- Throughout the 1950s, work progressed on building computers that copied the structure of human brains, with lots of interconnected neurons. These “neural networks” could use algorithms to learn. The first example of “machine learning” was in 1958, when a neural network learned to recognize basic shapes. Combining neural networks in multiple layers made more sophisticated “deep learning” possible. These “deep neural networks” could be trained to play games like chess and go. In 1997 a computer called Deep Blue beat world chess champion Garry Kasparov, and in 2016, DeepMind’s AlphaGo beat Lee Sedol, the world champion go player.