Uber driverless car in fatal accident

[Tuna]

loafer123 said:

What is important is that they have less accidents than cars driven by humans.

If they do (and they will) then insurance will be cheaper (probably embedded in subscription for car services) and fewer people will be killed and injured.

That is one enormous assumption there. The faith people put in 'clever AI' is completely unfounded at this stage, and things like the Uber trials were all about trying to test the theory. This isn't something you can try in the laboratory, and it isn't a science of absolutes. If you follow AI research, you'll know that a lot of the current state of the art is unpredictable, hard to analyse and distressingly easy to 'fool'.

The little bursts of hype we get from journalists, amazed at a tech demo made under carefully controlled circumstances, hides the fact that this is very much uncharted territory and in some cases little more than smoke and mirrors. Earlier in the week I was laughing at an overexcited YouTube video where some Tesla owners were putting the latest 'autopilot' update to the test. They were so pleased that it can almost handle corners and hills now (but not intersections). Doesn't that speak volumes about just how far we are from truly self-driving cars, and just how misleading the hype we're being fed is?

[RobDickinson]

Yep. Assuming the human safety driver was paying attention and the software too this may have been unavoidable given the pedestrians actions.

It might be the fault of the self driving car also. We just don't know. What we do know is it'll have been recorded in serious detail and if it was the fault of the self driving car it'll be fixed so it doesn't happen again. That never happens with human drivers...

[chris watton]

hutchst said:

Jasandjules said:

It was almost always going to happen. Though I must confess I was expecting a vehicle control failure which caused a fatality for a passenger.

Now, who is liable in criminal law for this death?

Theresa May is the popular answer these days isn't it?

I was thinking more along the lines of Russian bots/hackers!

[fatboy b]

Vanden Saab said:

I doubt it will be any different to any other crash the car's insurance co. will pay out if it is shown to be at fault.

You miss the point. Someone died, so who will be the one in the dock?

[Rovinghawk]

loafer123 said:

Rovinghawk said:

loafer123 said:

What is important is that they have less accidents than cars driven by humans.

That's probably not a high priority to the family of the deceased.

True, but it is for society as a whole.

I'm not sure that makes this death worthwhile for the greater good.

If it had been me or one of my family I'm pretty certain I wouldn't be looking at a potential utopian future so much as the fact that imperfect technology killed someone today.

[RobDickinson]

Rovinghawk said:

I'm not sure that makes this death worthwhile for the greater good.

If it had been me or one of my family I'm pretty certain I wouldn't be looking at a potential utopian future so much as the fact that imperfect technology killed someone today.

It'll certainly be imperfect but we are far from attributing any blame here yet.

It looks like the cyclist was cutting across the road where he shouldn't have been and not looking at traffic.

[Blaster72]

Most reports are saying it's a female pedestrian that's the victim with the only person in the car being a female Uber employee behind the wheel.

I wonder how long it'll be before the Uber PR machine rolls into action and tries to character assassinate the victim?

[loafer123]

Tuna said:

loafer123 said:

What is important is that they have less accidents than cars driven by humans.

If they do (and they will) then insurance will be cheaper (probably embedded in subscription for car services) and fewer people will be killed and injured.

That is one enormous assumption there. The faith people put in 'clever AI' is completely unfounded at this stage, and things like the Uber trials were all about trying to test the theory. This isn't something you can try in the laboratory, and it isn't a science of absolutes. If you follow AI research, you'll know that a lot of the current state of the art is unpredictable, hard to analyse and distressingly easy to 'fool'.

It isn’t really much of an assumption.

If they can’t make them safer than humans, they won’t be allowed to be used by the general public.

[RobDickinson]

Looks to me like it was a cyclist , the uber car was in the right hand lane and the cyclist came from the right.

[xjay1337]

Oh dear. Nasty way to go.


Still one does ask the question driverless cars seemingly cannot deal with this type of scenario.

[RobDickinson]

We don't know what kind of scenario it is yet.

How would you expect any car to cope with a cyclist riding out in font of it from the side at 20mph?

[Blaster72]

NTSB are saying pedestrian.

https://www.ntsb.gov/news/press-releases/Pages/NR2...

Perhaps she was pushing her bicycle across the road, maybe not. At least the Uber car will have plenty of cameras and footage to help investigate the cause.

[anonymous-user]

Setting up a tripod laser measure whilst leaving the body in place is all rather odd. That's not the way a UK police officer would manage the situation.

[RobDickinson]

jsf said:

Setting up a tripod laser measure whilst leaving the body in place is all rather odd. That's not the way a UK police officer would manage the situation.

I think that's just bags of stuff on the handle bars?

[Blaster72]

If you mean by the bike in Robs photo above, I thought that was just some bags.

[TwigtheWonderkid]

loafer123 said:

No-one believes that self-driving cars will not have accidents.

What is important is that they have less accidents than cars driven by humans.

I don't think that's true. I've lost someone close to me in an RTA. It was awful, it was human error, the driver was also someone I know (not that close) and the passenger who was killed was the person I was close too. But it was an accident, it happens, I accepted it and hold no malice towards the driver. She is a nice woman who made a mistake whilst driving.

But if I lost a loved one to computer error, that would be completely unacceptable.

[anonymous-user]

ah, ok, looked like a head and two shoulders on my screen.

[RobDickinson]

TwigtheWonderkid said:

I don't think that's true. I've lost someone close to me in an RTA. It was awful, it was human error, the driver was also someone I know (not that close) and the passenger who was killed was the person I was close too. But it was an accident, it happens, I accepted it and hold no malice towards the driver. She is a nice woman who made a mistake whilst driving.

But if I lost a loved one to computer error, that would be completely unacceptable.

But we should carry on doing nothing to fix it?

An AI driver making a mistake will have all the data available and it will be fixed so none of those AI drivers ever make that mistake again.

USA loose 30000 people in car accidents a year. Vast majority to driver errors. That number is going up...

[hyphen]

The Selfish Gene said:

wow - what was the idiot monitoring behind the wheel doing?

it was only a matter of time.

It's Uber, cutting corners is what they do.

Other companies both large like Google and smaller ones have not killed anyone despite having undertaken many more journeys... no surprise that it is the tts at Uber who are the first.

[anonymous-user]

fatboy b said:

Vanden Saab said:

I doubt it will be any different to any other crash the car's insurance co. will pay out if it is shown to be at fault.

You miss the point. Someone died, so who will be the one in the dock?

The difference is that objective evidence will be available for this death.

The vast majority of pedestrians killed crossing the road don't get that luxury. The driver just says "sorry i didn't see you" and that's that. This time, there will objective metrics to determine exactly what happened, and how to prevent it happening again.