Going deeper: artificial intelligence

　　It's been a big year for artificial intelligence, and in particular for a technique called deep learning, which already powers search engines, smartphone voice-assistants, translation systems and self-driving cars.

　　Its potential became apparent in 2012, when it led to much-improved accuracy in an annual contest called ImageNet;

　　last year deep-learning systems outperformed humans at image classification for the first time, and in 2016 they did even better.

　　During the year such systems also demonstrated superhuman performance in lip-reading and speech transcription.

　　Most prominently, in March a deep-learning-powered system called AlphaGo defeated Lee Sedol, one of the world's best players of Go—a board game so complex that computers had not been expected to master it for another decade.

　　Deep learning is still terrible at writing pop songs or stories, but expect the cat