I think the 'how do we decide' is when we can show autonomous driving is significant safer than humans. I suspect the problem will then be computers avoid many

I think the 'how do we decide' is when we can show autonomous driving is significant safer than humans. I suspect the problem will then be computers avoid many accidents humans make. But make mistakes that seem obvious to humans. So even though the overall risk is lower it becomes very hard to accept the computer did something wrong vs admitting a person was speeding and they stuffed up killing themself. To allow this tech to exist we might need legislation that autonomous car companies can't be sued unless there is clear malpractice in their systems vs some level of 'accidents happen'. In the same way a wife doesnt typically don't sue her husband that caused an accident that killed themself.