@donaldboyd the title is misleading and should instead be "the average human driver" because otherwise for most human drivers who died in automobile crashes, Fsd is already ahead.
@NicoTerry Statistically, humans are at 100Mil miles per fatality and that is even unfair because it includes drunk drivers. FSD is not self-driving so far, so we have zero data on it there.
@marktweise You misinterpreted my joke: The title is "a human driver." Some humans die in car crashes. For those drivers, almost any distance traveled by FSD would represent an improvement. A correct title would amortize over all human drivers (but if you wrote "better than all human drivers" most people would take that to mean "better than every human driver" which is a much higher benchmark; average is also too ambiguous, but the mean vs. median question is not well-approached using crash fatality statistics.