4 ms·

Accidents are not good, but how does FSD safety compare to human drivers in general? Human drivers run over and kill school children too, so which is actually

Accidents are not good, but how does FSD safety compare to human drivers in general? Human drivers run over and kill school children too, so which is actually worse statistically? Human or FSD? I'll side with the FSD / Elon haters if FSD is significantly worse. Until then I ask questions and remain neutral.

llamaimperative 2y ago
Tesla's decision not to release the data required to make such an assessment might be considered some sort of indication about what that data says
- taylodl 2y ago
  It's indicative of dishonesty and fraud.
rootusrootus 2y ago
I've tried FSD enough that I can say with perfect confidence that it is orders of magnitude less safe than an average human. Even a drunk one. Every few miles of driving it would do something that would most likely be a fender bender at minimum. That's pretty risky. Compare to human drivers, which average roughly 1 fatality for every 100,000,000 miles driven. And that's including drunk drivers. My guess is FSD in the current state would probably average 1 fatality for every 100 miles or so, if you were not closely supervising it. Assuming none of the wrecks along the way disabled the car.
- bboygravity 2y ago
  What version of FSD? In what year?
add-sub-mul-div 2y ago
> Accidents are not good, but how does FSD safety compare to human drivers in general? We'll never really know and tech companies aren't trustworthy. A new software update will invalidate what we've observed about the last version and we'll be back to knowing nothing. Humans in aggregate are predictable and nothing will make that entire fleet more dangerous overnight. Legacy car companies came about in an era when they faced accountability from both consumers and lawmakers that we can no longer count on.
- taylodl 2y ago
  > A new software update will invalidate what we've observed about the last version and we'll be back to knowing nothing. "Move fast and break things" is being replaced with "move fast and kill people."
Veserv 2y ago
It is Tesla's duty to provide rigorous, scientifically, and statistically sound evidence for their claims of safety as Waymo provides [1][2]. Instead, they present bare numbers with no base rate estimations, confidence intervals, methodology, or model supporting them despite already killing tens of people. They deliberately market falsified numbers [3] and their models are unable to detect individual sources of error that result in underestimating crashes by a factor of 500% (from a single well-known factor, ignoring any other source of error which we should assume are abundant given they failed to account for differences in airbag deployment rates which literal internet posters have been pointing out is a potential source of bias for years). They are either incompetent or intentionally misrepresenting their actual safety statistics. Until the experts at Tesla can find some rigorous evidence that their system is actually safer, it is safe to assume that it is not; they have had plenty of time for a competent team to produce sound estimates. [1] https://waymo.com/safety/ https://waymo.com/safety/ [2] https://assets.ctfassets.net/e6t5diu0txbw/54ngcIlGK4EZnUapYvAjyf/7a5b30a670350cc1d85c9d07ca282b0c/Comparison_of_Waymo_Rider_Only_Crash_Data_to_Human_Benchmarks_at_7_1_Million_Miles_arxiv.pdf https://assets.ctfassets.net/e6t5diu0txbw/54ngcIlGK4EZnUapYv... [3] https://static.nhtsa.gov/odi/inv/2022/INCR-EA22002-14496.pdf https://static.nhtsa.gov/odi/inv/2022/INCR-EA22002-14496.pdf
- bboygravity 2y ago
  Sounds reasonable. Weird that the (US) government can't force Tesla to publish that data as a per-condition to be allowed to offer FSD at all.
- aithrowaway1987 2y ago
  Waymo's numbers aren't dishonest and their accomplishment is real, but this data should be taken in its proper context. 1) The comparison lumps in a lot of humans which cynically and selfishly break the law. Currently Waymos attempt to obey the laws of the road, but Tesla FSD shows that self-driving companies can and will market towards people who want to speed, run red lights, etc. So I am deeply concerned that a formal comparison of law-abiding self-driving cars will be used to encourage development of autotaxis which compete with human Ubers on travel time by speeding or rolling through stop signs like a human would. It would ultimately be a correlation/causation error: Waymo's training and development correlates with better safety records when the causative variable is ultimately risk-adversity, with better engineering being a secondary factor. In particular, I suspect Waymos will eventually be safer than humans at legal driving but dramatically more dangerous at illegal driving, and Waymo executives using motivated reasoning to dispel the concerns. Again, FSD literally offered different options for how much you wanted to speed or roll stop signs. This isn't hypothetical. 2) This is less concerning than reckless driving, but these stats include a lot of human accidents due to mechanical failure. This is something I expect self-driving to handle very poorly without human intervention. The most cognitively demanding moment of my entire life was having a rear tire blowout while driving a heavily-loaded truck with little experience - as the end started to fishtail I quickly realized what happened and carefully applied the brakes, hit the emergency lights, closely paying attention to the tactile feedback in my feet and hands, then pulled over as soon as I possibly could. If I were an AI and this was an edge case the engineers forgot to train on, Robo-Me would have easily flipped the truck. Mechanical failures aren't occurring in Waymos now because the cars are still new. Eventually it might be a real problem.
taylodl 2y ago
Don't want to hear it so long as Tesla holds me legally responsible for whatever their software does. Unless and until Tesla is fully legally responsible for whatever their FSD does, then it is NOT "Full Self-Driving." Calling it "Full Self-Driving" is fraud and those who purchased it should file a class-action lawsuit.

llamaimperative 2y ago

Tesla's decision not to release the data required to make such an assessment might be considered some sort of indication about what that data says

taylodl 2y ago
It's indicative of dishonesty and fraud.

rootusrootus 2y ago

I've tried FSD enough that I can say with perfect confidence that it is orders of magnitude less safe than an average human. Even a drunk one. Every few miles of driving it would do something that would most likely be a fender bender at minimum. That's pretty risky. Compare to human drivers, which average roughly 1 fatality for every 100,000,000 miles driven. And that's including drunk drivers. My guess is FSD in the current state would probably average 1 fatality for every 100 miles or so, if you were not closely supervising it. Assuming none of the wrecks along the way disabled the car.

bboygravity 2y ago
What version of FSD? In what year?

add-sub-mul-div 2y ago

> Accidents are not good, but how does FSD safety compare to human drivers in general? We'll never really know and tech companies aren't trustworthy. A new software update will invalidate what we've observed about the last version and we'll be back to knowing nothing. Humans in aggregate are predictable and nothing will make that entire fleet more dangerous overnight. Legacy car companies came about in an era when they faced accountability from both consumers and lawmakers that we can no longer count on.

taylodl 2y ago
> A new software update will invalidate what we've observed about the last version and we'll be back to knowing nothing. "Move fast and break things" is being replaced with "move fast and kill people."

Veserv 2y ago

It is Tesla's duty to provide rigorous, scientifically, and statistically sound evidence for their claims of safety as Waymo provides [1][2]. Instead, they present bare numbers with no base rate estimations, confidence intervals, methodology, or model supporting them despite already killing tens of people. They deliberately market falsified numbers [3] and their models are unable to detect individual sources of error that result in underestimating crashes by a factor of 500% (from a single well-known factor, ignoring any other source of error which we should assume are abundant given they failed to account for differences in airbag deployment rates which literal internet posters have been pointing out is a potential source of bias for years). They are either incompetent or intentionally misrepresenting their actual safety statistics. Until the experts at Tesla can find some rigorous evidence that their system is actually safer, it is safe to assume that it is not; they have had plenty of time for a competent team to produce sound estimates. [1] https://waymo.com/safety/ https://waymo.com/safety/ [2] https://assets.ctfassets.net/e6t5diu0txbw/54ngcIlGK4EZnUapYvAjyf/7a5b30a670350cc1d85c9d07ca282b0c/Comparison_of_Waymo_Rider_Only_Crash_Data_to_Human_Benchmarks_at_7_1_Million_Miles_arxiv.pdf https://assets.ctfassets.net/e6t5diu0txbw/54ngcIlGK4EZnUapYv... [3] https://static.nhtsa.gov/odi/inv/2022/INCR-EA22002-14496.pdf https://static.nhtsa.gov/odi/inv/2022/INCR-EA22002-14496.pdf

bboygravity 2y ago
Sounds reasonable. Weird that the (US) government can't force Tesla to publish that data as a per-condition to be allowed to offer FSD at all.
aithrowaway1987 2y ago
Waymo's numbers aren't dishonest and their accomplishment is real, but this data should be taken in its proper context. 1) The comparison lumps in a lot of humans which cynically and selfishly break the law. Currently Waymos attempt to obey the laws of the road, but Tesla FSD shows that self-driving companies can and will market towards people who want to speed, run red lights, etc. So I am deeply concerned that a formal comparison of law-abiding self-driving cars will be used to encourage development of autotaxis which compete with human Ubers on travel time by speeding or rolling through stop signs like a human would. It would ultimately be a correlation/causation error: Waymo's training and development correlates with better safety records when the causative variable is ultimately risk-adversity, with better engineering being a secondary factor. In particular, I suspect Waymos will eventually be safer than humans at legal driving but dramatically more dangerous at illegal driving, and Waymo executives using motivated reasoning to dispel the concerns. Again, FSD literally offered different options for how much you wanted to speed or roll stop signs. This isn't hypothetical. 2) This is less concerning than reckless driving, but these stats include a lot of human accidents due to mechanical failure. This is something I expect self-driving to handle very poorly without human intervention. The most cognitively demanding moment of my entire life was having a rear tire blowout while driving a heavily-loaded truck with little experience - as the end started to fishtail I quickly realized what happened and carefully applied the brakes, hit the emergency lights, closely paying attention to the tactile feedback in my feet and hands, then pulled over as soon as I possibly could. If I were an AI and this was an edge case the engineers forgot to train on, Robo-Me would have easily flipped the truck. Mechanical failures aren't occurring in Waymos now because the cars are still new. Eventually it might be a real problem.

taylodl 2y ago

Don't want to hear it so long as Tesla holds me legally responsible for whatever their software does. Unless and until Tesla is fully legally responsible for whatever their FSD does, then it is NOT "Full Self-Driving." Calling it "Full Self-Driving" is fraud and those who purchased it should file a class-action lawsuit.