“Will I dream?” asks HAL 9000 in 2010. It turns out that computers do indeed dream after all, and what they imagine is beautiful, surreal and unlike anything anybody could ever foresee. Unless, of course, you are Philip K. Dick, who wondered if androids dream of electric sheep. They do.
Software engineers at Google tested the company’s servers to see if they were able to recognize commonplace objects, and because of how Google’s “neural pathways” operate, layering in pattern upon pattern to form an image, when that process is reversed, we get a glimpse of the AI’s “subconscious.”
“The Google artificial neural network is like a computer brain, inspired by the central nervous system of animals,” reports IFL Science. “When the engineers feed the network an image, the first layer of ‘neurons’ have a look at it. This layer then ‘talks’ to the next layer, which then has a go at processing the image.” Google’s engineers then “gave artificial neural network an object and asked it to create an image of that object. The computer then tries to associate it with specific features.”
Because of how Google’s search engines work, it seldom thinks of objects in isolation: asking it to create a dumbbell, for example, produces images of weights being lifted by a muscular arm, because that is most often how Google’s image search sees dumbbells. “To the computer, one of the defining features of a dumbbell is the arm that lifts it.”
Google’s neural network sees images in objects that would put Salvador Dali to shame: surreal and strangely beautiful artworks that exist by a weird internal logic. When designers fed the servers white noise, Google “pulled out patterns from the noise and created dreamscapes: pictures that could only come from an infinite imagination.”