• Home
  • The Poker AI Boom

The Poker AI Boom

Poker is a game where you bet that your cards are better than those of your opponents. Winning in poker is a matter of extracting the most value from your winning hands and minimising losses when you have a losing hand – something known as MinMax. It is not always the best player that wins, however. Sometimes a player’s tenacity and courage triumph over those with the best poker cards.

Like chess, poker involves imperfect information: each player knows their own cards but not those of their opponents. This makes it harder to model computationally and is why the poker AI boom didn’t happen until 2015, when computer scientists announced an algorithm that displayed essentially perfect play for two players in a limited version of the game with restricted bet sizes.

Since then, there have been a spate of commercially available poker solvers and in the space of four years we’ve had the first superhuman AI for multiplayer Texas Hold’em. In addition to confirming some of the folk wisdom on optimal poker strategy, these AIs have also overturned many maxims that expert players had gotten wrong, such as that it is never correct to ‘donk’ (initiate the first bet in a betting round) if you have the best possible hand.

In this episode Annie and Tyler explore how payoffs aren’t always monetary, the benefits and costs of probabilistic thinking, the magic thinking behind why people buy fire insurance but usually don’t get prenups, whether Trump has a tell, the number one trait of top poker players and more.