AI Experts Noam Brown and Tuomas Sandholm on the Future of AI in Poker

Noam Brown and Tuomas Sandholm, renowned AI experts, recently shared their insights on the future of artificial intelligence in poker and beyond. Their paper on solving games with incomplete information was recognized as the best at the annual NIPS-2017 conference. Additionally, they published a paper in the journal Science about the victory of 'Libratus' over professional players.

When asked about the applicability of their methods to 6-max poker, Brown and Sandholm stated that their methods will work in 6-max poker. However, they mentioned that poker is not the most optimal area for solving games with three or more participants. They explained that in games with three or more participants, it is possible to lose even if playing by Nash equilibrium. This is because Nash equilibrium does not guarantee a non-losing result in such games. Additionally, there can be multiple Nash equilibria, making it difficult to ensure that opponents will play by the same equilibrium.

Brown and Sandholm also discussed the potential of AI in other areas. They noted that their methods are well-suited for games with three or more participants. They mentioned that in poker, players often fold, and the more players at the table, the higher the probability of folding. They also stated that there are no team games in poker, and collusion with another player would be a violation of the rules. For all popular poker variants played online, there are already bots with superhuman artificial intelligence. However, an objective competition in 6-max is practically impossible due to the difficulty of controlling collusion between players.