Hacker News Reader

Comment by rbtprograms | Hacker News Reader

I think fault, specifically when it comes to litigation, is something that will end up holding back self-driving cars (at least in the USA). If there is an accident or someone is injured, who ends up getting sued? How does insurance resolve these claims?

Please note that I am not saying I think someone needs to get sued. I just think that the concept of liability is deeply ingrained in the American legal and insurance systems, and I almost forsee a future where we cannot come to terms with who is liable in different situations and it coming to preventing the adoption of and progression of self-driving vehicles.

View on HN

Reminds me of MYCIN health expert system from the 1970s which was quite good at diagnosing but ultimately failed because people would not know who to sue in the event of a misdiagnosis.

interesting note, thanks for sharing it.

https://en.m.wikipedia.org/wiki/Mycin

Expert systems where the previous iterations of what NeuralNets or Decision trees do today. They where hyper focused/trained in a specific domain. At that time NN where only an academic amusement given computer power.

Goes on to show that after all these years, we havent progressed in generalisation of artificial intelligence.

This isn't new though, is it? Surely there is a history of cases where the driver could have avoided the accident and changes in the features and manufacture of the automobile could also have avoided the accident.

There's various liability schemes for civil suits to settle the issues of partial fault. The states vary in which one they use.

Perhaps a solution would be a liability chain ? Get sued personally as the vehicle operator, and then sue the car manufacturer yourself ?

This way people will be pushed to pick the safest cars, or to adapt their attitude when the self-driving functionalities are unsafe.

It seems unfair due to the high costs though, meh quite a bad idea.

This doesn't help if you go to jail for the actions of the self driving car. (Generally. I'm not talking about the specific case linked).

Also, consumers can't be trusted to pick the safest cars. Most consumers choose on price, not safety features.

I may have a solution then,

In the future, let's imagine in 20 years, we may see "cars as a service", so essentially you can summon a Tesla, and the Tesla can bring you anywhere.

In that perspective, isn't it clear that the liability is to the company that organised the trip ? In similar way to when you get carried by a bus, a taxi or a train ?

But then why would people prefer this over driving and operating a car themselves?

One problem I‘ve always seen with the fleet model of widespread automated cars is that in 2020 cars are very much status symbols as much as they are modes of transportation. People own BMWs and Porsches to show off status first and get around second, and these types of people will always want “their” car.

I would expect that their status would come from being able to hail more expensive cars. They would spend extra to have an automated Porsche, BMW, etc. pick them up.

There will be cheap fleets and more expensive fleets, just as there is UberX, UberBlack and UberSelect. You could also imagine a BMW fleet that offers specific models only if you are an established frequent customer, so people can show off that they didn't just upgrade for this occasion but always travel in style.

Of course there are still plenty of edge cases where individual car ownership is nessesary, whether that's because too few cars of that type would be booked to make a fleet viable (Rolls-Royce) or because you are in a very rural area.

What you are describing is an automated taxi.

Rich people who want to call a chauffeured luxury car can do so in most major cities. Many of them do use such a service, but many of them also own expensive cars to show off their wealth.

How do you define and determine whether a consumer "chose on price" or "chose on safety features"? Surely consumers consider both the price and the desirable features when they purchase a car, just like for every other purchase. I suspect that most consumers wouldn't buy a $100 car than has a 50% chance of exploding every time you enter it, and most consumer wouldn't buy a $500,000 car that was 5% safer than a $40,000 car.

The point is that you shouldn't be able to sell or buy a $100 car that has a 50% chance of exploding every time you enter it. Or more to the point, you shouldn't be able to sell a shitty self driving system. The post I was responding to was basically saying "let the free market figure out which car has a better self driving system" which is a horrible way to go about it.

If manufacturers can be directly sued, then the cost will have to be incorporated into the cost of the cars. That being said, consumers already like being safe for their own benefit. And since safety is entirely in the hands of the manufacturer, consumers will probably be even more selective along this axis than currently.

But at the end of the day, we all share the road together, so the government will regulate some minimum safety level that all cars must comply with in addition to civil liabilities.

Should be the opposite. Companies would be incentivized to reduce risk if they alone face it.

Consumers don’t have any real information on the risk between a Waymo or a Tesla. So they aren’t in a position to mitigate.

Further, you’d incentive the car companies to cut corners since someone else would pay for it.

We don’t even need a new legal framework for it. We could just apply products liability law.

Total driving liability for a 17-yr old male with a speeding ticket or two (worst case driver) is still insurable for a few hundred dollars a month. There's no material economic difference whether the insurance companies are engaged by the automakers, car owners, or car 'driver/passengers'.

The law certainly makes a distinction, and bigger-pocketed defendants, in practice, are forced to pay more per wrongful death than 17-year-old Johnny with just an auto liability policy.

Not saying it's right, but that's how the law currently works.

Even if the liability does fall on the big players (imo it probably will) and autonomous cars are no safer than a 17 year old, at 10x that is only a few thousand a month in insurable liability - or much less than the cost of paying a cab driver a living wage + benefits.

Never mind that autonomous driving, once technically 'solved', is going to be a socioeconomic freight train that will have the resources to lobby and favorably influence legislation and judicial judgements.

It still doesn't work if the jury awards scale with the market cap of the SDC maker.

Reply on news.ycombinator.com