Hacker News new | ask | show | jobs
by NoZebra120vClip 1053 days ago
> Their job was to supervise an autonomous driving test

The "driver" performed her job flawlessly, by acting exactly as any other "driver" who's been rendered redundant by fully-automated systems. This "driver" exemplified the human indifference to the task when our attention isn't demanded by complete immersion in an interactive experience.

By causing the vehicle to start operation, the driver accepted responsibility and liability, per the current laws on SDCs. Therefore, "scapegoat" is a misnomer that wrongly implies innocence or detachment from the sinful deed(s). The driver accepted her fate by signing up and settling into the left-hand front seat.

3 comments

By causing the vehicle to start operation, the driver accepted responsibility and liability, per the current laws on SDCs. Therefore, "scapegoat" is a misnomer that wrongly implies innocence or detachment from the sinful deed(s). The driver accepted her fate by signing up and settling into the left-hand front seat.

I call bullshit.

Why was the "SDC" even on the road?

See, refusal to deal with this question is the attitude that increasingly pisses me off about tech companies. Oh, it's totally fine for me to deploy a multi-ton vehicle system with a "backup" operator that all human factors research reasonably tells us is going to be rendered unsafe by the levels of detachment induced by too good, but not good enough" levels of automation. Heaven forbid though that the company be ultimately culpable for the hazard they create. After all, they got an "independent contractor" to transfer risk onto. Likely one that poorly the nature, implementation, mechanics, and character of the system they were to supervise.

Gotta exploit the opportunity for massive profit, while saddling someone else with all the bite of the risk, because conventional testing and sane operational practices (which includes not deploying shit you know is going to be unsafe) is toooooooooo expensive and too much to ask.

Uber, apparently made the decision to deactivate the AEB system (likely because otherwise they couldn't convince anyone sane to ride in the car because they system wasn't ready). Uber decided to undertake risk transference to an independent contractor that, lets be real, every executive there knew they weren't going to have a great success rate at getting across the real operational envelope of these vehicles. The same exec team that brought you things like their networks kill switch to frustrate LE raids in other jurisdictions, and project Greyball, a system to gaslight LE in other jurisdictions as well. If anyone was "Jesus take the wheel"-ing that night, and every night leading up to it, it was Uber.

But nah... We all sit back and accept these types of shenanigans as just "cost of doing business". It's "too important" to develop safely.

>Oh, it's totally fine for me to deploy a multi-ton vehicle system with a "backup" operator that all human factors research reasonably tells us is going to be rendered unsafe by the levels of detachment induced by too good, but not good enough" levels of automation.

Except in this particular case, the driver wasn't just zoned out, she was streaming Hulu. I don't think she was so lacking in free will that not having a stream on was out of her control.

>Uber, apparently made the decision to deactivate the AEB system (likely because otherwise they couldn't convince anyone sane to ride in the car because they system wasn't ready).

You're saying they disabled the AEB system, otherwise people wouldn't ride in the car? How does that make sense?

>You're saying they disabled the AEB system, otherwise people wouldn't ride in the car? How does that make sense?

Their AEB technology had too high a false positive rate resulting in what a consumer would quickly coin "the perception of automated vehichle as deathtrap".

Once again, leading to "we must put something on the road now (despite it not being ready for prime time).

I'm inclined to disagree here. Their role here is akin to a driving instructor. Regardless of whether the primary driver is a human student or a AI, their job was to supervise the operation of the vehicle. They did not do it and someone died. There's room to pass the buck around to be sure, but I think that was reflected in the sentencing.
Per the current laws, but what about per objective ethics? I don't care what is written in the code book.
The person shirking their duty knew the risks and was even profiting from their negligent behavior. Meanwhile a person lost their life who has nothing to gain from this. I blame the employer and the employee. Employers shouldn’t set their employees up to fail, and employees should know a risky proposition when they see it.
The stated promise of self-drive is reduction in accidents.

The accident victim didn't sign up for this, but as a member of society she did stand to gain something; to wit, harm reduction. I do see the grim irony, but, nonetheless.

Generally, it's the trolley problem. (Although in this specific case there's an element of human fault.)

and if there is no accident, due to intervention of the driver, will the situation even be noted somewhere? in similar threads on hn there are many near miss stories.
This assumes employees have the privilege of walking away from a risky proposition.
> Vasquez had previously spent more than four years in prison for two felony convictions — making false statements when obtaining unemployment benefits and attempted armed robbery — before starting work as an Uber driver, according to court records.