A computer has finally beaten a human at the ancient board game Go – and rather dramatically. Developed by Google DeepMind in London, the program AlphaGo uses deep neural networks to get around the traditional problems of a simple search due to the size of the search space. The network training was from games with human experts and from random games of self-play, so in a sense there was a bit of human in the software, which some might find consoling because the score wasn’t close: AlphaGo 5, European Go champion 0. Expectations had been that this would not happen for at least another decade.