I think that will be a given.
You don't have empty captain's seats when the autopilot is on.
Is it? It should be, but many people here think these cars will be running around without anyone in them to take over if needed. For me, they're living in dreamland.
Once the technology is mature, and accident rates have dropped (and its rollout is extensive), then yes.
I imagine accident rates will be lower with autodrive cars than they are now with more fallible humans.
If the day comes when there is only these driverless cars on the roads, and they have been perfected, then maybe, but that's not going to happen is it?
They will be few and far between, and they wont be immune to the dimwits driving around anyway. What will they do if someone overtakes straight into their path? One small example of many.
If the day comes when there is only these driverless cars on the roads, and they have been perfected, then maybe, but that's not going to happen is it?
They will be few and far between, and they wont be immune to the dimwits driving around anyway. What will they do if someone overtakes straight into their path? One small example of many.
Exactly! Like my child running out into the road example, there will come a time when a driverless car has to make a "who / what" to hit decision. It's quite possible that in the majority of cases it will make a better decision than a human driver resulting in the maximum number of people living; however a computer doesn't have the selfish genes of a human and may sacrifice its owner to save 5 other people...
Exactly! Like my child running out into the road example, there will come a time when a driverless car has to make a "who / what" to hit decision. It's quite possible that in the majority of cases it will make a better decision than a human driver resulting in the maximum number of people living; however a computer doesn't have the selfish genes of a human and may sacrifice its owner to save 5 other people...
The car may of detected the child and alerted the other driverless cars of the childs location so they've all slowed down enough that when they run into the road they're half expecting them anyway so they can stop in time
Can you offer an example, without putting ideas in people's heads?
I'm genuinely intrigued.
You mean stuff like people getting "hacked roms" from the internet which'll allow them to twiddle the settings, drive above the speed limit or corner at higher speeds etc?
Must admit, the last time this came up, people were jawing about how marvellous it'd be to have a car drop you off and then go and "cruise" or find itself a place to park.
My first thought, there, was that such vehicles would make an easy target for thieves.
You spot a driverless car, you step out into the road in front of it, knowing that it'll stop, and a couple of your mates put the windows through and rummage around inside it while it obediently waits for you to get out of the way.
And good luck if your self-driving BMW decides to park itself in an out-of-the-way spot all night in Liverpool or Manchester.
No I can't offer any examples but I have a strong feeling that driverless cars out on ordinary roads will be prime candidates for the law of unintended consequences. Once bad things start to happen, whether or not it's the system - or how people use or misuse them, public opinion will rapidly turn against them. The media will highlight all such cases and whatever the safety or crime statistics prove will not matter.
Sooner rather than later, once they get out there in the hands of ordinary motorists, things will happen that absolutely turn the masses against them. Some of the possibilities are already being discussed in this thread but there are many other possibilities too. People are naturally extremely suspicious of the possibility of "death or crime by robot/computer" and I think those fears will soon come to the fore once the things are let loose on and for use by the general public.
Should a driverless car even attempt to take any action other than slam on the brakes and stop in a straight line once it has decided that a collision with someone or something is unavoidable?
Could present some interesting legal scenarios for the manufacturers if they start having to defend the moral programming of their systems!
All google need to do is teach it the 3 laws of robotics and we'll be fine (but does remember a scary robot car story from Asimov about how some robotic cars managed to bypass the laws to kill someone)
If the day comes when there is only these driverless cars on the roads, and they have been perfected, then maybe, but that's not going to happen is it?
They will be few and far between, and they wont be immune to the dimwits driving around anyway. What will they do if someone overtakes straight into their path? One small example of many.
They will react a damn sight better than you or I would react. That's the whole point of them - they can calculate changes in speed, distance and new objects in the field of vision within microseconds, rather than the half-a-second or so it takes a human brain to react, and apply all the controls exactly as necessary to avert an accident.
Indeed this technology is already incorporated into higher-end trucks on our road. If someone breaks in front of them or someone moves into its path, the truck will apply just enough braking to ensure it can never hit the vehicle in front. This is deemed to be far safer than relying on the dopey drivers to do it.
They will react a damn sight better than you or I would react. That's the whole point of them - they can calculate changes in speed, distance and new objects in the field of vision within microseconds, rather than the half-a-second or so it takes a human brain to react, and apply all the controls exactly as necessary to avert an accident.
Indeed this technology is already incorporated into higher-end trucks on our road. If someone breaks in front of them or someone moves into its path, the truck will apply just enough braking to ensure it can never hit the vehicle in front. This is deemed to be far safer than relying on the dopey drivers to do it.
True but like ABS, that's an example of providing specific assistance to the human in control, or limiting what he can allow (like speed limiters) in specific situations. Driverless cars go far beyond that and will inevitably occasionally get their decision-making 'wrong' in the eyes of the masses, or some people will find a way to misuse them. That's when the shit will hit the fan.
Insurance is based on risk and so far driverless cars have been shown to be far safer than human drivers so the risk to the insurance companies would be less which means the insurance premiums would be less.
If the car does make a mistake and causes an accident then the manufacturer would be at fault just as they would at the moment if there is a mechanical failure.
In the event of an accident the car would have a complete video log of the events leading up to the accident so it would be far easier for the insurance company to establish blame than it is at the moment.
Insurance companies would have to allow for the possibility that the car could be involved in an incident which led to a claim, even if the incident was not the insured's fault. And - unless insurance practices changed radically - you would still have to have a policy holder, other named drivers etc so some of the same factors would apply. For example, occupation of the policy holder and annual mileage: someone who only drives a few miles to the shops once a week would still be a better risk than someone who drives a fifty mile round-trip to and from their job as a journalist every day (I think journalism is one of the jobs that puts up premiums). These aren't factors that necessarily make the driver more likely to have an accident, they're factors that statistically correlate with the driver being more likely to have an accident, so we would have to wait for the statistics to change before seeing premiums fall for drivers of expensive cars who do high mileages for "risky" jobs.
Considering how much manufacturers charge for an optional Sat-Nav or CD changer, I'd imagine that all the additional sensors and computing power will be a tad expensive ... with depreciation being an absolute nightmare!
Sooner rather than later, once they get out there in the hands of ordinary motorists, things will happen that absolutely turn the masses against them. Some of the possibilities are already being discussed in this thread but there are many other possibilities too. People are naturally extremely suspicious of the possibility of "death or crime by robot/computer" and I think those fears will soon come to the fore once the things are let loose on and for use by the general public.
Uhuh.
I mean, a PC game is a fairly straightforward bit of software compared to an AI system that'd control a vehicle on a public road. Games get beta-tested and yet they're still released with huge, obvious, gaping flaws that anybody with half a brain can spot.
Unless you can test every possible scenario with every possible variable you really have no idea whether or not the system is capable of dealing with every situation it might encounter.
And, given the uproar over Toyota floor mats, how do we think people will respond when it turns out that a self-driving car loses it's shit when, say, it's maneuvering on black ice or when it's approached from different directions by several vehicles at a complex junction?
Out of interest, does anyone know how driverless cars deal with everyday situations that require interaction between one driver and another? For example, every time I drive to or from my house, as the road is only wide enough for one vehicle except where there are empty parking spaces, I have to make decisions about whether to give way to oncoming drivers or expect them to give way to me. Those decision might be based on factors like, how far is it to the nearest empty parking space (so who has to reverse the furthest), which vehicle is the bigger (so is the nearest space big enough to allow one car to squeeze past the other) and how happy the other driver looks about having to back up a couple of hundred metres. And, of course, whether the other driver flashes his lights to say it's OK for me to keep going.
So, how do driverless cars gather and assess all this information?
Unless you can test every possible scenario with every possible variable you really have no idea whether or not the system is capable of dealing with every situation it might encounter.
And, given the uproar over Toyota floor mats, how do we think people will respond when it turns out that a self-driving car loses it's shit when, say, it's maneuvering on black ice or when it's approached from different directions by several vehicles at a complex junction?
Yes, or if a car in front of it on a level crossing unexpectedly stops or reverses, or an oncoming car suddenly blocks its way and the lights then go red for a train. Me or you would either reverse back pronto, or squeeze up tight to the other car's bumper or at its side, scraping it if necessary or even smash through the barrier to escape if a train was imminent. There are many unlikely scenarios like that which, taken together, represent a significant risk to its passengers or passengers in another vehicle, train etc.
These kind of accidents might not be numerous enough to change the overall "driverless cars are safer" statistics but they sure as hell would panic the general populus!
They will react a damn sight better than you or I would react. That's the whole point of them - they can calculate changes in speed, distance and new objects in the field of vision within microseconds, rather than the half-a-second or so it takes a human brain to react, and apply all the controls exactly as necessary to avert an accident.
Indeed this technology is already incorporated into higher-end trucks on our road. If someone breaks in front of them or someone moves into its path, the truck will apply just enough braking to ensure it can never hit the vehicle in front. This is deemed to be far safer than relying on the dopey drivers to do it.
Will they see a potential idiot about to do something daft, by noticing what the driver is looking at, or doing? Will they be able to tell by another vehicles movement that it is best to take some action before the situation develops?
Comments
Fair enough but if it starts singing "Daisy" I'm out.
I think that will be a given.
You don't have empty captain's seats when the autopilot is on.
Look Dave, I can see you're really upset about this. I honestly think you ought to sit down calmly, take a stress pill, and think things over.
Is it? It should be, but many people here think these cars will be running around without anyone in them to take over if needed. For me, they're living in dreamland.
Once the technology is mature, and accident rates have dropped (and its rollout is extensive), then yes.
I imagine accident rates will be lower with autodrive cars than they are now with more fallible humans.
If the day comes when there is only these driverless cars on the roads, and they have been perfected, then maybe, but that's not going to happen is it?
They will be few and far between, and they wont be immune to the dimwits driving around anyway. What will they do if someone overtakes straight into their path? One small example of many.
Didn't people say that about cars?
what would you do, in that situation?
Exactly! Like my child running out into the road example, there will come a time when a driverless car has to make a "who / what" to hit decision. It's quite possible that in the majority of cases it will make a better decision than a human driver resulting in the maximum number of people living; however a computer doesn't have the selfish genes of a human and may sacrifice its owner to save 5 other people...
The car may of detected the child and alerted the other driverless cars of the childs location so they've all slowed down enough that when they run into the road they're half expecting them anyway so they can stop in time
No I can't offer any examples but I have a strong feeling that driverless cars out on ordinary roads will be prime candidates for the law of unintended consequences. Once bad things start to happen, whether or not it's the system - or how people use or misuse them, public opinion will rapidly turn against them. The media will highlight all such cases and whatever the safety or crime statistics prove will not matter.
Sooner rather than later, once they get out there in the hands of ordinary motorists, things will happen that absolutely turn the masses against them. Some of the possibilities are already being discussed in this thread but there are many other possibilities too. People are naturally extremely suspicious of the possibility of "death or crime by robot/computer" and I think those fears will soon come to the fore once the things are let loose on and for use by the general public.
Could present some interesting legal scenarios for the manufacturers if they start having to defend the moral programming of their systems!
If you could only rent them, then people wouldn't be able to mess around with the equipment..
They will react a damn sight better than you or I would react. That's the whole point of them - they can calculate changes in speed, distance and new objects in the field of vision within microseconds, rather than the half-a-second or so it takes a human brain to react, and apply all the controls exactly as necessary to avert an accident.
Indeed this technology is already incorporated into higher-end trucks on our road. If someone breaks in front of them or someone moves into its path, the truck will apply just enough braking to ensure it can never hit the vehicle in front. This is deemed to be far safer than relying on the dopey drivers to do it.
True but like ABS, that's an example of providing specific assistance to the human in control, or limiting what he can allow (like speed limiters) in specific situations. Driverless cars go far beyond that and will inevitably occasionally get their decision-making 'wrong' in the eyes of the masses, or some people will find a way to misuse them. That's when the shit will hit the fan.
Insurance companies would have to allow for the possibility that the car could be involved in an incident which led to a claim, even if the incident was not the insured's fault. And - unless insurance practices changed radically - you would still have to have a policy holder, other named drivers etc so some of the same factors would apply. For example, occupation of the policy holder and annual mileage: someone who only drives a few miles to the shops once a week would still be a better risk than someone who drives a fifty mile round-trip to and from their job as a journalist every day (I think journalism is one of the jobs that puts up premiums). These aren't factors that necessarily make the driver more likely to have an accident, they're factors that statistically correlate with the driver being more likely to have an accident, so we would have to wait for the statistics to change before seeing premiums fall for drivers of expensive cars who do high mileages for "risky" jobs.
You also have half a dozen people on the bridge, keeping an eye on stuff in case the autopilot screws up.
Besides, the sea is a big place so there's a greater margin for error.
Uhuh.
I mean, a PC game is a fairly straightforward bit of software compared to an AI system that'd control a vehicle on a public road. Games get beta-tested and yet they're still released with huge, obvious, gaping flaws that anybody with half a brain can spot.
Unless you can test every possible scenario with every possible variable you really have no idea whether or not the system is capable of dealing with every situation it might encounter.
And, given the uproar over Toyota floor mats, how do we think people will respond when it turns out that a self-driving car loses it's shit when, say, it's maneuvering on black ice or when it's approached from different directions by several vehicles at a complex junction?
So, how do driverless cars gather and assess all this information?
Yes, or if a car in front of it on a level crossing unexpectedly stops or reverses, or an oncoming car suddenly blocks its way and the lights then go red for a train. Me or you would either reverse back pronto, or squeeze up tight to the other car's bumper or at its side, scraping it if necessary or even smash through the barrier to escape if a train was imminent. There are many unlikely scenarios like that which, taken together, represent a significant risk to its passengers or passengers in another vehicle, train etc.
These kind of accidents might not be numerous enough to change the overall "driverless cars are safer" statistics but they sure as hell would panic the general populus!
Will they see a potential idiot about to do something daft, by noticing what the driver is looking at, or doing? Will they be able to tell by another vehicles movement that it is best to take some action before the situation develops?
Did they? If so, it didn't happen in their lifetimes, and this wont happen the way some seem to think it will in most our lifetimes.
It seems some think cars will be racing around the country next year, with people asleep in them, and not causing any problems. I don't think so.
It may well be some limited use will be available, but I cant see them not having to have a driver behind the wheel.