I agree with your intention, though isn't it inevitable self driving cars will kill people even if they are 1000x safer than humans?

If a CEO has many millions of cars running his software, their will always be a death.

In that case, nobody would want to take on that responsibility so nobody would build self driving cars.

Therefore humans will still drive cars and there will be 1000x more deaths, most preventable with self driving cars.

This is pretty much exactly the Trolley problem in ethics and philosophy. Read up on it :)

They don't need to be 1000x safer than humans. If they are 2x safer than humans, that is plenty for me to support using them. The issue is determining how safe they in fact are, to appropriately high confidence, before putting people at risk.

That is something that "tech" companies will in all likelihood never be able to do, since they seem uniquely designed to betray and lose the public's trust at the earliest opportunity.

"Tech" companies have become far too culturally accustomed to dishonest hype, "growth hacking", marketing BS, PR spin, etc. for their statements to be trusted, especially when there are life-and-death implications.

It's not always true. Waymo is a clear exception: their car was at current Tesla's level 5 years ago, but they chose not to run into production. Many Google engineers left from the team because they wanted to see their work in public, but Larry clearly wanted safe L4 driving before getting to the roads.

(of course Google did lots of ,,ungoogly'' things in the past few years, but at least they got Waymo right)

It's not the trolley problem because as yet we don't know how many people would be killed by rolling out self driving cars and, more to the point, the speed of rollout and safety are not necessarily inversely proportional.

By characterising self driving cars as a "trolley problem" you have presented what is known as a "false dilemma". Maybe you could read up on that ;)

Good point! Are there any publicly accessible crash/death statistics for cars with a similar price/safety/demographic profile as Teslas?

I feel that is the only way we could isolate for self driving and not other factors.

If individual engineers and executives were made to be legally responsible for potential fatal accidents resulting from technology, progress would grind to a halt.

I dont think that's the case. However, if the people who build it refuse to take responsibility for their defects killing people then I think that's the clearest possible signal that the technology isn't viable.

Obviously this would also mean that compensation would have to go up for self driving car engineers and execs to compensate for the risk. That's fair, and, if that cost makes a self driving car venture uneconomic - again, not viable. And that's okay.

People take responsibility for other people's lives every day when they get in a car. I don't think the idea that Uber engineers should do the same should be considered particularly controversial.

I mean, in cases where they are criminally negligent, they are already responsible for accidents.

The tricky part is figuring out what constitutes negligence.