Another point might be attribution. It's horrible when a driver is tired, drunk or simply misses something - but on a human level, you can understand what's goi

Another point might be attribution. It's horrible when a driver is tired, drunk or simply misses something - but on a human level, you can understand what's going on. Also, you know the guy at fault and that he will be punished (at least, somewhat). An autonomous vehicle, in comparison, is not relatable at all. Some equation delivered some wrong result, maybe it was something in the training set, so that's why the car suddenly drove into you. Bad luck. If you're lucky, you get some damage payments from the car company, if you happen to have enough money for a very good lawyer. Sure, as a technical person, I understand some of the possible errors, but it still feels like a complex system that one cannot relate too with hardly understandable edge case behaviour. That's probably why we're so much more lenient with humans.