misosoup7 writes: Smart programs have been beating humans are various games since the early 90's. The problem is that these programs are specialized and can only do one thing. The Deep Blue chess computer can only play chess, not even tic-tac-toe. The Google DeepMind team now has made a breakthrough against this "narrow" intelligence problem by creating an AI that can perform well across a variety of games. Published in Nature, the team describes their new algorithm. And you can watch the program play Space Invaders and Breakout at the bottom in the attached videos.