Teslas data doesn't prove anything. Comparing the accident rate with and without using autopilot might have told us something, but Tesla hasn't actually shared any such data. Even if they had, there's every reason to believe that it would be an unfair comparison since the autopilot isn't equally likely to be engaged under all conditions. What Tesla has said is that the number of miles on average between accidents is higher for Autopilot than it is for THE ENTIRE CAR PARK. You simply ASSUME - implicitly - that (a) the average Tesla owner is equally likely to crash as all other owners - including the poor, uneducated, young, jobless, drug-abusers and many other high-risk groups surely underrepresented in the Tesla owner population (b) Autopilot is engaged in comparable driving conditions (to non-autopiloting, I.e. ALL comditions!), so the data isn't skewed by the pilot being used more in conditions such as highways with a surface in good condition (and thus clear lane markings), in good weather with good visibility and so on. I'm NOT saying the autopilot is more dangerous, only that it is way too simplistic to claim Teslas data, at least anything published, gets close to "proving" anything at all!