AI Programming to be based on public opinion

[reohn2]

.......I've probably also mentioned this before. Long before self driving cars are prevalent you won't be able to buy a car that isn't capable of a fair degree of collision and hazard avoidance. By then you'll have a pretty good idea of just how safe these vehicles are likely to be.

I agree,it'll be a gradual process,little by little catch a 'monkey' perhaps?

In my experience 99.9% of 'incidents' I encounter daily are the result of having a monkey behind the wheel and *all* of them go away as soon as you put something in control that has patience and a basic understanding of vehicle dynamics.

Exactly!
As it is there's far too many power crazed monkeys behind the wheel IME,take the self important madness out of the equation and you solve the problem for the whole.

[reohn2]

It could be that the 'sacrifice people outside the car' is very rare and there will be other benefits for cyclists that considerably outweigh that.

E.g. if cyclists move out and take up the primary position, will self-driving cars and trucks always wait patiently behind until it's actually safe to overtake, then give good room? If so, then cyclists may quickly learn how to control the cars by our own actions and therefore increase safety as an overall balance.

Disclaimer: I have a doctorate in AI.

I think that cyclists with good road skills do control traffic ATM to a certain extent,and a good AI systm will wait as there's no self important baised head worx going on,presently we have the mad monkey disease to contend with and the problem is vulnerable road users can't tell the mad monkeys from the sane ones

[thirdcrank]

Some of this is almost on us. Amazon has just completed a huge (by anybody's standards but Amazon's) distribution centre just off the A1/M1 link road at J45, which has only been in use a few years (it was built in anticipation of the road serving it being developed from a track to a dual carriageway) is being extensively "improved" presumably with a chunk of dosh from Amazon paid in connection with the planning permission.

Bearing in mind that Amazon seem to be at the front of a lot of transport developments, I cannot see it being long before huge trucks to match the scale of that building are trundling around the motorway network with minimal human involvement.

I was reading recently about a new flexible bus system being built somewhere like London Docklands, the idea being that it only visited stops and routes where someone had either called the bus from a stop, or requested having got on. That's another system ideal for driverless operation, especially as on that type of development the land is either privately owned or controlled by a development corporation with the power to set their own rules.

However, once you optimise travel in a way like that, you risk throwing into question the whole private ownership of cars concept. Why own/ lease a very expensive bit of machinery that spends the majority of its time stationary and taking up a lot of space if you can summon a robotic one for the duration of the journey? Of course, in urban areas we are there already with taxis/ private hire but that doesn't seem to stop anybody who can afford it having a car or several as well.

It may be that the old gits have got it all wrong, but it will be the marketing people who run this, not the Tefal men. I mentioned the thing about some people liking to feel they have control of their car. On a typical automatic car of the executive type - where the manufacturers make their profits - in addition to a sophisticated automatic transmission, the driver will be able to select a sport mode which stays longer in lower gears to maximise acceleration, select a lower gear by pressing the accelerator right down for a similar purpose and then change up/down with paddles on the steering column. Why keep a dog and bark yourself? In this case, to feel in control. It may be illogical but it sells expensive cars, and IMO, this will influence the trends in cars. The pioneers with the money for the latest gadget will be in the marketing people's sights.

I'd agree that nobody will deliberately increase the risk of crashes but "safety" will increasingly be achieved by things like cycle routes behind bus stops.

https://www.youtube.com/watch?v=bsNwtBv3PI0

[mr bajokoses]

Maybe the car makers can see the writing on the wall for sales of 'premium' cars as they would offer no real-world performance advantage over cheaper and less powerful models. It's probably obvious to many of us on this forum that there is little that high performance cars can do that a modest small car cannot in the real world, but still many car buyers are seduced by the promise of incredible speed and roadholding.

If I were in the car building trade I would be concerned about the effect of a reduction in sales of my most profitable products.

[reohn2]

Maybe the car makers can see the writing on the wall for sales of 'premium' cars as they would offer no real-world performance advantage over cheaper and less powerful models. It's probably obvious to many of us on this forum that there is little that high performance cars can do that a modest small car cannot in the real world, but still many car buyers are seduced by the promise of incredible speed and roadholding.

I see so much of it as "fur coat and no knickers" all that style,bling and power stuck in an overcrowded motorway becoming evermore frustrated by their attempts to 'live the dream'

If I were in the car building trade I would be concerned about the effect of a reduction in sales of my most profitable products.

So much is sold on empty promises and I'm sure the marketeers will come up with a good plan.......

[ChrisButch]

Are there parallels with things that already exist, semi-automatic trains, lathes with cutoff, circuit breakers etc?

I would have thought he closest parallel would be with autopilot and other computer-controlled systems on airliners. I don't know much about them, but there must presumably be elaborate protocols for the circumstances in which automatic controls can or cannot be overridden. I seem to remember reading reports of aviation accidents where manual overriding of an automatic system, or the reverse, were crucial factors.

[chris_suffolk]

I'm probably a bit behind the curve here, but:

- Will real drivers be able to over-ride the AI, thus enabling them to be driven 'normally'. If so, still allows the monkeys to drive as they wish
- Never really understood the whole push towards self driving vehicles. Seems to be technology solving a problem yet to be invented, rather than a solution to an already existing problem - unless I've missed the whole point, which is entirely likely.
- Given that they are plenty of 10,15, 20 year old cars around, we will have a mix of driven and non driven cars for many years yet. And it's during this 'transition' phase that we will have most issues IMHO

[The utility cyclist]

It could be that the 'sacrifice people outside the car' is very rare and there will be other benefits for cyclists that considerably outweigh that.

E.g. if cyclists move out and take up the primary position, will self-driving cars and trucks always wait patiently behind until it's actually safe to overtake, then give good room? If so, then cyclists may quickly learn how to control the cars by our own actions and therefore increase safety as an overall balance.

Disclaimer: I have a doctorate in AI.

Exactly.

What makes me laugh is how this suddenly becomes an important issue. For decades we've accepted that self survival is the driving force for the monkey behind the wheel and that's OK but as soons as we have a computer that can make an actual decision in the time available to it then it becomes some sort of moral dilemma.

Even if the computer is instructed to protect the occupants it'll be no worse than what's out there now and that's before you factor in the reduction in likelihood of the vehicle being in a position where it has to make that decision anyway.
Not only that but if the computer has to make that decision it's unlikely to be a black and white one. If it has time to evaluate the threat it can also work out the odds of success and the numbers of casualties.
By far the cleverest thing here is the imagination applied to thinking up increasingly bizarre and unlikely scenarios that the computer *needs* to be able to solve.

I've probably also mentioned this before. Long before self driving cars are prevalent you won't be able to buy a car that isn't capable of a fair degree of collision and hazard avoidance. By then you'll have a pretty good idea of just how safe these vehicles are likely to be. In my experience 99.9% of 'incidents' I encounter daily are the result of having a monkey behind the wheel and *all* of them go away as soon as you put something in control that has patience and a basic understanding of vehicle dynamics.

I've never accepted it and never did even when I started driving so you can speak for yourself, my problem is not the computers/automation but those simply replicating the thinking we've had for decades with the programming.
Yes automation will bring some benefits but not if those programming the computers are using motorcentric thinking/bias and that of the self same public that are flawed and created the problem in the first place. Any hinty that it'll be no worse is no advancement at all, I want, no demand that we change massively in the direction of the vulnerable being protected not the thinking of sacrificing the vulnerable on a whim that simply replicates what we already have with humans in charge.

People saying that humans will simply risk themselves to hold up motor traffic is a load of baloney and has no basis in fact for such a statement, none that I've seen. It's like the premise that people on bikes deliberately put themselves in harms way, it's utterly preposterous

[old_windbag]

One intelligence not to copy in these cars is that of the cycle path dog walker. Crikey we'd be in serious trouble. I've just returned having experienced the best of idiocy they have to offer .

On a related note you'd think we'd have driverless trains all over by now as thats a hell of a lot easier to implement. Gets rid of a 45k driver on intercity for each train.

[kwackers]

I've never accepted it and never did even when I started driving so you can speak for yourself, my problem is not the computers/automation but those simply replicating the thinking we've had for decades with the programming.

Doesn't matter what you personally do. It's what the majority does that's important.

As for replicating that thinking with programming, how exactly does that work?
Do programmers make a list of bad things and write code to do them? Or do they take a list of rules and implement them?
If the latter where do they get those rules?

You have a bizarre world view of programmers if you think the code they write replicates their internal preconceptions. Not to mention the complexity such programming would require.
Take it from me. Programmers like a well defined rule set. It makes our task massively simpler. In fact what we really desire is a HWC along the lines of the German one with every small eventuality covered, it makes life so much simpler.

People saying that humans will simply risk themselves to hold up motor traffic is a load of baloney and has no basis in fact for such a statement, none that I've seen.

I can't be bothered going back to see who said that but I suspect the true meaning behind it is to simply make better use of primary. I do it all the time, I don't see that it puts me at risk, rather that it reduces it.
I also rather think that if we live with vehicles that are capable of evaluating overtakes properly there'll be little need to take primary or worry quite as much about road positioning as we do now.

[kwackers]

One intelligence not to copy in these cars is that of the cycle path dog walker. Crikey we'd be in serious trouble. I've just returned having experienced the best of idiocy they have to offer .

On a related note you'd think we'd have driverless trains all over by now as thats a hell of a lot easier to implement. Gets rid of a 45k driver on intercity for each train.

Driverless trains will never happen whilst the unions have a grip. It's not the hardest thing to make doors behave safely without several people to watch them but we don't have that either.

As for dog walkers, yep some are a pita no doubt. But having watched the way some cyclists behave around them I bet you don't have to look too hard to find some dog walkers who with some justification think cyclists should be banned from the paths or at least look forward to the day when self riding bicycles make the paths safer for them.

[Annoying Twit]

I've never accepted it and never did even when I started driving so you can speak for yourself, my problem is not the computers/automation but those simply replicating the thinking we've had for decades with the programming.

Doesn't matter what you personally do. It's what the majority does that's important.

As for replicating that thinking with programming, how exactly does that work?
Do programmers make a list of bad things and write code to do them? Or do they take a list of rules and implement them?
If the latter where do they get those rules?

You have a bizarre world view of programmers if you think the code they write replicates their internal preconceptions. Not to mention the complexity such programming would require.
Take it from me. Programmers like a well defined rule set. It makes our task massively simpler. In fact what we really desire is a HWC along the lines of the German one with every small eventuality covered, it makes life so much simpler.

For anything as complicated as driverless cars, it won't be (and isn't) a matter of the programmers coming up with rules. It'll be machine learning and cars will be taught how to drive. And the parts of it based on rules won't be programmers writing the rules, but specialists designing them and programmers implementing them.

This is a number of levels over and above something as simple as, say, Microsoft Word.

[The utility cyclist]

don't insult my intelligence
So you are saying there will be no set parameters, if you don't give a set of parameters/instructions/code then the AI will be given free reign to make unlimited, no boundaries decision making, where does it get the abaility to learn from, how without code(AKA programming) can it learn to start with? so what rules does that end up obeying/following without programming/parameters? Do you know what unrestricted AI will do in a life or death situation, is that one that aligns with civilised thinking of those presenting the greater harm should bear the brunt of the responsibility and also be sacrificed so that the innocent/vulnerable are saved? That's already shown not to be the case with those in charge asking the public what decision making it (AI) should take.

The Guy Martin programme was a perfect example of the supposed intelligence not being able to make a very simple decision because the programmers had not inputted what it needed to be able to make that decision and despite having already spent a lot of time, money and effort were massively short in something a human being would understand easily.
So without the programming/parameters where will will AI go, what direction with respect to decision making in those critical situations occurs, if it decides to sacrifice the innocent then that is clearly IMO an unlawful deliberate act and therefore we cannot trust AI to be an improvement in that aspect (As I said I believe it will be better in others)
So?

[kwackers]

For anything as complicated as driverless cars, it won't be (and isn't) a matter of the programmers coming up with rules. It'll be machine learning and cars will be taught how to drive. And the parts of it based on rules won't be programmers writing the rules, but specialists designing them and programmers implementing them.

This is a number of levels over and above something as simple as, say, Microsoft Word.

I know how it works

But my point which you've reiterated is a foil to old_windbags apparent belief that the rules will magically reflect the personality of the person doing the program.
This is a persistent belief; people aren't perfect therefore machines we make won't be perfect...

The interesting segregation that you touch on above is how the deep learning systems are 'taught' to drive.
Teaching isn't the same as programming and we know if you teach an expert system you can in fact teach it bad form; an example being to teach it to accelerate harshly rather than smoothly etc.
Obviously this isn't the same as teaching it to obey the rules since rules are not 'taught' they're hard coded.

[kwackers]

don't insult my intelligence
So you are saying there will be no set parameters, if you don't give a set of parameters/instructions/code then the AI will be given free reign to make unlimited, no boundaries decision making, where does it get the abaility to learn from, how without code(AKA programming) can it learn to start with? so what rules does that end up obeying/following without programming/parameters? Do you know what unrestricted AI will do in a life or death situation, is that one that aligns with civilised thinking of those presenting the greater harm should bear the brunt of the responsibility and also be sacrificed so that the innocent/vulnerable are saved? That's already shown not to be the case with those in charge asking the public what decision making it (AI) should take.

The Guy Martin programme was a perfect example of the supposed intelligence not being able to make a very simple decision because the programmers had not inputted what it needed to be able to make that decision and despite having already spent a lot of time, money and effort were massively short in something a human being would understand easily.
So without the programming/parameters where will will AI go, what direction with respect to decision making in those critical situations occurs, if it decides to sacrifice the innocent then that is clearly IMO an unlawful deliberate act and therefore we cannot trust AI to be an improvement in that aspect (As I said I believe it will be better in others)
So?

Insult your intelligence? Do you have experience of this sort of thing? If not then why does my explaining how it works insult your intelligence?
I don't consider my intelligence insulted when someone explains something to me.

If you want to take everything you know about a subject from an entertainment program, be my guest but don't fall into the trap of thinking you suddenly have acquired a deep understanding.

If it helps the "AI" knows how to drive, that's all.
The rules aren't taught, they're hard coded.
In a life or death situation the rules set what's required and the AI drives the car to the best of its ability.

When you drive most of what you do is done subconsciously; that's effectively what the AI does. The conscious bit of you that reads road signs and makes sure you stay within the rules, that's what the programming does and that's the hard coded bit.