3 ms·
The question is how that compares to human drivers. Imagine everyone would have been using autopilot in the past and steering wheels would have been introduced
by ArtTimeInvestor 3y ago
The question is how that compares to human drivers.
Imagine everyone would have been using autopilot in the past and steering wheels would have been introduced just recently. We would see horrifying reports about humans causing deaths by speeding, driving drunk, falling asleep etc. It would seem as if letting humans drive is an insane idea.
But would that really mean we have to return to autopilot immediately?
We don't know until we have some numbers. Like miles driven per fatal crash. For humans and for software.
Wikipedia says:
For driving, the rate was 150 per 10 billion vehicle-miles
https://en.wikipedia.org/wiki/Transportation_safety_in_the_United_States https://en.wikipedia.org/wiki/Transportation_safety_in_the_U...
In Teslas latest investor update, they state that FSD has driven about 1.2 billion miles:
https://digitalassets.tesla.com/tesla-contents/image/upload/IR/TSLA-Q1-2024-Update.pdf https://digitalassets.tesla.com/tesla-contents/image/upload/...
But I think that does not include autopilot?
So I'm not sure if there is enough public data for a comparison.
- hobs 3y agoNo it isn't the question is if the false marketing caused deaths, comparisons to human drivers are still irrelevant at this point.
- falcor84 2y agoI'd argue both are relevant - just like in medical research they look into both efficacy (performance under ideal and controlled conditions) and effectiveness (performance under actual real-world conditions).