To put it another way, the self-driving system did not alert the driver that it had detected something and did not know what it was. It wasn't an emergency, it was the car's normal operation.
If my skydive instructor doesn't deploy the backup parachute because I, the student, didn't alert them that the primary chute failed to deploy, it's entirely their fault if we hit the ground at terminal velocity.
If a lifeguard is working at a public pool watching Netflix on their phone and a kid drowns, you can't argue that the kid should have splashed more.
But how do you suppose we create FSD cars if we can't try them out before they are ready? There is just no other way to do it than vigilant drivers that watch what these cars do.
> To put it another way, the self-driving system did not alert the driver that it had detected something
Well of course not, who would expect that? If the car could positively identify the collision before it happened, it would have simply stopped, no need for a driver at all. The driver is there to prevent exactly this kinds of accidents, and this driver failed by getting our her phone and distracting herself with streaming videos instead of doing her one job. Plain and simple.
It's one thing to charge negligent homicide with a typical car. But the near-fraudulent claims of self-driving car hucksters at the time had a lot of people believing these vehicles were already far more capable than they will be for decades to come. And it's inevitable that the Koolaid drinking in this particular program within Uber was at its strongest. So I doubt the safety drivers were adequately informed of the actual capabilities of the cars or the risks involved.
Consider two training courses:
* instructors who are gung-ho on automation being "nearly there" and encouraging people to relax in the car and let the software do its work!
* instructors who are constantly impressing upon students that they need to be vigilant.
One can see how one party is 90% liable. And the other is 10% liable, etc.
The interesting part of this case comes down to the training. In that if it was lacking, it then makes it very clear to future companies that their training needs to be more rigorous.
There would be a different blame calculus if this were a production level 2 system like autopilot. In that case, it's not a paid test pilot, it would be the paid purchaser of a certified aircraft.
However, I don't think it's reasonable at all to expect someone to remain attentive while looking at a self-driving car. This is the same problem train drivers face, which has been mitigated with all sorts of methods, least of which a dead-man switch. Some countries let their train drivers mention every signal they come across to themselves, with Japanese train drivers even pointing at signs to ensure they're paying attention.
This was a vehicle that had been modified to reduce certain safety features (because Uber couldn't get them to work properly) with someone at the wheel expected to be 100% focused on the road while giving them nothing to do at the same time. You can only go through so many hours of sitting in a card doing nothing before you go crazy.
From a revenge-seeking perspective it's easy to blame the one person who could've stopped the car for her obvious disregard for safety (streaming video on the job), and I suppose a criminal justice case might be in order. However, I think Uber should be mainly responsible for the loss of life because their flawed design not only made the car less secure but also completely disregarded the human psychology when they designed how their human safeguard driver should do their job. Even human-operated cars will beep and yell at you if you don't pay attention while you're driving in cruise control, if such safety features were omitted in the self-driving design then clearly the driver was set up to take the fall when something bad would happen.
I strongly believe Uber only put that woman in there because local law wouldn't let them test their car without a human at the wheel, not because they wanted to ensure their car didn't kill anyone.
I think pointing out things in the environment is a great idea for safety drivers in this kind of setting. It helps keep them engaged, possibly helps the system notice when they're distracted, and possibly provides additional useful training data.
No, the last articles and discussions were quite clear that this accident wouldn't have happened with a real driver. The road conditions were good, the visibility was good. A driver would have seen the woman crossing the road well enough and slow down.
lets just acknowledge inefficiencies in the system
I don't even know if an expert witness was there, thats how grand juries work.
And that a corporation didn't tell a full truth and threw an "maybe employee" under the bus? Not really that much to "posit".
Ironically also trending on HN at time of writing: https://news.ycombinator.com/item?id=24488350
When I saw the dashcam footage, I just thought to myself, I probably would have avoided that, had I been driving.
I've watched my own dashcam footage before and people on the periphery come out less visible than in reality because the headlights blow out the video - the camera has less dynamic range than our eyes.
Also trending on HN today: https://news.ycombinator.com/item?id=24488350
Oh, right...
This is part of the general attitude that rights and freedoms don't matter if a machine violates them. When you walk out of a store post purchase and the security alarm goes off you have zero obligation to disclose what is now your property to the loss prevention experts. But somehow it's acceptable to assume you're a criminal because a machine said so.
Driving while watching a video certainly qualifies normally, but purely hypothetically, if Uber had internally said it was totally safe to drive distracted, then maybe she has a defense.
Only to a crime whose required mental state is intent, which rules out crimes of negligent or recklessness (or strict liability, for that matter.)
> The state has to provide criminal negligence, which in most states requires very risky behavior.
“A motorist can be convicted of negligent homicide for causing the death of another person while driving in a criminally negligent manner. A person acts with criminal negligence by unknowingly doing or failing to do something that creates a substantial and unjustifiable risk to others. The risk must be of such nature and degree that the failure to perceive it constitutes a gross deviation from the standard of care that a reasonable person would use in like circumstances.”
https://www.drivinglaws.org/resources/arizonas-vehicular-hom...
I don't think that that's a hard fit for the publicly-reported facts of this case.
> Driving while watching a video certainly qualifies normally, but purely hypothetically, if Uber had internally said it was totally safe to drive distracted, then maybe she has a defense.
No, if “Uber” said that there is an even greater case than there already is under simple respondeat superior for the actions of the driver alone as agent of the firm for Uber to be charged with negligent homicide, manslaughter, or even 2nd degree murder.
But I don't think any representation by Uber could reduce the standard of care owed by a safety driver to members of the public to something below “not watching a video on the phone while supervising the safety of the operation of a self-driving vehicle”.
>I don't think that that's a hard fit for the publicly-reported facts of this case.
Agreed, unless (again 100% hypothetically) Uber misled the driver about the capabilities. Even then it would still require that a reasonable person would believe a (hypothetical) incorrect statement about the car's capability.
>No, if “Uber” said that there is an even greater case than there already is under simple respondeat superior for the actions of the driver alone as agent of the firm for Uber to be charged with negligent homicide, manslaughter, or even 2nd degree murder.
Not sure respondeat superior applies to criminal law. In either case, I think operating a fleet while (again 1000% hypothetically) misleading the drivers would be itself the act of criminal negligence. So I don't think you'd even need to resort to master/agent liability.
>But I don't think any representation by Uber could reduce the standard of care owed by a safety driver to members of the public to something below “not watching a video on the phone while supervising the safety of the operation of a self-driving vehicle”.
Not my area of expertise and I haven't looked at any case law on this point, but I think a believable statements made by Uber would be considered as the circumstances wrt "standard of care that a reasonable person would use in like circumstances."
If a reasonable person would believe it was safe to watch the video while driving, it wouldn't qualify as criminal negligence.
The reason why Uber had a human monitor, one required to have a valid motor vehicle license, is because of the potential lethal outcomes. You learn all of this when preparing to test for a license. It doesn't stop being one's responsibility to be a vigilant driver because it's convenient to do something else. Even considering the general stupidity levels of American drivers, ignorance of the law is no excuse.
What Uber (hypothetically might have) said is completely irrelevant; she was the driver, she is responsible.
IMO this is much more applicable to Uber's share of the responsibility for this incident.
I dont even know what license beyond a driving license exists for this job. Clearly not every driving license holder is fit to hold this position.
protip: You probably want to let them figure out what is the thing beeping while you're still in the shop. It's not great to get home only to realize that they forgot to take off the anti theft device on some of your beers.
https://hoodline.com/2020/08/driver-who-killed-pedestrian-at...
As for "how many have videos": when Amelie Le Moullac was killed by a truck, the police didn't even check videos. There had to be a grassroots effort to go get the video.
You're kidding me.