Tech News: Meet the New AI Challenging Human Poker Pros



Here’s another recent update from the tech world!

Meet the New AI Challenging Human Poker Pros

In Texas Hold’em, there may be up to 10^160 possible plays at some point in the game. With so many possible game outcomes at a given point, AIs in the past faced difficulties in solving all outcomes in advance to determine the best move. To address this problem, Sandholm and Noam Brown have created Libratus algorithm to prune certain paths over time and create a smaller game tree, speeding up the processing times. This will allow the AIs to take a perfectly balanced approach to the game, including methods of bluffing and unusual strategies. To this current date, AIs have not been able to beat human pro poker players. Beginning tomorrow, a match of Texas Hold’em will be held between human players and AI.

Read the full article at IEEE Spectrum