Blog

DeepMind company which develops a neural network AlphaGo has recently announced that it was preparing to teach artificial intelligence to play poker. However, Johannes Heinrich and David Silvera, project managers, are aimed to teach the bot to think like a man using neural networks. Admittedly, the company has already had positive experience, they have created a program that had succeeded to win four of five games against the Korean gou-professional Lee Sedol.
However, gou isn't a poker. Poker is a game of incomplete information where in contrast to the curly game bot doesn't see the the opponent's position and can't create its strategy based on this. Moreover, people could make unconventional and controversial moves that will stand out from the general logic of the draw (from the perspective of the bot) using their knowledge, intelligence and psychology. Now scientists are going to storm the last bastion of human benefits - psychology. A new bot DeepMind will be trained on it.