Hacker News Reader

Comment by Brybry | Hacker News Reader

What if "Autopilot"/"FSD"/computer-assisted driving hallucinates and drivers using it run over children but at a lower rate than people who are not using computer-assisted driving run over children?

If progress for automobiles is that it has to be perfect (and not just better) then the only answer is to use automobiles less or have stricter training standards for drivers.

I do think there needs to be strong proof that it is better to avoid liability. And I agree Tesla's marketing is bad and should make Tesla liable.

View on HN

It comes down to the simple fact that somebody is going to hold the liability bag.

Tesla and similar obviously are not pushing to take on that liability, courts/legislatures aren't particularly interested in putting that liability on the manufacturers. So that leaves the drivers to try and offload the liability from themselves.

Unfortunately drivers are not particularly well equipped to be pushing that liability off themselves so there it stays. Insurance companies are better equipped for that sort of thing but they're only interested in the financial liability. The criminal liability, is definitely not going anywhere anytime soon.

> drivers are not particularly well equipped to be pushing that liability off themselves so there it stays

It's premature to conclude this. We don't have enough real-world cases off which to construct a framework.

> Insurance companies are better equipped for that sort of thing but they're only interested in the financial liability

Even this hasn't been cleanly severed in any competent jurisdiction.

> It's premature to conclude this. We don't have enough real-world cases off which to construct a framework.

Precisely, it's premature to conclude that anybody but the driver is at fault. Until a sufficient body of law has been built towards the alternative it will remain by default with the driver.

I think it depends on how the car ran over the children (or anyone in that case). If it was an unavoidable freak accident, then you can't really blame the car.

But, if the car literally didn't see a person in front of it (where reasonably a human would 99.9999% of the time) because it's cameras malfunctioned or the LLM read it as something else, then those cars should not be on the road.

There is no LLM in Tesla cars. Otherwise correct. I'd rather expect they use a convnet or a vision transformer, probably the former.

I actually disagree, if it can be shown to be safer overall then I don’t think it should be required to be a strict superset of human abilities.

AI will always behave differently to human cognition, we should expect them to have different illusions and different failure cases. That doesn’t mean worse, if they perform better than a legal human driven car on average then I don’t think there’s a justification to exclude them.

That said, they will probably have to be _significantly_ better than human drivers in order to survive the media and public perception, so this might be irrelevant in the end.

To play devil's advocate. Say that on average it was 10x or 100x safer than human drivers, but for whatever reason, there was a 1/1,000,000 chance that the self driving car would plow into traffic at a red light because some black-box instruction told it it was the thing to do, very likely injuring or killing you. But overall it was better in all other cases. Would you take that risk?

It’s not called “almost autonomous driving”, it’s named incorrectly and as a result falsely marketed.

> not called “almost autonomous driving”, it’s named incorrectly

Article says Autopilot was engaged, not FSD. You can't just check out after engaging autopilot on a plane.

Reply on news.ycombinator.com