As a society, we have not figured out who is liable in cases like this -- the owner or the car company that engineered the AI.
For the most part none of this is novel to the law.
So long as the owner didn’t do anything negligent themselves, then it seems a strong claim of product liability (defective product). It seems there is very little chance of contributory negligence of the owner with this function, unlike crashes in self-drive mode, where the driver was supposed to pay attention and maintain their hands on the wheel, but turns out they were sleeping or in the backseat live streaming.
I do think there should be legislation adopted to protect the victims by giving them a statutory private right of action against both the owner and the car manufacturer to hold them jointly and severally liable for any accident involving self driving vehicles, then in practice the owner/manufacturer (i.e. their insurance companies) to sort out liability.
Assuming that's the case, I would argue that this specific sort of accident is squarely in the domain of the user.
Not even mentioning the explicit warnings to keep line of sight when you use smart summon, as well as the fact that summon is automatically limited to only within 20 meters of the car.
You may say multitasking is hard or impossible but we humans manage to approximate it anyways because it’s necessary for life. For example, if you are the parent of two fussy children in a busy parking lot, you better know how to multitask or one of those kids will escape your sight!
"The person in the driver's seat is only there for legal reasons"
from the 5+ year old marketing video still featured on https://www.tesla.com/autopilot
Asking not because I don't believe it, but because the only results I get on google by searching for that quote are your comments on HN from this year in multiple threads that include it.
If the Tesla website indeed says it (which I am yet to find), I would write it off as just future-promise-marketing. You can call it misleading in that case, which would be fair.
At the end of the day, that's not how it currently works, and you have to hold the button for the car to move during summon, and you gotta be within 20 meter radius. As soon as you let the button go, the car stops. This isn't me just saying based on what internet says, this is what I have observed myself and what the car manual says[0] (both the online and the print version, in multiple spots).
0. https://www.tesla.com/ownersmanual/model3/en_us/GUID-6B9A1AE...
It's interesting how Google narrows your world view.
> If the Tesla website indeed says it (which I am yet to find)
Here you go:
https://www.tesla.com/blog/introducing-software-version-10-0
> I would write it off as just future-promise-marketing
No, it's just lies and advertising.
Now, here, maybe the user could have prevented the car accident by letting go of the button. But, what if the user had faith (however misplaced) that the car was going to avoid the collision until it was too late to let go? Or what if the user had suffered a stroke at that particular moment and couldn’t let go? It’s an interesting thought experiment.
Some things, like the decision to trigger a weapon in an act of war may be something societies want to have humans, and only humans make the ultimate call. That is, there is a desire for a human-in-the-loop.
On the other end of the spectrum are decisions we want to take out of the human hands because we think automation is more reliable and better for our society.
And then there are the unintended consequences. Looking back the past 10 years, it's arguable that the 'share' and 'retweet' features of social media that gives the power of virality to ordinary people have had a profound effect on our society and culture, both for good and for bad (and for many cases where we can't even agree on what's good or bad).
I argue that, while some of this is not novel to the law, I don't necessarily think that those precedents cover all of the changes tech has wrought to how we live and govern ourselves.
If my car is parked in a parking lot, and someone rams my car so that it rolls and hits another car, I am liable. This is why parking breaks are important
I don't understand why self driving cars would be any different. I am liable to make sure I am using my car properly. Even if it has fancy tech.
Relying on images only requires you to perfectly identify every object in the scene in order to infer its bounding box. And that's an issue like we see here where you don't expect to see private jets near a Tesla and so it may not be modelled.
With LiDAR, Radar etc. at least they know "something" is there even if they don't know what is.
https://www.theverge.com/2021/5/24/22451404/tesla-luminar-li...
https://www.patentforecast.com/2021/07/22/lidar-competition-...
https://www.patentforecast.com/2021/09/02/waymo-is-driving-l...
Fwiw, the best use of summon is simply getting out of a tight parking spot (e.g., some residential garages).
Expecting it to be a valet is ambitious in its current state.
So which is the definitive statement?
That said, some of the Tesla marketing, like in the example you provided, is very far off the mark. I really wish they wouldn’t do that, since it makes people question the other things marketed about the car that are (more or less) accurate.
Fwiw, the summon feature is a nice, minor convenience. I would even say that it is well-implemented for simple use cases. That said, the existing feature does not match the dreams that Tesla is currently selling (imho).
I bet it'd work great on Mars, though. Not a whole lot of airplanes there, now or in the foreseeable future.
I find disturbing how the car just keep going... This is not a sensor issue.
Tesla stops when it sense a imminent collision even if you have your foot on the pedal.
But, based on the incident, it seems to rely on the cameras and radars. Not on engine performance data.