria Konnikova, she shares what she learned from an interview with Tuomas Sandholm, a computer scientist at Carnegie Mellon University, regarding poker and artificial intelligence (AI) and its limits. As a game of skill but also chance, a few scientists in the past have become interested in poker, as it shows the perfect model for human decision-making. This requires players to find a balance between skill and chance in every choice – that’s what game theorist John von Neumann thought. Sandholm has no knowledge of poker nor is he a fan, but he is interested in the game for the same reason Neumann was.
“Poker is the main benchmark and challenge program for games of imperfect information,” Sandholm explains to Konnikova, per her paper, “What Poker Can Tell Us About the Limits of A.I.” This scientist is the one behind the creation of Claudico, Libratus, and, most recently, Pluribus, which are AI-based robots designed to play poker against humans. The goal is not to beat players at poker, but to create algorithms that are effective for the decision-making process that involves playing poker – a game of imperfect information – and apply it to other scenarios with similar situations like military, business, government and cybersecurity, among others.
Claudico, the first attempt, was basically beaten by human poker players easily. Libratus, on the other hand, was able to defeat US-based pro poker players in heads-up matches. Libratus is actually capable of playing in situations that present unknown probabilities. One of the aspects of playing poker is that, in real-life applications, every possible option is weighed, which a poker algorithm doesn’t do. In poker, this process is a lot simpler, as it only focuses on maximizing the value, which for humans is more difficult to extract from given situations, so these robots learn to make decision based on maximizing the value.
“Logic is good, and the AI is much better at strategic reasoning than humans can ever be,” he explained.
“It’s taking out irrationality, emotionality. And it’s fairer. If you have an AI on your side, it can lift non-experts to the level of experts. Naive negotiators will suddenly have a better weapon. We can start to close off the digital divide.” In the future, what AI learns from poker algorithms can assist humans in the process of decision making in front of certain situations that can be easily influenced by human emotions.