Hacker News Reader

Comment by alexd127 | Hacker News Reader

How would user vs developer be distinguished? Looks like it mentions both users and developers but doesnt distinguish how one would be found liable

View on HN

> This bill would prohibit a defendant that developed or used artificial intelligence, as defined, from asserting a defense that the artificial intelligence autonomously caused the harm to the plaintiff.

I am not sure the distinction matters?

If you’re driving a self-driving car, you can’t blame it on the car.

You can blame it on a car even if the car is not self-driving

It’s your responsibility to ensure that your breaks etc work.

Presumably it’s also your responsibility to pick and maintain a working self driving system.

Vehicles have defects and sometimes the manufacturer is at fault for accidents.

The upstream poster is correct. This new law has no relevance to who is liable. It would simply remove "the AI did it" as a valid defense in any case (with whatever exceptions are defined in the existing law referenced).

Like the resale of stolen property. The victim can always just sue you, you may be able to sue the person you got it from.

You can always be sued, and this may be a difficult defense to make in practice, but drivers are not generally liable for accidents caused by mechanical failure if the cause is manufacturing defect.

Sure, if you can demonstrate that there’s no reasonable way the fault could have been predicted or avoided then that can get you off the hook.

However if your tire blows out you’d be expected to demonstrate that you regularly inspect them for ware or damage and there hasn’t been a recall etc. That same level of proactive care is going to be applied to self driving systems.

I don’t see anyone really disputing that. To the extent I’ve participated in this thread it has been to make clear:

1. Both car manufacturers and car drivers can be liable, even with self-driving cars. Any confusion here is likely due to conflating the car with the car manufacturer.

2. The proposed law wouldn’t assign liability, it would simply remove “the AI did it” as a possible legal defense.

What I think people are ignoring is choosing to be an early adopter of fully autonomous self driving vehicles is itself going to be questioned.

Being the first member of the general public to use a 100% self driving car the first day it’s available might even be considered reckless if it then crashes that day.

Later of is a model is preforming poorly operating such a vehicle could be called into question etc.

I don’t think it’s that binary. You have some duty to make an effort of maintenance but there can still be accidents that are just technical faults no one is really responsible for.

Good luck arguing that to a jury.

It’s true in some cases the manufacturer or car mechanic etc is at fault rather than the owner, but it’s difficult to offload responsibility to a 3rd party.

People (or rather their insurers) successfully argue that to a jury all the time. If your vehicle is serviced by a professional mechanic according to the manufacturer's recommendations, it's very difficult to argue that you're liable for the consequences if your brakes suddenly fail. You took all reasonable steps to ensure that your vehicle was in a roadworthy condition. If you didn't bother to follow the manufacturer's recommendations, then you're on your own.

You’re making this seem easier than it is. Even just convincing people the brakes actually failed is a hurdle.

Someone trying to defend themselves by saying the brakes failed needs to show the brakes alone are the cause of the accident rather just a contributing factor. So there was no alternative like using a parking brake and the driver didn’t get into an unsafe situation.

Similarly the failure must be sudden and not predictable etc.

I think you meant “brakes”.

>Presumably it’s also your responsibility to pick and maintain a working self driving system.

I guess it's reasonable to say I should apply updates and get the car serviced and make sure the sensors are not obstructed, yes. Then the difference between "self driving" and "assisting" technology would be a matter of guarantee the manufacturer advertises.

Would it (hypothetically) be reasonable for me to expect that the thing will suddenly break on a highway in normal conditions without anything obstructing the way forward? No, I don't think so. Can I do something to prevent it or foresee it, except not using the self-driving technology at all? Can I choose to have a different self-driving tech installed in my car and retrain the model or control it's behavior in any way at all?

There are different answers to that and I guess the answers will also change over time.

I think there's a potential confusion here between different kinds of "blame."

If your robot punches a bystander, you're liable for their hospital bills, separately from whether you're culpable for battery.

Perhaps you can cite when this was tested in court, because "robot punches man" isn't something I've noted in the headlines. Otherwise, it's nothing but speculation.

That wouldn't work anyway. In basically every jurisdiction, operators of vehicles are expected to retain control of the vehicle regardless of self driving status.

I think it’s useful for legislators to do their job and explicitly define and clarify the boundaries of the law. You can’t necessarily just rely on precedent for new things.

Not a new question: "the AI did it" is an equivalent of "the gun did it" or "the car did it".

> or "the car did it".

Tesla gives you the old with the new.

Accountability hacking: https://www.motortrend.com/news/nhtsa-tesla-autopilot-invest...

Liability sinks is the word

On the other hand, “the corporation did it” has been pretty successful so far

Reply on news.ycombinator.com