A few weeks ago, Google released its DeepDream project. Basically, it is a visualization tool to show how artificial intelligence (in plain language: a robot) sees our world. They feed the A.I. millions and millions of different pictures of something they want it to learn to recognize, until it does. Obviously, this isn’t perfect yet: it’s simply impossible to feed it millions and millions of pictures of everything, which makes the A.I. want to see things that are not actually there. But how does this bias affect the way the A.I. would see things? Well, that is what DeepDream visualizes. The result? Some psychedelic, and often beautiful pictures. The application isn’t available in a very user-friendly version yet, but Google has made the source code available for the public, so that is bound to be available at some point. The video underneath shows you what happens when every single frame of a scene out of Fear and Loathing in Las Vegas is run true the process. Some marvelous examples are available here, here and here (porn! :-)).
Deep Dreaming Fear & Loathing in Las Vegas: the Great San Francisco Acid Wave