The biggest problem with self-driving cars is not the technology. It’s the humans. The world’s biggest carmakers are pouring billions of dollars into solving the problem of how humans and robocars can share the planet. After several trips in self-driving cars recently, I’m astonished at the strides that have already been made towards that goal — but also daunted by some of the unexpected areas of friction.

First, the achievements: when Ford put its autonomous cars through their paces for a group of journalists in Miami recently, the robocar that I was riding in (like most autonomous cars being tested on US roads today) braked a little too hard at many stops, accelerated just that bit too abruptly and sometimes stopped in the middle of the road, for reasons known only to its robo-brain.

But it also did something extraordinary: it demonstrated a remarkable ability to act like a car driven by a trained, experienced and skilled human driver. (In other words, it drove at least as well as my two teenage learner drivers back home). And in at least one instance, it drove better than the human sitting behind its steering wheel as a “safety driver”.