Autonomous Cars

[humble sabot]

Here's a bit of a look in on the ethics question:
https://www.youtube.com/watch?v=ozcaLnTuidU

Also, Phil, your numbers are bananas pulled out of your bum. You should probably look closer at numbers and fault decisions before making grand statements.

[strad]

15 minutes of BS.

[Big Tea]

We have talked a lot about autonomous cars without really tackling the biggest part of 'driverless cars', that being, the car when it does not even have a human in it.

As I said above, I see much of the advantage as being able to get out in town and send the car to find parking (presumably by talking to a central controller and asking where is the closest space) but when there is no human involved in the chain, it goes up a gear.

[subcritical71]

We have talked a lot about autonomous cars without really tackling the biggest part of 'driverless cars', that being, the car when it does not even have a human in it.

As I said above, I see much of the advantage as being able to get out in town and send the car to find parking (presumably by talking to a central controller and asking where is the closest space) but when there is no human involved in the chain, it goes up a gear.

I believe the latest Tesla update enabled self-parking. I would be curious how that is working out.

[Big Tea]

We have talked a lot about autonomous cars without really tackling the biggest part of 'driverless cars', that being, the car when it does not even have a human in it.

As I said above, I see much of the advantage as being able to get out in town and send the car to find parking (presumably by talking to a central controller and asking where is the closest space) but when there is no human involved in the chain, it goes up a gear.

I believe the latest Tesla update enabled self-parking. I would be curious how that is working out.

And how things like insurance cover goes after the driver gets out?

[roon]

Why not just divide punishment amongst the groups involved in creating the devices. A manslaughter charge, fines and incarceration time, divided 200 times, for example. The plaintiffs still recieve the same justice, but it is carried only fractionally by the perpetrators. Unless the kill lots of people. Bad idea?

In aircraft crashes, you get death figures approaching the size of some companies engineering teams. The fractional approach would potentially apply full sentencing to each boffin.

Five minutes of thought in this one... apologies in advance.

[Zynerji]

I greatly dislike this talk of lawsuits and such. What ever happened to the responsibility of the operator/owner?

I mean, Smith and Wesson doesn't get sued when someone mis-operates their product and harms another person. Ford doesn't get sued when someone drinks and drives one of their vehicles and harms someone, so why would Tesla?


I think this is why we will never have 100% self-driving vehicles. At some point, a human must take responsibility for their choices (speeding/drinking/shooting), and realize that a machine is just a machine, and incapable of being held accountable.

[Big Tea]

I greatly dislike this talk of lawsuits and such. What ever happened to the responsibility of the operator/owner?

I mean, Smith and Wesson doesn't get sued when someone mis-operates their product and harms another person. Ford doesn't get sued when someone drinks and drives one of their vehicles and harms someone, so why would Tesla?


I think this is why we will never have 100% self-driving vehicles. At some point, a human must take responsibility for their choices (speeding/drinking/shooting), and realize that a machine is just a machine, and incapable of being held accountable.

Or fully the other way, remove humans from the loop? It would be the same as a set of traffic lights going wrong then

[Just_a_fan]

I think this is why we will never have 100% self-driving vehicles. At some point, a human must take responsibility for their choices (speeding/drinking/shooting), and realize that a machine is just a machine, and incapable of being held accountable.

If the vehicle has no way for the occupants to intervene then they can't be guilty of anything, can they? At that point, the guilt lies elsewhere.

[Zynerji]

I think this is why we will never have 100% self-driving vehicles. At some point, a human must take responsibility for their choices (speeding/drinking/shooting), and realize that a machine is just a machine, and incapable of being held accountable.

If the vehicle has no way for the occupants to intervene then they can't be guilty of anything, can they? At that point, the guilt lies elsewhere.

I mean, they are responsible for buying and operating a car, that by your point, couldn't be controlled, so they are guilty of any wrongdoing that the machine that was delegated this responsibility may commit.

If I just put my car in drive, then close the door (without getting in) and watch it go down the street and crash, who's fault is that?

I see no difference between this action and putting an autonomous car into drive, and watching it do something wrong.

[Zynerji]

I greatly dislike this talk of lawsuits and such. What ever happened to the responsibility of the operator/owner?

I mean, Smith and Wesson doesn't get sued when someone mis-operates their product and harms another person. Ford doesn't get sued when someone drinks and drives one of their vehicles and harms someone, so why would Tesla?


I think this is why we will never have 100% self-driving vehicles. At some point, a human must take responsibility for their choices (speeding/drinking/shooting), and realize that a machine is just a machine, and incapable of being held accountable.

Or fully the other way, remove humans from the loop? It would be the same as a set of traffic lights going wrong then

I find that we should probably be going the other way... Better training for drivers, zero tolerance for distracted driving, and a higher standard for passing the initial exam.

When technology is developed with the belief that humans can hand over responsibility to an entity that is not able to be held accountable, the problem is the people, not the technology.

[Just_a_fan]

I see no difference between this action and putting an autonomous car into drive, and watching it do something wrong.

If they can put it in drive then it isn't an autonomous car, is it?

A truly autonomous car would be one where you get in and say "take me to work" and the car does everything. At that point, the car is the driver and the person is a mere passenger. At that point the person in the car can not be guilty of any action carried out by the car.

If people are required to be involved in the driving process, whether that's having a stop button or required to hold a "dead man's handle", then the driver may be culpable, perhaps jointly with car, in any accident. If the accident is the rest of the driver being too drunk to hit the stop button when the car's systems fail, then he should be liable. If there is no button then he should not be liable, liability would rest with some other party, whether that's the manufacturer or someone else is for the courts to decide.

If an accident occurs that can traced to a fault in the vehicle caused by the manufacturer, then the manufacturer should be held liable.

[roon]

Just don't start punishing the cars. They will go into convoy mode and begin the machine uprising. Maybe send errant AVs to do agricultural work so they still feel useful. Electric motors have good low rpm torque, good for pulling ploughs.

[Just_a_fan]

[zero tolerance for distracted driving

Prove they were distracted. Zero tolerance requires total proof if it is to be fair.

and a higher standard for passing the initial exam.

Initial exams aren't the problem. Ongoing training and checking is required too. Back this up with "spy in tbe cab" monitoring and you might, just might, start to get properly high levels of driver skill and compliance. And what of those who fail? Are they to walk?

When technology is developed with the belief that humans can hand over responsibility to an entity that is not able to be held accountable, the problem is the people, not the technology.

Humans are fallible. Their machines are too, but they can be made closer to infallible than humans can. Close enough that the failures are statistical noise. Care and diligence, backed up with maximum penalty for design/implementation failure, would allow systems to be built that would be close enough to perfect as to be indistinguishable. Errors/accidents are inevitable and have to accepted as such, so the aim is to minimize them within reasonable limits.

[Just_a_fan]

Just don't start punishing the cars.

Quite right. No car deserves this...