autonomous cars

bazster

IanRi wrote: »

Computers crash, seize up and go wrong!

But with nothing like the frequency that human beings do.

bazster

New technologies tend to develop at an exponential rate. This is going to happen far sooner than some of you think.

andyfr_2

You will always have the resistance of the drivers who do still want to drive.

almillar

Humans, even lazy ones, are still way better at handling novel situations than AI

What novel situations? Expect the unexpected? Problem is, lazy humans are often terrible drivers for the non-novel stuff, and that's why they crash so much. For the novel stuff, they won't be paying attention so mightn't be able to do anything about it till it's too late! Aren't 90% of accidents down to driver error or inattention? What if autonomous cars got rid of all of these?!

Anyway if you think you're a better driver than a computer, this probably isn't for you. It's for people who find driving a chore, or scary, not for people who actually enjoy it, and come onto internet forums to talk about it...

bazster

almillar wrote: »

Anyway if you think you're a better driver than a computer, this probably isn't for you. It's for people who find driving a chore, or scary, not for people who actually enjoy it, and come onto internet forums to talk about it...

True dat...

bsms1147

I'm hoping for the day when there's national personal rapid transport-automated car hybrid system. No need for anyone to own their own cars. No need for public transport.

Request a car, enter your destination and away you go.

macman

The maim obstacle is surely not technical, but legal? When your autonomous car does malfunction and mows down a family on a pedestrian crossing, which insurer is going to pay out? Your own, the car manufacturer, or the third party provider who make the hardware, or even the software developer? Current safety systems are usually designed only to intervene at a late stage when the driver clearly isn't going to react in time: for example the various emergency city braking systems now available only activate 'late and uncomfortably', the intention being not to encourage drivers to become reliant on them.
Handing over complete responsibility to the software from start to end of the journey is another matter entirely.

welfayre

bazster wrote: »

As Tobster86 says the machine would do what it was programmed to do. The value decisions would be made by the designers, not the machines, and would be based on a huge range of factors, many of them statistical, carefully considered at great length.

Whereas a human driver faced with the scenario outlined would be totally incapable of weighing up the situation and coming to a reasoned conclusion in a split second. He or she would simply do something totally irrational in a blind panic.

Which is why I said in my first post that I don't think it will happen for a long time. I'm not arguing that it can't/won't happen just that there's a lot of issues, not just technical, to be addressed first.

In my scenario a human panics and kills the child, the police investigate and decide whether there's any blame with the human and either prosecute or don't and the drivers insurance company takes care of compensation. It happens daily and we are conditioned to accept it as part of modern life

If an AI car decides to hit the child what happens? The police can't investigate a machine, the insurance company will argue they can't be held liable for something that's pre programmed to happen. The victims family just get told "sorry your son/daughter's been killed but some guy from VW wrote a programme that decided your kids life was worth less than the guy in the car." It'll take a while for the general public to accept that as part of life.

victor2

welfayre wrote: »

Which is why I said in my first post that I don't think it will happen for a long time. I'm not arguing that it can't/won't happen just that there's a lot of issues, not just technical, to be addressed first.

In my scenario a human panics and kills the child, the police investigate and decide whether there's any blame with the human and either prosecute or don't and the drivers insurance company takes care of compensation. It happens daily and we are conditioned to accept it as part of modern life

If an AI car decides to hit the child what happens? The police can't investigate a machine, the insurance company will argue they can't be held liable for something that's pre programmed to happen. The victims family just get told "sorry your son/daughter's been killed but some guy from VW wrote a programme that decided your kids life was worth less than the guy in the car." It'll take a while for the general public to accept that as part of life.

Similarly, there will inevitably be the AI equivalent of a motorway pile-up in fog, and the ensuing arguments of "it was unavoidable" against "humans would have avoided it".

bsms1147

victor2 wrote: »

Similarly, there will inevitably be the AI equivalent of a motorway pile-up in fog, and the ensuing arguments of "it was unavoidable" against "humans would have avoided it".

Why? Autonomous cars don't need to 'see' like we do, why wouldn't they be better in fog, or in the dark, than us?