Human vs. Computer - Poker style!

If anyone remembers, a few years back there was a chess match where a chess champion took on a computer program called 'Deep blue' as an experiment in Artificial Intelligence (AI). Now a team of scientists took up a harder challenge, getting a program to play a game of 'imperfect knowledge' a.k.a limit holdem.

From the University of Alberta website:

"A team of computer science researchers at the University of Alberta are pitting Polaris, their poker-playing computer program, against two of the best Texas Hold 'em card players in the hemisphere. The purse is $50,000 in the 2,000-hand match between...Phil Laak and Ali Eslami and the Alberta team, led by Jonathan Schaeffer." (Link)

First I must say, I put my chips on Laak and Eslami; not because my lack of faith in AI but for two other reasons: First because humans can change their style which will cramp what the program learned and the predictions it makes. The second reason is a quote by Schaeffer stating: "There is a mathematically optimal rate at which you should bluff". Math is a big part of poker, no doubt, it is impossible to play poker in a high level without understanding pot odds, implied odds etc. However, it is just one tool of many, optimizing you're bluffs depends not only on the maths but on reads, betting pattens, what is showing on the board and much more.

It remains to be seen if the program can really 'intelligently' understand motivations and anticipate random changes in how a player plays, not just relay on maths. I will however follow the story and the game, which will be presented as live updates on the site (linked above).

(In the near future I will also be writing a post about the danger (and opportunities) such programs can present to online poker)

- Random -