Google researchers teamed up with Stanford University computer scientists to build a "deep learning" neural network across 16,000 computer processors. The team fed it 10 million images from YouTube videos in the hope that the system would learn on its own—and it worked. As Jeff Dean told the New York Times: “We never told it during the training, ‘This is a cat.’ It basically invented the concept of a cat." And it's not just cat images -- that same neural network architecture is why speech recognition on your smartphone's Google Search app is so accurate.
