Computers, gaming
You can teach a computer to play games, but better that it teach itself
HOW good are computers at learning to play computer games? The chart shows the performance of a machine using artificial intelligence to play a selection of classic video games, compared with that of a professional human tester.
The work—presented in Nature—is more than just a bit of fun. It demonstrates the power of modern “machine learning” techniques, which allow machines to figure things out for themselves. The computer was not given any information about the rules of the games. Instead, like a human, it inferred how to play by watching the screen and the score, figuring out concepts like “alien” and “power up” for itself. Such self-teaching algorithms are a hot topic for computing firms, which use them for everything from facial recognition to targeted advertising. Google is particularly keen: it bought Deepmind, the company behind the work, for a rumoured price of $400m last year.
This article appeared in the Science & technology section of the print edition under the headline "Computers, gaming"
More from Science & technology
Many mental-health conditions have bodily triggers
Psychiatrists are at long last starting to connect the dots
Climate change is slowing Earth’s rotation
This simplifies things for the world’s timekeepers
Memorable images make time pass more slowly
The effect could give our brains longer to process information