Featured Tesla Q2 2024 Safety report:

So probably people are using autopilot more on highways where you get more miles without crashes?

 

Like others adaptive cruise control and lane keeping, Autopilot can work at slower speeds, but without the stop reading abilities of FSD, most people likely only use it on highways and stretches of road without stops.

 

This is called "selection bias". It has been pointed out for years, but we still aren't getting the data transparency needed to gauge or address it.

 

Would drivers who use AutoPilot while it is under investigation by NHTSA be subject to liability lawsuits in the event of an accident?
This was some time ago but before this system was under investigation:

Tesla driver is charged in a deadly crash involving Autopilot : NPR

Tesla Autopilot Lawsuits (klinespecter.com)

Tesla must face vehicle owners' lawsuit over self-driving claims | Reuters

 

Autopilot do read stop signs and red lights but will only warn you, not stop. I wish they included that functionality to manual driving too. It could prevent so many t-boning.

 

I use it on highways and regional roads unless the traffic speed on those is faster than 10 km/h over the speed limit (hard limit by AP on roads other than highways), then it's just TACC. I tried the Volvo lane assist on those same regional roads and after a few km, I disengaged it as it can't follow the curves on those roads.

 

I think the selection bias has been with each single autopilot crash, people say autopilot has been killing people. Instead with and without autopilot are much safer than the average car. We have a whole thread assuming that bad assumption.

Most teslas are newer cars, and older cars may be much less safe. Tesla buyers appear to be more affluent which may make them safer drivers. What can not be said is autopilot is unsafe because it has been involved in a tiny percentage of accidents where the driver should have taken over. I don't know if teslas are more or less safe than a new toyota with safety sense, but certainly autopilot is not killing the huge percentage of people in the database.

Investigation has nothing to do with if it was subject to lawsuits. Normally in wrongful traffic deaths people go after the driver and car company, as the car company has more money. But the incidents I have read about the driver should have been paying more attention so its unlikely autopilot is at fault.

 

For drivers, there should not be any added liability, compared to those not using AP. For one, AP isn't legally self-driving, so the liability split is technically already 100% human driver / 0% machine AP. For another, "under investigation" normally means no legal action has (yet) be taken, and there is no certainty (yet) that any action will ever be taken. Drivers have not (yet) been given any form of legal notice beyond what is in the Owner's Manuals.

Drivers using AP and other ADAS systems properly, should have nothing to worry about. It is only those who let their expectations get ahead of current technical reality, using AP / ADAS as a crutch to do things they wouldn't do in its absence, who should have any liability concerns, along with the manufacturers who encourage or entice them to do those things.

 

That would be an additional segment of selection bias, not "the" (as in only) selection bias. You acknowledge more than one selection bias in your next paragraph, by listing some of them:

I don't believe I've suggested that AP per se, is unsafe or is killing people. It is drivers, either of their own volition or through the encouragement of marketeers or fan-bois, putting their expectations ahead of current technical reality, who might be doing so. Common human nature does push people in that direction, and any single human's lifetime experience is too small a sample base to derive an answer of "no problem". The current opaque non-independent statistics are not allowing us to see a proper answer.

 

Autopilot hardware is standard on all recent Teslas. Autosteer is all one gets in the base price. Perhaps the middle tier Autopilot has been discontinued since I last looked in December 2023 now that the full FSD (supervised) has dropped from $12,000 to $8,000 or $99/month?

At $1,200 I'd take it! If facing a long road trip perhaps I'd rent it for a month. But at current prices I can pass, Autosteer is pretty good on interstate.

Not happy with panic braking when a vehicle crosses in front of me, or when a big truck is coming around a left hand curve toward me while I'm going right in the curve. When I see this before the adaptive cruise control sees it, I put foot on accelerator to override.