No big deal or anything. Don’t be alarmed. But Google’s secretive computer network simulating the human brain has learned to recognize cats. On its own. With no hints from its mortal creators. AH-HA! So this is how Skynet will begin.
The project, of course, comes out of Google’s clandestine X Labs, the same futuristic outfit responsible for augmented reality on your face and cars that drive themselves. Geeze, they just can’t make humans obsolete fast enough, can they?
Operation First Lets Start with the Kitties has been several years in the making, reports John Markoff at The New York Times. Scientists built one of the largest “neural networks,” mimicking the human brain, by connecting 16,000 computer processors with more than one billion connections. Then, they proceeded to feed it random thumbnails of images from 10 million YouTube videos.
The results in the cat test were about twice as accurate as prior efforts. But what’s perhaps more remarkable, is that unlike most commercial vision technology, which has humans “supervise” (scare quotes his!) the process and label features, GOOG gave its brain no help.
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.
Neuroscientists have discussed the possibility of what they call the “grandmother neuron,” specialized cells in the brain that fire when they are exposed repeatedly or “trained” to recognize a particular face of an individual.
Wait, if cats are the first faces the machine learned to the recognize, does that mean they’ll be spared . . . or targeted?