Uber driverless car in fatal accident

[98elise]

captainaverage said:

TwigtheWonderkid said:

You may be right, but fatal RTAs are emotional. 1750 people a year die on UK roads. We don't like it but we accept it. If you could cut that by 75%, I don't believe the British public would accept 440 people being killed a year by computer error. They just won't.

I could be wrong, but I doubt I am.

We are more accepting of a human mistake if it's an honest mistake. Computers are expected to be perfect and quite rightly so they should be. No room for a computer mistake regardless of honest or not. Very very restrictive and tightly controlled legislation should and will be needed.

No computers are not expected to be perfect. They are expected to work in the way they are programmed, until they break. Humans do the programming so they are prone to bugs and flaws. If they can do something safer and better than a human then that's the better option.

[anonymous-user]

There seems to be a number of posters that are worried, scared, dismissive, upset, concerned, or all of the above regarding Autonomous vehicles on our roads.

This point of view seems very strange to me as I find the technology exciting, interesting, and potentially very useful.

The idea of autonomous cars doesn't worry me at all.

Can I ask all the naysayers why YOU are worried about it?

[Eric Mc]

98elise said:

If they can do something safer and better than a human then that's the better option.

ALL humans - all 2 or 3 billion drivers ion the world?

Who sets the mark?

What is the mark?

How do we know when it is achieved?

Autonomous cars will arrive at some point in the future - but only when human driven cars no longer exist. I do not think the two can co-exist or that autonomous cars can exist where humans move freely.

[Eric Mc]

Lord Marylebone said:

There seems to be a number of posters that are worried, scared, dismissive, upset, concerned, or all of the above regarding Autonomous vehicles on our roads.

This point of view seems very strange to me as I find the technology exciting, interesting, and potentially very useful.

The idea of autonomous cars doesn't worry me at all.

Can I ask all the naysayers why YOU are worried about it?

And that's the problem isn't it. Those who get all excited about technology are prone to ignoring inherent dangers and risks associated with it.

New technologies are great - but only if they serve to improve our lives and make the world a better place. I am yet to be convinced that autonomous vehicles aer really going to do that for at least a number of decades.

[4x4Tyke]

98elise said:

No computers are not expected to be perfect. They are expected to work in the way they are programmed, until they break. Humans do the programming so they are prone to bugs and flaws. If they can do something safer and better than a human then that's the better option.

Rather like air crash investigations to prevent the same thing happening again, rather easier to do this with autonomous cars than with every human driver.

[Kawasicki]

Lord Marylebone said:

There seems to be a number of posters that are worried, scared, dismissive, upset, concerned, or all of the above regarding Autonomous vehicles on our roads.

This point of view seems very strange to me as I find the technology exciting, interesting, and potentially very useful.

The idea of autonomous cars doesn't worry me at all.

Can I ask all the naysayers why YOU are worried about it?

I am a naysayer - dismissive of the current tech, because I am involved with the current tech. You should be slightly worried, people who develop this tech are.

[Eric Mc]

4x4Tyke said:

Rather like air crash investigations to prevent the same thing happening again, rather easier to do this with autonomous cars than with every human driver.

You don't need an autonomous car to have data recording technology in a car.

[Coolbanana]

Eric Mc said:

Autonomous cars will arrive at some point in the future - but only when human driven cars no longer exist. I do not think the two can co-exist or that autonomous cars can exist where humans move freely.

Riiiiight.

So you have decided in your ***Expert*** opinion that the two cannot exist. Or where humans roam freely.

Let's see it then, show us your own evidence that demonstrably proves your case - no Googling anyone else for support - this has to be your own work, based upon your own tests and logic.

[Eric Mc]

Coolbanana said:

Eric Mc said:

Autonomous cars will arrive at some point in the future - but only when human driven cars no longer exist. I do not think the two can co-exist or that autonomous cars can exist where humans move freely.

Riiiiight.

So you have decided in your ***Expert*** opinion that the two cannot exist. Or where humans roam freely.

Let's see it then, show us your own evidence that demonstrably proves your case - no Googling anyone else for support - this has to be your own work, based upon your own tests and logic.

Silly point.

Autonomous cars are very rare at the moment so meaningful stats don't exist yet. So far, the signs for their practicality and safety don't look that great to be honest.I am just saying that there is along road to go before they are let loose on the world.

I actually think many uber tecchies despise human beings and would rather that all the world was run for and by machines.

[4x4Tyke]

Eric Mc said:

You don't need an autonomous car to have data recording technology in a car.

So find out who is at fault, rather than prevent them re-occurring.

[Tuna]

98elise said:

No computers are not expected to be perfect. They are expected to work in the way they are programmed, until they break. Humans do the programming so they are prone to bugs and flaws. If they can do something safer and better than a human then that's the better option.

Except that modern AI is not 'programmed' in that sense. It's shown thousands of images and told "that's a car", or "that's a person" and it turns out in some cases we don't really know what it is that it actually picks up on.

When an AI system sees a road sign, it doesn't do character recognition to read it, it decides if it 'feels' more like a Stop than a 60 sign. It turns out it's easily confused, and there's a lot of recent research into how little an image has to be changed to be read incorrectly. Faded paint or a few mud spatters can be enough to make the difference, even though we can 'see' it's clearly still a Stop sign.

Another example is the Microsoft image classification tool. It turns out that it's learnt that any picture of a green field with pale objects in it is "sheep". It's amusing to see it classify a field with a few flowers in as "sheep, grazing" - but think about the consequences here. Nothing is "programmed" in the traditional sense, and no human has "checked the algorithm". There isn't one. It's a few million bytes of weights and probabilities that we have trouble even interpreting, let alone debugging. We only find out that it gets it wrong when someone stumbles across a situation where it gets it wrong.

There's a lot of hype in this area, particularly because we're in the middle of a Silicon Valley bubble and there are billions of dollars being handed out to companies in this space. It's a hard problem and there's a winner takes all mentality, not least because it's not a solved problem and no-one has (yet) got an answer for how we get from here (cars mainly get it right in carefully controlled situations) to there (cars consistently get it right in a dense urban environment). The fact that Google have stepped back from promising Autonomy should give people pause for thought. If they say it's too tough to crack right now, that's a pretty big sign that Autonomy is not landing soon.

The company that achieves that is going to make a lot of money. So, in the absence of a proven answer, all the companies out there are gathering as much data as they can, in the hope that there is a tipping point where AI systems get very much better. It's what Tesla, Uber and most of the others are doing at the moment, and the reason for all the driverless trials going on. They're not testing a solution, they're hoping a solution will emerge if they gather enough data.

[anonymous-user]

Eric Mc said:

I actually think many uber tecchies despise human beings and would rather that all the world was run for and by machines.

That is complete and utter nonsense, and you know it.

You are verging on being hysterical over this whole matter.

Yes, I believe people will die at the 'hands' of autonomous vehicles at some point. And you know what will happen? Very little.

The tech will continue to be developed and implemented.

People have been killed in the development of technology ever since we started building things.

Some posters have pointed out that even the people developing the tech are worried, and that's fine. They will therefore only release it into the wild as and when they are less worried.

[anonymous-user]

Anticipation is the key. See school kids you slow down. An algorithm cannot program stuff like that.

[so called]

Kawasicki said:

Lord Marylebone said:

There seems to be a number of posters that are worried, scared, dismissive, upset, concerned, or all of the above regarding Autonomous vehicles on our roads.

This point of view seems very strange to me as I find the technology exciting, interesting, and potentially very useful.

The idea of autonomous cars doesn't worry me at all.

Can I ask all the naysayers why YOU are worried about it?

I am a naysayer - dismissive of the current tech, because I am involved with the current tech. You should be slightly worried, people who develop this tech are.

I'm a naysayer too.
The reason I am currently of this view is quite simple.
I have read on PH in the past where pro-automatous supporters were so convinced of the superiority of this technology and its benefits that they naively ignore reality.
I recall one guy saying, in consideration of all of the variables, possibilities and eventualities, that "they" (car manufacturers) will have considered all and everything that could or may happen = safe as houses.
The 'considerations list' will be subject to a design budget just like everything else.
Risk assessment and mitigation will also have a budget.
Some things will only get 'upgraded' when something (insert death here), happens.
Example - Uber car kills woman crossing the road. Withdraw from service, identify failure, restart service with pomp and assurance.

[Eric Mc]

Lord Marylebone said:

That is complete and utter nonsense, and you know it.

You are verging on being hysterical over this whole matter.

Yes, I believe people will die at the 'hands' of autonomous vehicles at some point. And you know what will happen? Very little.

The tech will continue to be developed and implemented.

People have been killed in the development of technology ever since we started building things.

Some posters have pointed out that even the people developing the tech are worried, and that's fine. They will therefore only release it into the wild as and when they are less worried.

Nope - I've seen the reaction of people on here who are gung ho tecchies. If things go wrong, they always want to blame the human being. It's a fairly obvious trait in many.

The have forgotten the reason why technology exists. It is not there for technology sake - it is to better our lives. If it doesn't do this - it's pointless.

[DonkeyApple]

Eric Mc said:

In 1966 over 6,000 died on UK roads. We've come a long way.

We definitely have and there is still more to go, albeit possibly into diminishing returns now.

But a glimpse into the future suggests two crashed autonomous cars connecting by WiFi to their insurance brokers to talk to two autonomous AI call centre bots and as all four robots have been programmed by IT blokes then none are going to admit they’ve made a mistake, get into another dispute over which Stars Wars film was better until a woman turns up and they reboot at the sight of a pair of boobs.

[Eric Mc]

A very realistic prediction

When the AIs take over, - we're finished.

Where's The Terminator.

[otolith]

Eric Mc said:

The have forgotten the reason why technology exists. It is not there for technology sake - it is to better our lives. If it doesn't do this - it's pointless.

And you don't think that our lives would be improved by reducing death and injury on the roads, or enabling independent personal mobility for those with disabilities which stop them driving or even just by enabling people to go to a country pub for lunch and a few pints?

[98elise]

Thesprucegoose said:

Anticipation is the key. See school kids you slow down. An algorithm cannot program stuff like that.

Why not?

if you detect school kids you slow down, That's not remotely hard to program. More to the point if you do it will always slow down.

The issue at the moment is getting the computers to accurately understand whats being detected. Sensors can detect far better than we can, and computers can think faster, and obey rules far better than we can. When the combined perception of objects and people is better than ours then Autonomous cars will be a no brainer.

[captainaverage]

98elise said:

Have you read the article? At the moment it looks like it was the woman's fault.

Autonomous cars will not guarantee no accidents or fatalities though. They just need to be better than humans. A fatality that is attributed to an autonomous car is highly likely at some time.

Yes I have hence why I mentioned the "random" event. Since most people say autonomous cars are so so much better, they need to be perfect and quite rightly so.
So far considering the number of autonomous cars on the road, they have already been involved in many accidents whether at fault or not, if they're so good they should be able to avoid ALL or 99% of the accidents. Like it or not folks, the fantasy has been slightly delayed