I stand corrected. I will say I was speaking about the ratio of miles per dead human, and not about crashes, but I definitely take your point and now understand that the amount of human interference in autonomous cars is still significant. This is what needs sorting out, for me, regarding this specific incident. from the Scientific American article you quoted - What happened? Was this guy not paying attention? Was he trusting too much in the car? Was he acting like a passenger, and not a driver? In some ways, i consider this person to be responsible for Herzberg's death, and not the AI program. It's clear that Uber knows that its AI still needs to be babysat, and they had a babysitter. The babysitter, however, didn't do his job. Taking that metaphor further, if the "baby", the AI, managed to injure someone else (knock over a lamp which hit's someone, let's say), what would be the first question asked? "Why weren't you watching them?"Utterly, completely and irredeemably false.
An Uber employee was in the Volvo XC90 SUV acting as a safety operator but told police he did not have time to react to avoid hitting Herzberg