AI and Poker, here we are…

Facebook researcher Noam Brown made an AI that is able to beat professional poker players in a multi-pot game.

This is fascinating.

In March 2016, Google team AlphaGo defeated the world champion in Korea in the game of Go. That was an important accomplishment because Go has multiple orders of degrees of freedom over Chess. This means brute force computing wouldnt have been able to solve that problem easily, i.e. we would need to throw far more computing at it than was used.

Playing poker is yet another significant milestone. Poker is a game of incomplete information. A player doesnt know what another player holds. This is unlike Go or Chess where each player knows the layout of the board. So from an uncertainty point of view, Poker is complex. A multi-pot game is even more complex.

AI being able to play a game of incomplete information is important because that is what real life is – full of uncertainty and hidden information. Investing is another one of those games.

Leave a Reply