What do you and 16,000 computer processors hooked up to Google's secretive machine learning neural network have in common? Well, you and the neural network both have something resembling a human brain, allegedly. Also: Cats. You both spend an inordinate amount of time looking for cats on YouTube.
In their never ending quest to eliminate humans altogether, Google researchers at company's skunkworks lab in Mountain View (cleverly named the "X laboratory" like some kind of secret lair) created one of the largest neural networks designed to think and learn like a real human brain. To see if their doomsday device could learn on its own, researchers from Stanford and Google asked the network to identify a list of 20,000 different items in still images taken from 10 million random YouTube videos. The result: Google's network taught itself to spot a cat without first knowing what a cat even was. The New York Times explains in their breathless, jargon-laden foreshadowing of the downfall of humanity:
The Google brain assembled a dreamlike digital image of a cat by employing a hierarchy of memory locations to successively cull out general features after being exposed to millions of images. The scientists said, however, that it appeared they had developed a cybernetic cousin to what takes place in the brain’s visual cortex.
In other words, Google's computer network put together a fuzzy image of a kitten because it began to recognize and identify cats from their frequent appearances on YouTube. While Google's research team says their neural network is still fairly small compared to a human brain, it's probably just a matter of time before the Google starts using humans for batteries.
Anyhow, here's Keyboard Cat to play us (humanity, that is) off into the post-Skynet Terminator apocalypse: