Futurology

It's also the nature of the accidents. There are human-causes of accidents that we understand because we are human. We can punish irresponsible drivers more harshly than those who just had a freak accident.

Automated systems on the other hand will fail in completely unexpected non-human ways. We will look at the circumstances of a collision and say something like "it was completely clear [to a human] that the pedestrian was crossing the road, how could the car not see them?" and this will fuel a contempt of the automated car for being incompetent in ways which should disqualify it from driving, as an incompetent human would be, even if the car has a fraction of the accident rate.

We can drive defensively by predicting the mistakes or bad behaviours of other (human) drivers. But when there are drivers on the road that are completely unpredictable and make mistakes in unexpected ways, it makes all of us less safe, and less able to drive safely.

It is a question of legality more than the harm itself. When a driver kills someone, someone is responsible. When an AI kills someone, I'll be damned if the car company will accept liability.

I guess the public is an unfortunate loss they're willing to accept.

Tesla doesn’t use LIDAR, and their lane assist (FSD) is a clear, proven threat to the public.

“Close to one million people lose their lives every year to car accidents. If a technology company builds a vehicle that kills one person every year, that’s one-millionth of the difference, but it will have trouble to survive,”

You mean, like, a train? A tram? A bicycle, even?

The alternative is government to legislate slowly adding more tech to normal cars to turn them into nanny vehicles watching and correcting every move you make.