Re: Crash Test Dummies.
Why does it need to be 10x safer? 2x safer is a huge improvement.
What I haven't seen is the number of Tesla autopilot miles that have been racked up compared to the number of deaths (3 now?) compared to the number of manual pilot miles that have been driven compared to the number of deaths in a country (probably the US at this stage) from that driving, as this will give us some good, hard data on miles per death/crash ratio.
Although, something that I've never seen brought up in these discussions is how computers think* differently to humans. When a Tesla crashes on autopilot, it's always seen as something utterly stupid that any human driver would have avoided but what we'll never get data on is the number of crashes that didn't happen because it turns out that computers are much better at avoiding that kind of accident than people.