Essentially that’s what it is.  A company acquired by Google developed a fairly general “AI” that can play a series of Atari games.  Nature published a video interview with the researchers and it’s pretty interesting.  Turns out [3:40] it’s a neural network that takes input as what’s on the game screen and then looks at what action (what button to push next) will help to maximize its score.

The problem is that it isn’t generalizable enough to master games that need more time for feedback.  Interesting comments on a Reddit Thread.  Lastly, I initially thought that this was just another application of a genetic algorithm but it was, as mentioned before, a neural network focused on Reinforcement Learning (Stanford).

Here’s the full video from Nature: