Semiotics 2- Google's "Deep Dreams"

It was hard to choose just one of these pictures, check out this gallery if you don't immediately know what is going on here. This was originally a totally normal-looking picture of a beautiful countryside, which was processed through Google's latest artificial intelligence engine specifically designed to identify objects and relations between objects in digital pictures. Google has structured some of its most powerful computers to process and analyze information in ways similar to the neurons in our brain, with both mesmerizing and near diaper-requiring results. Its practical applications have already been proven a few months ago with the release of the new Google Photos app which takes facial and object recognition to insane levels. For example, it found and correctly identified pictures of my 30 year old cousins in their pre-teens with bowl cuts and glasses, and can differentiate and identify any number of categories from animal species to which sport is being practiced, all on its own.  Now all of this awesomeness required extensive machine learning, modeled after traditional human learning patterns, and so some guys decided to show the world what was going on inside this virtual brain.

Hope that's enough background to get on with my semiotic analysis. You can compare the original image to this one in the link I provided above, along with several other examples, and the similarities between each rendered image tickles the curiosity. We see several building related structures and lots of animal-lookin things that seem to have evolved from a nuclear holocaust. Sure it's kinda amusing to look at, but what's more amusing is trying to understand how the computer is trying to understand. Seeing how sophisticated this understanding has become has prompted several industry leaders to prophesy an imminent iRobot-eske doomsday if we don't pull the plug on artificial intelligence. But how biased do you think Hollywood has made us? Do we really think Google will start genetically modifying horrible creatures to wipe out the human race because that is what it is "thinking"? Are some people scared because of the image or are they simply scared of an unknown object they can't seem to identify?

But what is truly groundbreaking here is the notion that this technology is no longer just reproducing and regurgitating information we give it, but that it is starting to creatively produce its own material from a blank slate. Here are some examples of Google's imagination running wild when prompted to identify objects and scenery from white noise and blank space. These "dreams" as the engineers have dubbed it, are even more fascinating. What elements of consciousness will we be able to flesh out and understand with these new tools? Superficially these images all seem like random hodge-podges of nonsense, but there are several layers to what is going on here.