4 ms·
The relevant metric should be accident rate per million miles driven. What is the rate with autopilot vs without autopilot.
by d_theorist 2y ago
The relevant metric should be accident rate per million miles driven. What is the rate with autopilot vs without autopilot.
- HPsquared 2y agoNot all "miles driven" are the same. Autopilot is going to be engaged on highway driving which already has a lower accident rate per mile driven. So that biases a simple comparison. Edit: and even within highway miles, autopilot will be enabled more in the "easy" stretches and turned off in roadworks etc.
- FireBeyond 2y agoPrecisely. If these systems aren't or can't be engaged on a Pittsburgh street at night in a whiteout, then I don't much care. Tesla jukes the stats by "Oh, this won't go well, so we're not even going to try", which isn't an option for human drivers, short of "don't drive at all".
- Someone 2y agoOn similar roads, in similar traffic, in similar weather. If you look at fatalities or define “accident” as cases where somebody had to visit a doctor or something like that, you also need to compare to similar cars. Most self-driving cars are relatively new and relatively large, and both decrease those (probably even accident rate because they statistically have newer brakes and tires and have more active safety features) It’s far from trivial to decide whether some program is a better driver than the average human driver if the program gets the newer bigger car and can decide to cop out in cases where we do not expect humans to do so.
- d_theorist 2y agoI agree you need to control for other factors. But I don't think that is particularly beyond the wit of man.
- lurkingmba 2y agoYou can add, "at a similar time of day". Hours after midnight are very dangerous. To my knowledge, Tesla has never put out data on how Autopilot is "safer" on accidents per mile that was close to being apples to apples.