> While sad, mile for mile FSD is better than humans.
For this statement to be correct, we’d need to have full disclosure of all travel using FSD at any point, accidents which happened anytime FSD was active or had been recently deactivated (for example, that guy who fell asleep counts even if FSD deactivated a minute before the vehicle crashed), and be able to compare that to the same trips driven by human drivers. You especially need to avoid including incidents in the human stats which are in conditions where FSD would do even worse.
For this statement to be correct, we’d need to have full disclosure of all travel using FSD at any point, accidents which happened anytime FSD was active or had been recently deactivated (for example, that guy who fell asleep counts even if FSD deactivated a minute before the vehicle crashed), and be able to compare that to the same trips driven by human drivers. You especially need to avoid including incidents in the human stats which are in conditions where FSD would do even worse.