Tesla and Uber Unlikely to Survive...
Discussion
RobDickinson said:
CABC said:
No AI is going to mitigate against someone jumping in front of a car.
No of course it won't. How could it? If people suddenly decide to leap in from of fast moving vehicles that's their problem we already have laws to deal with that.Fsd just needs to be safer than average human driver. People are not getting better or safer, ai will.
neural nets will get better at predicting pedestrian behaviour than humans. That is technologically at its best. The law will need tweaking a little to make it clearer at police and magistrate level about who is liable. More procedural.
And we'll need a new offence: "deliberately provoking cars to manoeuvre without due cause". Otherwise Teslas are going nowhere on Friday night in town centres!
Can anbody name 1 automatic machine/thing where it’s acceptable for anyone to get injured let alone killed? The nearest I can think of is medicine but that kills the person taking the drug and typically with their consent.
The point being, 2x safer than humans isn’t good enough, 10x isn’t good enough either. It needs to be accident free unless there is a major legal shift.
The point being, 2x safer than humans isn’t good enough, 10x isn’t good enough either. It needs to be accident free unless there is a major legal shift.
Heres Johnny said:
Can anbody name 1 automatic machine/thing where it’s acceptable for anyone to get injured let alone killed? The nearest I can think of is medicine but that kills the person taking the drug and typically with their consent.
The point being, 2x safer than humans isn’t good enough, 10x isn’t good enough either. It needs to be accident free unless there is a major legal shift.
2x safer than a human is better than the alternative surely?The point being, 2x safer than humans isn’t good enough, 10x isn’t good enough either. It needs to be accident free unless there is a major legal shift.
jamoor said:
2x safer than a human is better than the alternative surely?
Surely it depends on in what circumstances it's saferWho's going to like it If its less safe in situations that humans can deal with
eg the Tesla that followed the wrong side of a carriagway marking into the end of a crash barrier
or where there was a mahoosive stationary truck across the way
dukeboy749r said:
There may be two 'elephant in the room' issues though:
1) I, as a human being, still prefer to do most of the things in my life - agreed a degree of automation is acceptable, welcome even, but not at the expense of me deciding and being, you know 'spontaneous'.
2) Since email, bank/credit cards, most IT of things can/already are/may well be hacked - I am even less sure I'd welcome my car being somewhat (if not totally reliant) on external network factors and the (even remote) chance that someone else decides to throw in a glitch.
1) human beings won't even walk 5ft to change the TV channel! Every generation wants things easier. My great grandmother washed clothes by hand. My grandmother washed them in a twin tub. My mother did hers in a proper Washing machine. If I could buy a washing machine that collected washed and ironed my clothes I would.1) I, as a human being, still prefer to do most of the things in my life - agreed a degree of automation is acceptable, welcome even, but not at the expense of me deciding and being, you know 'spontaneous'.
2) Since email, bank/credit cards, most IT of things can/already are/may well be hacked - I am even less sure I'd welcome my car being somewhat (if not totally reliant) on external network factors and the (even remote) chance that someone else decides to throw in a glitch.
2) even a 1970's car can be "Hacked". You simply reach under the car and cut the brake lines.
Tech moves on, and the only issues is people's fear. Everything you take for granted today was the previous generations new scary technology.
Heres Johnny said:
Can anbody name 1 automatic machine/thing where it’s acceptable for anyone to get injured let alone killed? The nearest I can think of is medicine but that kills the person taking the drug and typically with their consent.
The point being, 2x safer than humans isn’t good enough, 10x isn’t good enough either. It needs to be accident free unless there is a major legal shift.
No such thing as risk free. You encounter risks ever second of your life. ‘Safe’ is measured by math. Injury and deaths per million or whatever. Are current collisions and deaths always someone’s fault. The point is the stats will determine if it’s safe. Deaths etc will be dealt with no different to today. What went wrong, who was at fault etc. The point being, 2x safer than humans isn’t good enough, 10x isn’t good enough either. It needs to be accident free unless there is a major legal shift.
Burwood said:
Heres Johnny said:
Can anbody name 1 automatic machine/thing where it’s acceptable for anyone to get injured let alone killed? The nearest I can think of is medicine but that kills the person taking the drug and typically with their consent.
The point being, 2x safer than humans isn’t good enough, 10x isn’t good enough either. It needs to be accident free unless there is a major legal shift.
No such thing as risk free. You encounter risks ever second of your life. ‘Safe’ is measured by math. Injury and deaths per million or whatever. Are current collisions and deaths always someone’s fault. The point is the stats will determine if it’s safe. Deaths etc will be dealt with no different to today. What went wrong, who was at fault etc. The point being, 2x safer than humans isn’t good enough, 10x isn’t good enough either. It needs to be accident free unless there is a major legal shift.
RobDickinson said:
If/when a self driving car has an accident we should 'rip the code apart' and fix it, then one less accident situation.
Every time you do that you make self driving safer.
You cannot do this with human drivers, humans are not getting better at driving.
And so we should, but Musk doesn’t see it that way, he think ‘twice as safe’ is the bar whereas zero is the bar. The only reason AP is on the road is because the driver is ultimately in control, the first time there’s an accident when it’s the computer there will be an investigation, if it’s somebody crossing the road at night, no case to answer, we carry on, but there will be accidents when the car calls it wrong, what then?Every time you do that you make self driving safer.
You cannot do this with human drivers, humans are not getting better at driving.
Guys, stop this simplistic logic. it's more emotional.
4,000 people die on the roads, it's not newsworthy unless it's a multiple death.
one plane crashes and it's a headline.
single death car crashes involving autonomous cars will be newsworthy for years to come.
this is the real reason why airplanes are so safe now. and why cars will eventually be much safer too.
4,000 people die on the roads, it's not newsworthy unless it's a multiple death.
one plane crashes and it's a headline.
single death car crashes involving autonomous cars will be newsworthy for years to come.
this is the real reason why airplanes are so safe now. and why cars will eventually be much safer too.
Heres Johnny said:
RobDickinson said:
If/when a self driving car has an accident we should 'rip the code apart' and fix it, then one less accident situation.
Every time you do that you make self driving safer.
You cannot do this with human drivers, humans are not getting better at driving.
And so we should, but Musk doesn’t see it that way, he think ‘twice as safe’ is the bar whereas zero is the bar. The only reason AP is on the road is because the driver is ultimately in control, the first time there’s an accident when it’s the computer there will be an investigation, if it’s somebody crossing the road at night, no case to answer, we carry on, but there will be accidents when the car calls it wrong, what then?Every time you do that you make self driving safer.
You cannot do this with human drivers, humans are not getting better at driving.
There is no zero harm option. It only has to be measurably better to be accepted.
Take the air crash example. Planes crash, often due to the very systems designed to keep them I the air. We accept that because it's still significantly safer than before. Errors get fixed and planes continue to fly. But we know for certain that there will be another crash and many fatalities. It's unlikely we will every get an entirely safe Aircraft.
98elise said:
That's not the world works. People die every directly or indirectly due to the systems and procedures put in place to make things safer. It's less people than would die without those system and procedures then it's a good thing.
There is no zero harm option. It only has to be measurably better to be accepted.
Take the air crash example. Planes crash, often due to the very systems designed to keep them I the air. We accept that because it's still significantly safer than before. Errors get fixed and planes continue to fly. But we know for certain that there will be another crash and many fatalities. It's unlikely we will every get an entirely safe Aircraft.
If a plane crashes there's a detailed and systematic review, the fault is identified, if its a potential design flaw, even though millions of air miles may have been safely flown, the associated planes are grounded. How long was concord flying before the fuel tank ruptured on a freak accident and they grounded them all until they could improve the protection?There is no zero harm option. It only has to be measurably better to be accepted.
Take the air crash example. Planes crash, often due to the very systems designed to keep them I the air. We accept that because it's still significantly safer than before. Errors get fixed and planes continue to fly. But we know for certain that there will be another crash and many fatalities. It's unlikely we will every get an entirely safe Aircraft.
I'm not saying that unless it can be proven that there will always be zero accidents, I'm saying that zero is the target. We don't look at a plane and think well, I've a 1 in 1,000,000 chance of dying on my flight to london but hey, if I drove it would be 1 in 4,000,000 so thats ok - we tolerate car accidents every day, we don't tolerate a single public transport fatality thats caused by public transport. The extrapolation of this is that accident rates for autonomous cars have to be tiny, and thats thousands of times safer than us driving, and each time it happens there will be hell bent drive to work out what failed and fix it.
All this chit chat about self driving cars, if you have driver assist which is essentially part of autonomous driving it’s pitifull, a real world example..........driving in France along the motor way I’m cruising at motorway speed and the sign for the motor way states 90 km 50 km the cars camara pics this up and displays it as the speed limit rather than the exit speed.
Now as a human I know I want to carry on down the motorway, aha I hear you say but if that was sat navvdriven the car would recognise it needs to reduce speed.
However I know I can negotiate the exit at a slightly higher speed as we all do but the autonomous car would ridgidly stock to the limit .
Now AI is just that it’s a recognition system it has no thought no consciousnesses there are no ifs buts or maybes that is the current state of the technology it’s imperfect and far more so than any human.
Now as a human I know I want to carry on down the motorway, aha I hear you say but if that was sat navvdriven the car would recognise it needs to reduce speed.
However I know I can negotiate the exit at a slightly higher speed as we all do but the autonomous car would ridgidly stock to the limit .
Now AI is just that it’s a recognition system it has no thought no consciousnesses there are no ifs buts or maybes that is the current state of the technology it’s imperfect and far more so than any human.
RobDickinson said:
DonkeyApple said:
And if you can find an article that discusses this in the context of AI cars and not just humanoid robots then I will be genuinely impressed.
Predicting pedestrian road-crossing assertiveness for autonomous vehicle control :https://www.google.com/url?sa=t&rct=j&q=&a...
A Scenario-Adaptive Driving Behavior Prediction Approach to Urban Autonomous Driving:
https://www.google.com/url?sa=t&rct=j&q=&a...
Communicating Awareness and Intent in Autonomous Vehicle-Pedestrian Interaction
https://www.google.com/url?sa=t&rct=j&q=&a...
Edited by RobDickinson on Wednesday 17th October 09:05
So, you’ve just picked three unrelated articles.
None of these articles even touch remotely on human behaviour and how it changes not just between types of human but between humans and machines nor the concept of crowd/herd mentality.
To revert to the original remark, I can see how AVs can operate on simple road networks such as motorways or in simple social environments such as NZ or Canada but I do not see anyone discussing how they will operate in the more complex urban environments such as Central London if fear of injury is removed from the equation.
CABC said:
Mr2Mike said:
RobDickinson said:
This. This is meaningless drivel. It doesnt make sense. You dont make sense, you have zero clue as to what work is being done on this. Billions is being spent on it. Waymo has been doing this for near 10 years.
Yet all you can come up with is that drivel and they might not have thought about something?
So what is the solution they are implementing? DonkeyApple asked the question and you attack him for it, but then fail to provide a single answer apart from a bit of vague hand waving. Poor form.Yet all you can come up with is that drivel and they might not have thought about something?
No AI is going to mitigate against someone jumping in front of a car, unless 10mph is max speed. so they'll be a combination of law and many camera angles to apportion liability. The law will have to accommodate people deliberately jumping out to provoke emergency braking too, bit like shouting "fire" in a theatre.
AP needs testing in Switzerland. They have a habit of walking along, eyes forward, and then suddenly turning 90 degrees by a pedestrian crossing as though to catch a driver out.
As someone who has lived and driven in central London all my life what piques my interest is that there are key junctions and turns where as a driver you must deliberately drive at pedestrians in order to make them stop at the curb or to get out of the way. Every single day you see the driver who hesitates and once one pedestrian keeps going on a red light then hundreds follow and the driver is then stuck for that whole cycle of the lights. In the Square Mile there is a different scenario where City workers will deliberately step out in front of oncoming vehicles that they are almost certain will slow for them. I’ve seen similar in NY, Chicago and many of the highly developed financial centres.
This is the aspect of AV that is of real interest. How humans will interact with machines en masse. Not how these machines will deal with individual situations which is what all these papers seem to be based around.
Absolutely anyone who has ever driven in South London will instantly see the issue that an AV programmed to protect human life not to threaten it would not be able to make any form of progress anywhere where there were pedestrians being watched by police there pedestrians. It’s already perfectly common to have a local deliberately walk out in front of you, staring right at you attempting to ‘fierce you out’ to look well hard to their fellow pedestrians. Some kind of gibbon resides in Essex and plenty of other places.
Many years ago, when I first visited Italy on business, a colleague gave me some sage advice about using road crossings as a pedestrian. In the UK, you look up the road, to make eye contact with the driver(s) and to ensure cars see you, whereas in Italy she assured me, doing this is fatal - if the drivers know you see [i]them[i/], they will carry on because they assume you'll get out of their way. So one set of circumstances with two sets of rules in just one continent.
Digga said:
Many years ago, when I first visited Italy on business, a colleague gave me some sage advice about using road crossings as a pedestrian. In the UK, you look up the road, to make eye contact with the driver(s) and to ensure cars see you, whereas in Italy she assured me, doing this is fatal - if the drivers know you see [i]them[i/], they will carry on because they assume you'll get out of their way. So one set of circumstances with two sets of rules in just one continent.
Once all cars are AP then we won’t need ped crossings as you’ll just be able to walk wherever you want with your eyes closed. 
Gassing Station | EV and Alternative Fuels | Top of Page | What's New | My Stuff