26 minutes ago, OldTony said:

Human beings have accidents when driving cars - Insurance assessors can no doubt provide statistics on the risk factor. It seems likely to me that cars controlled by AI will also have accidents, but be developed to the point that statistics show them to be safer than humans. I can see a person in such a car that has an accident giving as an excuse "I gave control of the car to a system that statistics show improved safety on the road". Even more than that, when AI control is acknowleged as the safer option I can't see why the human occupant will even need to be able to drive at all. Is that a beautiful dream or a nightmare ?

There are already autonomous robotaxis and what we currently learn is that it obviously cuts down on human failings, such as cell phone use and other distracted driving, DUI etc. And therefore outperform drivers especially under good conditions. The issue is a bit murkier under bad weather, dusk/dawn situations, but this could be areas of improvement. It should be noted that this is likely not exclusively a software-side issue, though companies like Tesla treat it as such, but also a sensor-side issue. LIDAR significantly outperforms camera only, for example.

If we get tot the point that autonomous driving becomes the norm, I would assume that liability will be similar with respect to robotaxis, i.e. the manufacturer will be liable, for crashes while in autonomous mode. Though there will be questions regarding shared liability with respect to maintenance or use beyond the terms that will likely exist (e.g. driving outside of established grids).

But again, I don't think that autonomous driving, at least for the foreseeable future, is an AI-only question.

57 minutes ago, OldTony said:

Human beings have accidents when driving cars - Insurance assessors can no doubt provide statistics on the risk factor. It seems likely to me that cars controlled by AI will also have accidents, but be developed to the point that statistics show them to be safer than humans. I can see a person in such a car that has an accident giving as an excuse "I gave control of the car to a system that statistics show improved safety on the road". Even more than that, when AI control is acknowleged as the safer option I can't see why the human occupant will even need to be able to drive at all. Is that a beautiful dream or a nightmare ?

We’re in a discussion of legal liability, so who is responsible when the accident occurs. Is it the AI company, or the non-driver? If it’s the latter, then people are going to have to be convinced that it’s worth assuming the liability.

When governments use AI and the AI errs and violate someone’s rights, they aren’t really in a position to pass the buck

2 hours ago, swansont said:

It’s not nebulous that a human is involved somewhere, making a decision - turning on a machine, etc. The legal responsibility is nebulous when the law doesn’t exist that holds people accountable. That’s why we need laws and legal precedent, as some of us are discussing. Unfortunately the law is sometimes absent or written such that it shields the people who are morally responsible.

Consider why motorists are so heavily shielded from charges of recklessness simply for bringing an inherently dangerous piece of machinery into an environment shared by others.

The origin of 'jaywalking' is instructive:

The term is not a historically neutral one.^[2]

The arrival of the automobile in the opening decades of the 20th century led to increasingly deadly conflicts in the street, and the public was generally unsympathetic to motorists or to early attempts to legislate pedestrian behavior.^[3][4] In response, the US automobile industry and associated organizations undertook public campaigns to frame pedestrians, newly impugned as jay-walkers, as a problematic element in the new automotive age.^[3] The first widely successful criminalization of jaywalking was enacted in Los Angeles in 1925, using legislation drafted by the auto lobby that inspired similar ordinances in other American cities.

If a man takes his imported pet leopard for walks down the village street, and every so often it bites a lump out of a passing yokel, who is morally responsible for this? The leopard? The yokel?

Pray explain why there should be any difference regarding machinery whether AI controlled or not. Indeed this argument seems to render the AI factor irrelevant.

1 minute ago, sethoflagos said:

Consider why motorists are so heavily shielded from charges of recklessness simply for bringing an inherently dangerous piece of machinery into an environment shared by others.

The origin of 'jaywalking' is instructive:

That would be an example of the kind of shielding that I mentioned. The US, at least, heavily favors automobile use to anyone else on the road.

1 minute ago, sethoflagos said:

If a man takes his imported pet leopard for walks down the village street, and every so often it bites a lump out of a passing yokel, who is morally responsible for this? The leopard? The yokel?

Pray explain why there should be any difference regarding machinery whether AI controlled or not. Indeed this argument seems to render the AI factor irrelevant.

Indeed. I can’t think of why there should be one, but there are lots of examples of corporations lobbying their way into being shielded from accountability

5 hours ago, OldTony said:

Human beings have accidents when driving cars - Insurance assessors can no doubt provide statistics on the risk factor. It seems likely to me that cars controlled by AI will also have accidents, but be developed to the point that statistics show them to be safer than humans. I can see a person in such a car that has an accident giving as an excuse "I gave control of the car to a system that statistics show improved safety on the road". Even more than that, when AI control is acknowleged as the safer option I can't see why the human occupant will even need to be able to drive at all. Is that a beautiful dream or a nightmare ?

I don't think you can hide behind statistics.

Every accident, human or otherwise, is liable to the determination of responsibility.

It is no good saying, for instance

"I am driving a Volvo since they are safer cars", if you then cause an accident in one.

As regards manufacturer responsibility motor bike are very definitely less safe than cars.

Should we blame the idiots who drive them irresponsibly or the manufacturers?

4 hours ago, swansont said:

The US, at least, heavily favors automobile use to anyone else on the road.

Quite the reverse situation in the UK.

In 2019 the 'Highway Code' was strengthened by introducing a formal hierarchy of precedence with pedestrians at the top of the list.

Formerly horseriders were top, but they now come second, with cyclists third.

Unfortunately too many drivers think they have precedence in all situations, when there are actually very few such situations indeed.