The distinction between narrow and general artificial intelligence is crucial. Humans are so effective because they have general intelligence: the ability to learn from one situation and apply it to another. Recreating that kind of intelligence in computers could be decades away. Progress, though, is coming. Researchers at DeepMind, a London-based company owned by Google, made what they called “baby steps” towards artificial general intelligence in February when they unveiled a game-playing agent that could learn how to play retro games such as Breakout and Space Invaders and apply the skills to tackle other games.

But Nigel Shadbolt, professor of artificial intelligence at Southampton University, stresses that the hurdles which remain are major ones. “Brilliant scientists and entrepreneurs talk about this as if it’s only two decades away. You really have to be taken on a tour of the algorithms inside these systems to realise how much they are not doing.”

“Can we build systems that are an existential threat? Of course we can. We can inadvertently give them control over parts of our lives and they might do things we don’t expect. But they are not going to do that on their own volition. The danger is not artificial intelligence, it’s natural stupidity.”