NHTSA Gives Green Light To Self-Driving Cars

New submitter tyme writes: Reuters reports that the U.S. National Highway Traffic Safety Administration (NHTSA) told Google that it would recognize the artificial intelligence in a self-driving car as the "driver" (rather than any of the occupants). The letter also says that NHTSA will write safety rules for self-driving cars in the next six months, paving the way for deployment of self-driving cars in large numbers.

Comments

Instance or class?

[iapetus]

So is each individual instance of an AI a driver? Each version of the software? Each combination of hardware and software?

If a single car is found to be doing something that would have its license revoked, does that car lose its license, or are all Google cars immediately banned from driving? Would a version tweak cause that license to be reinstated, or would Google be out of the self-driving-car business?

Re:Instance or class?

[Anonymous Coward]

If I own a self driving car, is my insurance insuring the AI as the driver ? Is the driving record of that AI individual to my car, or to AI's of that software version ? Can I sue the AI, or am I suing the AI manufacturer. Is the AI the car, or separate from the car? i.e. am I suing Google or Ford ?
 
So many questions - and no clear way ahead. While I like the idea of an automated car, until such a time as ALL private cars are automated, its a terrible idea.

Re:Instance or class?

[Your.Master]

I would imagine that's part of the rules being written over the next six months. Rationally, we know the correct solution is "it depends". A software error could get the license revoked across-the-board until it received an update to resolve the situation, and a mechanical error would get an individual fixed.

Re:Instance or class?

[Anonymous Coward]

The more interesting implication IMO is that since the AI is considered the driver, no occupant will be required by the Feds to pay attention to the road. That is good, it allows people to utilize the commute time for whatever they chose.

Re:Instance or class?

[tnk1]

I wouldn't be surprised if the "driver" was considered a "minor", and the person in the driver's seat was considered to be in an overseeing position with overall responsibility for ensuring that an accident did not occur. Obviously, that would require that the "adult" have the ability to take over safely and at least get the car pulled over to the shoulder or to evade a problem. More to the point, the "adult" would have to be paying attention to some degree.

I doubt that anyone is going to allow the driver to be completely off the hook for this, although they could simply set it up so that the owner was responsible if they were not keeping their car patched.

Re:Instance or class?

[wiggles]

Actually, I've seen answers to all of those questions.

> If I own a self driving car, is my insurance insuring the AI as the driver?

Yes. Google has stated they will assume liability. Other companies pursuing this say the same.

> Is the driving record of that AI individual to my car, or to AI's of that software version ?

This one is actually easier. The insurance industry will have much better figures on the probability of having a claim to pay for the AI drivers, since all those drivers will drive the 'same'. They will be able to say that cars of model X get into .00001 accidents per car per year (or whatever) resulting in $2000 payouts per accident on average (or whatever) and thus will be expected to pay .00001 x $2000 x $INDUSTRY_MARKUP for insurance. Of course it gets a lot more complicated when you have to weigh in modifiers such as the weight of the vehicle (heavier cars cause more damage), the paint job (red cars get more tickets), the environment the car is in (urban cars get hit more), and etc.

> Can I sue the AI, or am I suing the AI manufacturer. Is the AI the car, or separate from the car?

The manufacturer gets sued. The manufacturer would keep insurance and lawyers for these lawsuits.

> am I suing Google or Ford ?

You sue whoever sold you the car. One throat to choke.

Re:Instance or class?

[Coren22]

You'll sue the car, it will then have to get a job as a taxi to work off the judgement.

Re:Instance or class?

[XxtraLarGe]

This one is actually easier. The insurance industry will have much better figures on the probability of having a claim to pay for the AI drivers, since all those drivers will drive the 'same'.

They might always drive the 'same' but the conditions on the road are always different.

Re:Instance or class?

[U2xhc2hkb3QgU3Vja3M]

Shouldn't the "red car" factor disappear with A.I. drivers?

Re:Instance or class?

[fluffernutter]

Why should I accept any accident that is 0% my fault, even if it is only .00001% likely??

Re:Instance or class?

[suutar]

I don't see what you're accepting. If the AI is at fault, Google is paying. If the AI is not at fault, and you're not at fault, the other person's insurance is paying. It doesn't look like you're on the hook unless you personally are at fault, which is as it should be.

Re:Instance or class?

[ShanghaiBill]

Actually, I've seen answers to all of those questions.

Yes, since all of these questions have already been answered, because SDCs are already on the road in many states. The only thing this NHTSA ruling does is help pave the way to private ownership of fully autonomous cars. But we already have corporations operating SDCs on public roads, and we already have private ownership of semi-autonomous cars, such as Tesla Autopilot, where the software is making decisions. So this ruling doesn't really change anything. It is just another incremental step.

Re:Instance or class?

[Joe_Dragon]

keeping their car patched may only be part of the issue.
Like what if ford says your car is 2 years old no more updates buy a new car!

Rented / non owned car?

The state fails to update the software on traffic monitoring systems that sends out bad info.

The firemen in a rush to save people failed to set the road blocked flags in the right way.

The city waterworks crew who was called in on call to fix a pipe did not have an system to flag the road off.

The cable subcontractor is to cheap to buy the flagging system so they do without.

Re:Instance or class?

[Anonymous Coward]

Really? Google will assume liability? How many appeals are we talking?

Re:Instance or class?

[Anonymous Coward]

Yup. And the AI car might actually make it cheap enough to measure that for it to be a factor.

In the UK traditionally insurers asked "Were you convicted of any motoring offences (except parking) in the last 5 years?" and if you said "Yes" they'd ask for details. But although the answer affected your premiums, and in theory your insurance could be invalidated if you lied, they didn't check.

They _could_ have checked, but it meant telephoning an agent at the relevant government agency, going through a security process, waiting while the employee brought up the details. If that cost them £1 per time, it caught 1-in-1000 drivers lying, and catching one lie saved £800 in premiums, they were losing money to bother. So they rarely did unless other red flags appeared.

Then the UK government told the agency to make an HTTPS API that does the check. Instead of a £1 phone call by a human, it's a 100ms round trip HTTP transaction. They do it for every driver, don't even bother asking "Were you convicted ...?" because the government will tell them the answer faster than you can click the button. And the government won't forget that 2000 is now SIX years ago not five, nor will it forget that "Drunk Driving" isn't a type of parking offence even if you were caught in a pub car park at the time, so it gives more reliable answers.

So I'm sure insurers would like to ask "Do you drive like a fucking maniac along unlit country roads?" or "Are most of your trips boring low speed commuter journeys to a local business park?" but today they know they'd never get a useful answer. If they can ask the computer "Tell me your 10th, 50th and 90th percentile speeds while on public roads" and "What percentage of your miles are driven between dusk and dawn?" that might actually make a difference to the premiums.

On the other hand, just as with health insurance, even asking might be bad policy. It turns out it's better to just insure everybody, at a low group price for millions of people, than to have everybody seek insurance individually, most for a bit less but some paying millions of dollars a year due to their high risk profile.

Re:Instance or class?

[SecurityGuy]

For the "driver" to be considered a minor, they'd have to be considered a person at all first. That's a massive stretch. I don't see it happening.

Re:Instance or class?

[Anonymous Coward]

The driver of the "red car" is only half the equation.

Red cars stand out more to officers pointing a radar gun.

Perhaps the "red car" factor will be statistically less significant with A.I. drivers?

Re:Instance or class?

[Bill_the_Engineer]

The manufacturer gets sued. The manufacturer would keep insurance and lawyers for these lawsuits.

This is not a given. The plaintiff has to right to sue all parties that the plaintiff believes contributed to the accident. It's up to the courts to dismiss the suit against the owner of the vehicle or the person determined to be in a position to take corrective action.

I don't believe the court will automatically dismiss the suit since the person in the vehicle still played a role in the accident by not taking corrective action such as hitting the big red stop button. The owner of the vehicle could still be held liable if the court decides that the owner allowed the use of his vehicle by someone unable to take corrective action, but this type of liability is usually dismissed unless the plaintiff can show that the owner knew that the person they allowed to use the car would be unable to take corrective action.

It would probably be more accurate to say that the manufacturer can assume the liability and court costs for the third parties but I think, if you look in the fine print, you'll see that they only assume liability for any defects or deficiencies in the AI software. I bet the owner will still be held liable if the owner didn't maintain the vehicle and its sensor in good working order.

Re:Instance or class?

[Bill_the_Engineer]

I think that the only way Google could make those promises is if they operated as a livery service, where they remain the owners and maintainers of the vehicles and the people subscribe to the transportation service.

Re:Instance or class?

[fluffernutter]

Ok so as long as Google covers my insurance, and the cost of automated cars is not higher here where there is frequently ice on the road as opposed to in California, and I am not liable for any extra costs and I am not open to possible premium increases in what I am paying, then that's totally fine.

Re:Instance or class?

[cerberusti]

My bet would be that the AI will refuse to drive if it is not maintained well enough to do so safely. These systems will be redundant, so one failure should not cause an immediate crash.

Re:Instance or class?

[Anonymous Coward]

Actually, it will get a job as a garbage truck to pick up the trash on the side[s] of the road. Likely in some orange and yellow car wrap.

Re:Instance or class?

[Anonymous Coward]

Considered a minor as in having diminished capacity and requiring a licensed driver paying attention and capable of taking over operation of the vehicle.

Re:Instance or class?

[tnk1]

To be sure, I think the terminology will not be "minor" vs. "adult". I was meaning to imply that the AI might be considered to be similar to a permitted learner or restricted driver who requires certain supervision. I doubt we would attempt to consider the AI to be a person and that is a entirely unnecessary development for this level of computing intelligence.

Re:Instance or class?

[Anonymous Coward]

I think that the only way Google could make those promises is if they operated as a livery service, where they remain the owners and maintainers of the vehicles and the people subscribe to the transportation service.

They could call it 'Goober'....

Re:Instance or class?

[Anonymous Coward]

Now here is an odd question:

Someone runs a light, hits a self-driving car. Because they tend to be small, they will spin around due to momentum in an intersection. It does so, and hits another car. When I witnessed a small car about the size of a Honda Fit or Mini get rammed due to a red light runner, then spin into another vehicle, the small car, even though it was not at fault with wreck #1 got sued by the driver of the car it hit.

So, with multi-car wrecks, how do the vehicle makers handle that, or will the responsibility be on the owner? I asked a lawyer about this, and in a multi-car wreck, like a rear-ender, if his client is car 5 in line, he files his lawsuits against drivers 1-4.

Of course, this is a reason why I don't like small cars -- bearing responsibility for injuries and damage to other parties, because another vehicle had more momentum.

Re:Instance or class?

[Anonymous Coward]

They might always drive the 'same' but the conditions on the road are always different.

That's the beauty of the insurance industry. At a certain point it doesn't matter. You don't need minute by minute predictions about how a car/driver will perform. You only need aggregate statistics over the entire contract period. While you can't predict which automatic cars will be involved in accidents, or under which conditions, the law of large numbers means that you can be pretty darn confident that over a year, say, 0.001% of the cars will be involved in an accident. Higher numbers of accidents due to higher than normal numbers of wet leaves on the road in Ohio in September will be offset by better-than-average visibility on Pennsylvania nights in November.

The major risk you take is if there is a systematic change in the accident rate. For example, if a software patch accidentally causes the self-driving car to reduce its following distance too much under certain conditions. But that's why there's the reinsurance industry (insurance companies take out insurance on their insurance policies!). The reinsurance agencies will take a look at the risk of a accident-increasing regression, and compare it to the chance that patches will improve the safety of the self-driving cars. They'll come up with an overall risk, and issue insurance to the insurance companies to cover the chance of systematic failures.

Re:Instance or class?

[Immerman]

A whole lot of those would seem to fall under the "don't hit things" rules the AI is going to be operating under. You shouldn't have to use any special flagging system, just put out the proper safety cones to block off your work area.

Of course I've seen some really lousy traffic barrier placement too, to the point of following what was to all appearances the proper path, only to find myself clearly on the wrong side of the barriers. That might overtax the AIs "situational awareness", but hopefully it would only take a few accidents before road crews start making sure their barriers are placed unambiguously.

Re:Instance or class?

[lgw]

Someone runs a light, hits a self-driving car. Because they tend to be small, they will spin around due to momentum in an intersection. It does so, and hits another car. When I witnessed a small car about the size of a Honda Fit or Mini get rammed due to a red light runner, then spin into another vehicle, the small car, even though it was not at fault with wreck #1 got sued by the driver of the car it hit.

Anyone can sue anyone for anything. That wont change.

Usually in multi-car accident, only the instigator (or his insurance) will end up paying.

Why would you think self-driving cars tend to be small? The Tesla's pretty heavy as modern cars go. Self-driving will probably come first to luxury cars, and those all tend to be heavy.

Re:Instance or class?

[Anonymous Coward]

Clearly you are not a lawyer. You sue all of them. Let the court sort out responsibility.
Geez newbies.

Re:Instance or class?

[shoor]

Why shouldn't the cost be higher where there is frequently ice on the road if that increases the cost to the provider? Are you expecting people in ice free areas to subsidize you?

Re:Instance or class?

[fluffernutter]

No, it should be for Google to have as part of their insurance policy, with cost distributed across all buyers of the product.

Re:Instance or class?

[stub667]

I wouldn't be surprised if the "driver" was considered a "minor", and the person in the driver's seat was considered to be in an overseeing position with overall responsibility for ensuring that an accident did not occur. Obviously, that would require that the "adult" have the ability to take over safely and at least get the car pulled over to the shoulder or to evade a problem. More to the point, the "adult" would have to be paying attention to some degree.

I doubt that anyone is going to allow the driver to be completely off the hook for this, although they could simply set it up so that the owner was responsible if they were not keeping their car patched.

You are assuming there is a driver seat. The rules are for self-driving cars, and will need to consider vehicles with no adult humans on board whatsoever. How these rules define the liability issues will define the industry, and whether the US has autonomous or semi-autonomous vehicles.

I imagine if Google gets liability for their cars, allowing autonomous vehicles, you won't have a choice about keeping your car patched and screwing with the process the equivalent or cutting another drivers break cables.

Good ...

[gstoddart]

Then Google can pay the insurance, right? This should save consumers billions.

Oh, wait, what ... Google is going to make you pay to insure their product from defects?

Thought so.

Re:Good ...

[MightyYar]

I'm not sure what you are suggesting, or where the savings would supposedly come from. The market price for a car with insurance included would certainly be higher than the market price for a car with no insurance included.

Re:Good ...

[gstoddart]

I'm suggesting if Google is driving, and the passengers are passengers, then why the hell would anybody pay for things like liability insurance for an AI?

Could it be because it's still going to have a "fuck it, you drive" mode which passes responsibility to the human so Google can claim they're not responsible?

A self driving car becomes useful when I can have no controls, and be asleep in the back. I don't pay liability insurance on a bus, train or taxi ... why the hell would I pay it when something created by Google is in charge of driving it?

Re:Good ...

[Anonymous Coward]

I would expect the manufacturer of self-driving cars (*not* necessarily Google) to insure it when it's self-driving, and possible the user to insure it for issues relating to poor maintenance or manual overrides.

It's already possible that car manufacturers put in manufacturing defects of every other possible variety that could lead to car accidents, and they can indeed be liable for those problems. Why would self-driving technology be especially different? Because it's a computer?

Re:Good ...

[Anonymous Coward]

There's a reason it's called a "Driver's License" and not an "automobile operation permit".
http://freeinhabitant.info/travel-rights/dmv-gives-exempt-plate-to-not-for-hire-driver.htm

Re:Good ...

[brianerst]

You routinely buy insurance that protects you from mechanical defect or wear-and-tear. If the defect is egregious enough, lawsuits will be filed (presumably by the insurance companies or as class actions) and you will get some sort of recompense (if only in lower insurance bills than you would otherwise have gotten). You are still making the decision to use an autonomous car, so you are in some sense responsible for that choice. ("Yeah, I know Android Car is a better driver than Microsoft Car, but I got a great deal on it...") This doesn't even get into issues that may arise if the software has a problem dealing with a condition caused by the owner - e.g., your tires are bald or you forced it to drive in conditions it's not rated as capable of handling.

Google will also have counterbalancing claims if you (or the car company using Google's software) don't keep the software updated - there can be no "I preferred to stay on Jellybean" for Android Cars.

Re:Good ...

[gstoddart]

There's a reason it's called a "Driver's License" and not an "automobile operation permit".

There's also a reason it's called a self-driving car.

It's driving, or I'm driving. This isn't Schroedinger's driver.

From your link:

Click on the above, you will see that the âoestateâ requires liability insurance on the Mercedes. Travel by right requires no regulation and that means no requirement for Liability insurance, all you need to understand is that you are liable under common law.

No, because I'm not driving. I'm sleeping in the back.

If this is to be a hybrid model where the car drives until it blames you, then just drive the damned thing yourself.

Re:Good ...

[ooloorie]

Oh, wait, what ... Google is going to make you pay to insure their product from defects?

Google isn't "making you pay" anything. It's your choice to buy or not buy a self-driving car from them and purchase the necessary insurance for it. If you impose liability on Google, they buy their own insurance and add it to the purchase price of the car. Then it's still your choice whether to buy the car or not.

It's the same when you buy a Ferrari: nobody is "making" you buy the car or insure it, but if you choose to buy it, you have to get the required insurance, which may be different than for regular cars.

I suspect pretty soon, the risk from self-driving cars is going to be judged lower than the risk from human driving, meaning that your insurance rates are likely to be lower if you buy a self-driving car. That's going to be even more so for people with prior accidents or who are under 25. So it makes sense to actually price and purchase insurance separately from the car.

Re:Good ...

[Anonymous Coward]

Actually it might not. It's likely Google would be so large it would self-insure and its possible it could be so large with all these self-driving cars that the aggregate cost to insure passed to the individual car could be quite small. To the point where you might not even notice. Consider if self-driving cars are radically more safe than cars with drivers the premiums would drop significantly and this could then be passed on by Google to all of its business activity.
Then boom, no insurance hike of any kind.

There is potential for this to be an example of how economies of scale potentially eliminate a cost from a product in its entirety.
Google might even need to offer it to get people to purchase self-driving cars to begin with.

There is a business case to be made for this to eliminate insurance costs in automobiles. Or at least drastically reduce them.

Re:Good ...

[Anonymous Coward]

No, the federal or state government is going to make you pay to insure your car to indemnify every other driver for mistakes you are liable for, just as they do now.

Google doesn't make the rules regarding mandatory car insurance. Why would you think they did?

If there's a point you're trying to make here, you're failing.

Re:Good ...

[ooloorie]

I don't pay liability insurance on a bus, train or taxi ... why the hell would I pay it when something created by Google is in charge of driving it?

If there is a risk from self-driving cars that needs to be insured, you are going to pay for it one way or another. If you don't buy your own insurance, Google will simply have to add it to the purchase price.

It's similar to malpractice insurance: you may think you force your doctor to pay for it, but they just add it back to your bill.

In the end, it's you who wants to operate the product, the companies providing it aren't charities, so you end up paying for everything: material, labor, and risk of operating it.

Re:Good ...

[Coren22]

That article didn't read like English. What exactly was the goal of the owner of the Mercedes? What was s/he exempt from? Why was the driver surrounded by police and the plate confiscated if it was a legally obtained plate? How did that link have ANYTHING to do with driver's vs operator's license? What is it you are even trying to say with that statement?

Re:Good ...

[gnupun]

Then Google can pay the insurance, right? This should save consumers billions.

Not if Google adds thousands to the car's price tag, effectively making you pay for the insurance.

Re:Good ...

[gmack]

Could it be because it's still going to have a "fuck it, you drive" mode which passes responsibility to the human so Google can claim they're not responsible?

Well no, if it had a "fuck it, you drive" mode then there would need to be no regulatory changes. Also, it's right in the article that Google wants to avoid the driver panicking and taking control of the car when something goes wrong.

Re:Good ...

[IgnitusBoyone]

Of course insurance will be different between a self driving car and a human driving car. Just like it is different between my self and my co-workers based on our driving history. However, the question is what is the insurance for. If I am buying liability insurance for a autonomous object then the rationale like car insurance must be such the vehicles are so likely to be in an accident that I need a method of paying off other individuals before I can recoup my losses from google. This is akin to needing public transit insurance to pay off people hit by a city bus I was ridding. I was not in control of the vehicle at the time of the incident I had no way of prevention the accident from occurring and it would have occurred weather I was on the bus, so the insurance is paid for by the City who operates the bus and not the passengers.

Re:Good ...

[fluffernutter]

It's driving or I'm driving.

I agree.. otherwise your self driving car becomes a game of Russian roulette. Will you have accidents and have to claim responsibility for them? Maybe, maybe not is not a good enough answer.

Re:Good ...

[fluffernutter]

If google pays it then I'm fine with that.. because then it becomes part of what the market will bear. I just don't want my premiums jacked up because there was a wicked snow storm last week and my automated car decided to get into an accident on its own accord.

Re:Good ...

[Anonymous Coward]

Insurance premiums should drop quite a bit once everyone converts to self driving cars... so yah, I'm sure it will be a difference of 100x to insure an AI vs a human being, and teenagers will just be priced out of the market completely... e.g. $100/year to insure your driverless car vs $10000/year to insure your 17 year old kid to drive themselves around.

Re:Good ...

[fluffernutter]

IF that is going to be the scenario, that a car has a 0.0001% risk of getting in an accident and you are responsible for covering it when it happens, then I'll just keep driving myself thanks. I can't see how an automated car would ever sell that way. Too big of a risk.

Re:Good ...

[gstoddart]

That's my major problem with this technology: there's an awful lot vague answers to specific questions.

A "self driving car" means you put little Timmy in it, send him to school, and monitor it on your cell phone to confirm he gets out in the right place and a teacher has collected him ... or it means you come out of a bar, fall into the backseat, and say "home, James" ... or it means grandpa who has lost his vision and his driver's license can get in and say "take me to my doctor's appointment".

No driver's license or legal responsibility for operating the vehicle at all. You are livestock being transported. You're not driving or operating, you simply told it your destination.

This bizarre model in which the car drives, except when it doesn't, and with no clear demarcation between is damned near impossible to make sense of.

If the car decides it's got no idea what to do, and it just says "you're in charge", and before you even know what's happening you're in an accident .. and the logs say "human was driving, his fault", you're screwed. Or, worse, someone builds in code which lies and just says "human was driving" 5 minute before any crash is triggered (so they can avoid liability).

There can't be a gray area between who is in charge and who isn't. And paying for liability insurance when the computer is in charge sounds moronic to me, why would you do that? Are you accepting liability on behalf of the computer or something?

Self-driving-ish cars? Autonom-ish cars? It just seems like everybody is pretending this is a solved issue, and I don't believe it is.

Re:Good ...

[Anonymous Coward]

It's from a Freeman On The Land, basically they're a type of crazy person who doesn't believe in government. They have all sorts of odd ideas, but the common theme is that they shouldn't have any responsibilities, they should just be able to do whatever they like. Sometimes it's easier to let them carry on with their nonsense, but often they eventually cause too much trouble to tolerate and we have to lock them up. They believe a lot of weird conspiracy stuff, often related to superstitions about documents they don't really understand.

For example, some companies in my country include a pre-printed slip with their bills that lets you pay the bill over the counter in a bank very easily. Convenient if you somehow haven't entered the 21st century. But Freemen On The Land think it's actually money, that they should be able to turn the bill into cash ...

Re:Good ...

[ShanghaiBill]

Then Google can pay the insurance, right? This should save consumers billions.

Not if Google adds thousands to the car's price tag, effectively making you pay for the insurance.

SDCs have already driven millions of miles on public roads, and have a far better safety record than human drivers. As software improves, and hardware gets faster, their safety record will get even better. So the cost to insure them will be much lower, regardless of whether the cost of the insurance is incorporated into the car price, or purchased separately.

Consumers will indeed save billions. I would not recommend investing in auto insurance companies. Their business model is due for disruption.

Re:Good ...

[ooloorie]

This bizarre model in which the car drives, except when it doesn't, and with no clear demarcation between is damned near impossible to make sense of.

There is no "this model"; people just don't know yet what form self-driving cars will take. There may well be several different classes, depending on speed, kind of AI, and kind of driver.

There can't be a gray area between who is in charge and who isn't. And paying for liability insurance when the computer is in charge sounds moronic to me, why would you do that? Are you accepting liability on behalf of the computer or something?

In that regard, a self-driving car seems no different from a horse or a dog to me: you are using it for some purpose, but you aren't fully in charge of it since it can act autonomously, and you are liable for any damage it causes. Unlike dogs or horses, self-driving cars at least have a track record (by brand).

Re:Good ...

[drinkypoo]

I'm suggesting if Google is driving, and the passengers are passengers, then why the hell would anybody pay for things like liability insurance for an AI?

You are going to need basic liability insurance no matter what, but it should be a lot cheaper in a car that you're not allowed to drive, because you won't be able to cause an accident.

Could it be because it's still going to have a "fuck it, you drive" mode which passes responsibility to the human so Google can claim they're not responsible?

For the foreseeable future, cars are going to have a human-driven mode, so you're going to need liability insurance for that. If you're willing to let your insurer into your car's data, perhaps they will give you a discount if you don't actually use it.

A self driving car becomes useful when I can have no controls, and be asleep in the back. I don't pay liability insurance on a bus, train or taxi ... why the hell would I pay it when something created by Google is in charge of driving it?

Mechanical failure. Again, your rates should go down if you're not driving, but there will still be opportunities for failure.

Re:Good ...

[drinkypoo]

Self-driving-ish cars? Autonom-ish cars? It just seems like everybody is pretending this is a solved issue, and I don't believe it is.

It isn't, but the cars aren't ready to drive themselves 100% of the time anyway. They need a lot more data. So they're going to put the cars with the technology out there, and start collecting it. Then they'll determine how it's going to work as they go along... and by "they" I mean automakers and regulators alike.

Re:Good ...

[Macdude]

You're arguing a situation that won't happen.

It's simple, if the car runs into a situation it doesn't know how to handle it will come to a stop. At that time the human operator can take control. The car won't just hand off control without warning, in fact the car won't hand off control at all -- the human would have to take control.

Re:Good ...

[gstoddart]

You assert this as a fact. Citation? Or are you just deciding it's true? (If it's true, I'd love to know.)

If the car won't just hand off control without warning, then I should be able to be asleep in the back. If I can't be asleep in the back, then I don't believe what you say.

If it's full stop, change control, start driving ... then I shouldn't physically be in the driver's seat, to make it 100% explicit.

So far your "simple" scenario has yet to be validate by anybody, and so far all these tests require a driver in the seat ready to take controls.

I'm afraid you're arguing a scenario which thus far isn't real.

Re:Good ...

[Whorhay]

That risk is very likely astronomically lower than the risk you, or any human, pose as a driver. So far as I've heard every accident that a SDC has been involved in was clearly the fault of the human driver in the other vehicle.

Re:Good ...

[drew_kime]

It was solved in Demolition Man. Human drives from home to the freeway on-ramp. Engages auto-pilot and the wheel stows itself. Coming up to the exit the wheel comes back out and the human disengages auto-pilot.

Re:Good ...

[fluffernutter]

The difference with that is, of course, it is possible to keep dogs and horses under our control. An automated car is NOT under our control by its very definition and functional purpose.

Re:Good ...

[fluffernutter]

And if the automated car has to come to a stop when there is no safe place to do so and someone slams into it? This becomes the other driver's fault?

Re:Good ...

[Kjella]

I'm suggesting if Google is driving, and the passengers are passengers, then why the hell would anybody pay for things like liability insurance for an AI?

Same reason I can lend my car to someone with a driver's license but no car and thus no insurance of his own. Google's driving but it's still your property and that makes you liable. Say you walk into a store and a light fixture falls on your head. Maybe it's a manufacturing error, maybe it's shoddy work in construction, maybe it's sabotage (unlikely) or whatever. It doesn't matter to you because you sue the store, the store manager can't just pass the buck. Even if they find it was a manufacturing error and the manufacturer is bankrupt he still can't pass the buck. I'm guessing they'll keep it as an insurance because then they can also set the conditions of insurance, like if the car is out of spec in any way it can refuse to drive, it might demand to be kept up to date with the latest driving logic, traffic regulations, road maps and whatnot. A product liability is just that the product was in a normal condition on delivery, not beyond.

Re:Good ...

[fluffernutter]

Not realistic. That solution requires cooperation from the roadway. I think the assumption there is that the entire freeway is designed in a way that interacts with cars as to totally prevent the possibility of an accident. That's just not going to happen.

Re:Good ...

[fluffernutter]

Well.. let's just say I'm waiting for them to start testing on ice covered routes. Then I will decide how safe they appear to be.

Re:Good ...

[GabeGhearing]

teenagers will just be priced out of the market completely... e.g. $100/year to insure your driverless car vs $10000/year to insure your 17 year old kid to drive themselves around.

teenagers do have the highest fatal crash rates, but experience and deteriorating physical capabilities also play roles.

• New drivers of any age are very high risk for a fatal accident fir first 6 months or driving.
• drivers over 80 have more fatal accidents per mile than any other group(including teenagers)

If new human drivers are too expensive to allow to drive, then it shouldn't take long until we don't have human drivers. By old age there is no reason to allow humans to drive anyway.

Re:Good ...

[Anonymous Coward]

It doesn't need to prevent the total possibility of an accident, merely do better than humans. And Google's car is currently outperforming humans without assistance from external sensors or designed roads.

Re:Good ...

[Anonymous Coward]

Of course it's the other driver's fault. The fact that they hit the car in front of them meant they were either driving too fast, following too closely, or not paying attention. Whether the car they hit is automated or not isn't relevant.

Re:Good ...

[The-Ixian]

I am sure that the details will be hashed out over the years. You don't know what you don't know and you can't hold back the ocean so basically do your best and write laws that seem to make sense and then let the courts deal with hashing out the details when we have real-world issues.

Re:Good ...

[demonlapin]

It's your car. If a tire has a blowout and your car runs into something, it's your problem - just as it's your problem if your tree falls on your neighbor's house. Failure to properly maintain it creates the liability. If you don't want the risk, pay someone else (e.g., Uber) to take it. And you'll still have to insure against accidental or malicious damage (e.g., hail, tree limbs, graffiti/keying) on a car you've borrowed money to buy.

Re:Good ...

[Firethorn]

This bizarre model in which the car drives, except when it doesn't, and with no clear demarcation between is damned near impossible to make sense of.

If the car decides it's got no idea what to do, and it just says "you're in charge", and before you even know what's happening you're in an accident .. and the logs say "human was driving, his fault", you're screwed. Or, worse, someone builds in code which lies and just says "human was driving" 5 minute before any crash is triggered (so they can avoid liability).

Hell, they're already doing just the opposite. Remember the Hyundai superbowl commercial? Within a certain speed range the car will emergency brake itself to prevent a collision - and that's with a human driver at the wheel.

Given the VW scandal, I think that car companies are going to be under more intense scrutiny for a while. The only time I've heard about self-driving cars that will toss control to a driver were extreme-alpha builds, manned by professional drivers. Modern self-driving cars have the opposite problem - they're designed to stop safely if there's a problem, and not proceed if they don't understand what's going on.

Tossing control to a driver while traveling at 50+ mph moments before an accident isn't something any professional is going to allow. "Cattle car" is a good analogy. Worst case, it stops safely on the side of the road and buzzes for assistance. That's an acceptable failure mode.

Re:Good ...

[Firethorn]

So far your "simple" scenario has yet to be validate by anybody, and so far all these tests require a driver in the seat ready to take controls.

From all the documentation about google cars I've seen, while a professional driver ready to take control is required, the self-driving car will continue to drive until the driver takes positive action to over-ride it.

That's actually how one of the accidents happened - the google car was braking to a stop, the pro disabled it and hit the gas into the back end of the car ahead of him. The car attempts to keep itself safe. Getting 'stuck' is a bigger problem than getting into an accident.

Re:Good ...

[fluffernutter]

Ok, agreed. Except for the fact that stopping in the wrong place is a moving violation and a ticketable offense. It is difficult for the other driver to defend themselves in that case, but not unheard of.

Re:Good ...

[fluffernutter]

Well, 'better than an average human' is not good enough for me. Perhaps if it gets to the point where there are 1 or 2 accidents a year caused by AI in North America, in all climates we see in North America. Otherwise I'll do the driving, since I am also better than an average human as well. It had better be better then any human can do and then some.

Re:Good ...

[Macdude]

When the car in unable to determine a safe route to travel and stops, if it is in an area that stopping is a ticketable offense the driver can simply take control and drive and avoid getting a ticket.

How likely do you think it is that an autonomous car will have difficulty on a stretch of road where stopping is not allowed (e.g. a bridge or tunnel) where it's not actually unsafe to continue? You won't get a ticket for stopping on a bridge because there is a traffic jam in front of you, for example. Or if you stop because of white-out conditions, for another example. If you get a flat tire on a bridge the car will know you're on a bridge where stopping is not allowed and will drive off the bridge ruining your wheel.

Re:Good ...

[castionsosa]

If I have a 1 in ten thousand chance of getting in an accident with an automated vehicle, those are pretty good odds. On average, in the city I live in, I get nearly hit once every 100 times.

So, taking a 1 in 10,000 risk of an AI flub-up, even if it goes on my insurance, that's fine with me. It doesn't go on my driving record as a conviction, so I don't have to worry about points. With one free driver fsck-up every 3-5 years, where rates don't get hiked even if I am found culpable, I'll take those odds.

Toss in some dash cams, and it reduces those odds even more.

Re:Good ...

[fluffernutter]

I guess it depends how grievously you or your family members are injured. If it is one in ten thousand are scraped and/or bruised then fine. Yet if one in ten thousand are sent to the hospital, I am unlikely to accept those odds.

Re:Good ...

[fluffernutter]

In 'unsafe to stop' I was thinking more of the unpaved shoulders on great lengths of highway where I live, or a shoulder that is covered by a large snow drift.

Re:Good ...

[castionsosa]

True. I'm assuming relatively low speed, due to congestion. Usually metro wrecks consist of rear-enders, light-runners, someone yielding into a car, not a lane, or other fender-benders. However, on country roads, there would be far fewer wrecks... but far more grievous. However, a vehicle's AI usually has fewer factors to figure out in those cases, and can react (brake, swerve, accelerate) far faster than a human can.

Regardless, the odds go up. Yes, one can wind up with the epic fail condition that a human could have avoided where an AI couldn't, but I'd suspect the other way round is far more likely.

Re:Good ...

[fluffernutter]

And I say again, I'm taking a wait and see approach on safety since I live in a very hostile climate for driving.

Slowing down to a stop or an almost stop is one solution, but it is rarely the best solution since it snarls traffic.

Re:Good ...

[Lodlaiden]

Sidebar: Any chance that was an actual experiment to document what would happen? I'm not saying it would be right to jeopardize people like that, but you have to know how the system will react.

Re:Good ...

[fluffernutter]

The thing is, AI bears the burden of being foolproof in this case. People like control. People are always going to feel safer if they have control and therefore more willing to accept accidents from themselves than they are ever going to accept accidents from AI.

Re:Good ...

[Whorhay]

All that does is change the physics calculations which the computer would need to use. Which a computer is far better suited to do on the fly than a person. The car could also easily be equipped to identify dangerous road conditions before the car is actually in the midst of them. And once the car is on ice it would be able to react to changing conditions on a scale of milliseconds while a person at best deals in tenths of seconds.

Re:Good ...

[fluffernutter]

The thing is, my concern is that it will adapt to icy conditions by just slowing down to a walking speed or less, and that that will be terrible for driving. Sometimes you have to spin the tires, and sometimes you have to slide around a bit. If AI can calculate how to turn the tires so that the front wheels slide but the whole car moves where it should go, or how to alter the course so that the tires land in the visible ruts as opposed to staying in the lane, then bravo for AI, it's doing better than a human. But slowing to a crawl at the immediate appearance of ice would be a disaster and prevent anyone from getting anywhere. Also, missing ruts by a bit can send a car out of control and human drivers make ruts that don't stay in the lane. I'm sure there are a million other adjustments that I make in order to get from A to B on a winter day in a reasonable time frame, those are just ones that come to mind.

Re:Good ...

[Anonymous Coward]

But you posited a situation where the card HAD to stop when there was no safe place to stop. So the driver (be they human or robot) can hardly be blamed for stopping in the "wrong" place. If the system is trying to find that driver at fault, then the problem is with the legal system, not the driver. Again, automated or not doesn't matter here.

Re:Good ...

[Firethorn]

Sidebar: Any chance that was an actual experiment to document what would happen? I'm not saying it would be right to jeopardize people like that, but you have to know how the system will react.

Nope, the professional driver was just being a moron. The cars have extensive logging though, so they knew precisely what happened.

Re:Good ...

[kqs]

Otherwise I'll do the driving, since I am also better than an average human as well.

See also the Dunning–Kruger effect.

But sure, drive yourself. No problem. My self-driving car will recognize the cars driven by unsafe humans and avoid them, so you'll still be much safer (since you'll only be in danger from other idiot humans). Seems like a loss for natural selection but otherwise a win for everyone.

Re:Good ...

[kqs]

liability insurance for a autonomous object then the rationale like car insurance must be such the vehicles are so likely to be in an accident that I need a method of paying off other individuals before I can recoup my losses from google

Really? I pay homeowners insurance even though 95% of the possible claims would be caused by weather, or natural disasters, or poor wiring done by an electrician, or poor wiring in a device I didn't manufacture. How is self-driving car insurance any different?

So the insurance is paid for by the City who operates the bus and not the passengers.

No, the insurance is paid by you via your fare. In the same way that if you own a house, you pay the property tax directly. If you rent an apartment, you pay the property tax via your rent. If you eat at a restaurant or shop at a store, some tiny part of your bill pays for the insurance and property tax of the restaurant.

The benefit of self-driving cars is that the insurance is likely to be far, far lower than human insurance. But we'll see; insurance companies are quite good at determining risks, so in a few years we'll see how the insurances compare.

Re:Good ...

[ooloorie]

Why would youcare what the risk is as long as someone is willing to insure it for you cheaply enough? Your insurance for a self-driving car is likely going to be lower than for a regular car.

Re:Good ...

[fluffernutter]

Well if there is a 0.0001% change of me dying in the car from a mistake the car could make then I do care about that. Also, I have a problem accepting liability for an accident I may not be able to do anything about. I'm not willing to accept that it is ok for the car to get in an accident because it is safer than me. As long as it may get into an accident in a situation that I would have been able to handle, I should not have to accept any liability for it. These cars need to be 100% infallible.

Re:Good ...

[ooloorie]

Well if there is a 0.0001% change of me dying in the car from a mistake the car could make then I do care about that

Why? Your lifetime risk of dying in a car accident is about 1.2%. A 0.0001% chance of you dying from a mistake the car could make would be completely negligible.

Also, I have a problem accepting liability for an accident I may not be able to do anything about.

You don't accept liability, your insurance company does. Furthermore, if you drive, you already do the same thing: you pay insurance to cover you and your passengers for the mistakes of other people, like when an uninsured motorist plows into your car.

Re:Good ...

[ooloorie]

If I am buying liability insurance for a autonomous object then the rationale like car insurance must be such the vehicles are so likely to be in an accident that I need a method of paying off other individuals before I can recoup my losses from google.

No, it doesn't mean that at all. It simply means that Google sold you the car without assuming liability. Maybe you're just a student and don't own much, but as people get older, they tend to own more "autonomous objects" that can cause injury to others. That's why many people get umbrella liability policies. Unless you're dirt poor, it's a good idea to have something like that.

This is akin to needing public transit insurance to pay off people hit by a city bus I was ridding.

You are paying for the insurance of that bus, both through your ticket and through your taxes. And you're likely paying a lot more for insurance that way than if you paid for it on a case-by-case basis.

Re:Good ...

[fluffernutter]

If I'm paying for insurance, then I am accepting liability.. or at least lets call it, bearing the burden of it. I am in a car I am not controlling. I might not even be in it, I might have a box in it. I'm not paying a dime to protect myself from that. These people say it can go out into the world, then it can go out into the world. I wash my hands of it.

Re:Good ...

[ooloorie]

If I'm paying for insurance, then I am accepting liability..

No, you are transferring that liability.

I'm not paying a dime to protect myself from that

Well, then don't buy a self-driving car. Better don't buy a regular car, a house, a yard, or a dog either, because owning any of them involves risks that you have no control over at all.

Re:Good ...

[fluffernutter]

Well, I control a regular car, a yard. That's why people have to leash their dogs. So they control them. If I had a gun, I would control that too. If it went off without my control then I sue the gun company. I don't control how my house was built. but that's what building standards are for, so the builder is liable for that part. I control everything in it, and the parts that I don't control are covered by their various vendors. Things are pretty safe these days. See the pattern? The aspects that I control, I pay for insurance on. The aspects that I do not control become the legal responsibility of the supplier. These cars I do not control. Anyway it doesn't matter because if insurance agencies are allowed to charge people what they will want for this, the whole autonomous car idea is a passing fad, because no one is going to pay the vehicle costs with insurance premiums on top.

Re: Good ...

[Anonymous Coward]

No matter how well the car can drive, there will always be fools who are cocksure they can drive better still. Your lot will cause most of the remaining accidents.

Re:Good ...

[ooloorie]

Well, I control a regular car, a yard. That's why people have to leash their dogs. So they control them. If I had a gun, I would control that too. If it went off without my control then I sue the gun company. I don't control how my house was built. but that's what building standards are for, so the builder is liable for that part. I control everything in it,

OK, so you live under the delusion that if anything goes wrong with anything you own, you're just going to sue and recover from the manufacturer. Good luck with that.

Any sane person has insurance coverage to cover these cases, and the insurance company will simply pay out to the injured party and not sue the manufacturer (except in cases of gross negligence).

because no one is going to pay the vehicle costs with insurance premiums on top.

Those insurance premiums are likely going to be lower. But given your gross errors in estimating risk ("if there is a 0.0001% chance"), it's not surprising that you have these misconceptions about insurance and premiums.

I predict, though, once these cars are out, you'll just buy one, like everybody else.

Re:Good ...

[fluffernutter]

Maybe... It really depends how safe they become in winter for me. I just really, really doubt that insurance companies are going to charge that much less to cover them. It's not going to be 'oh these are 1000 times safer so we will charge 1/1000 of a regular car'. It's going to be more like 'these are safer so you can have 10% off the cost of the regular car'. It may even be 'these are new technology, and very expensive to fix, so you must pay 2x the amount of a regular car'.

We'll have to see.

Re:Good ...

[MightyYar]

I agree that scale might make it easier for Google to reduce costs over and above what an individual can get. But I think we're talking about a few percentage points - profit margins for insurance companies are in the 10% range. At the end of the day it will still be a matter of paying for your insurance up-front with the cost of the vehicle or paying as you go.

Re:Good ...

[Lodlaiden]

...Nope, the professional driver was just being a moron...

Ah, more people being people.

Re:Good ...

[dave420]

You sure are better, until you have an accident the AI wouldn't have. That's the thing - human drivers are terrible, yourself included.

Re:Good ...

[fluffernutter]

AI is not human, therefore it is going to get confused in an entirely different situation than a human would. I'm not saying I can drive better than AI, I'm saying that it is entirely conceivable that AI would get confused in a situation that I would not. It's like how chess grand masters need to change their game entirely when they play computers. AI just does not act like a human.

Longer commute, here I come

[Lodlaiden]

If I don't have to lose an hour each way on maintaining moderate concentration, moving out of the suburbs into the country suddenly becomes feasible. Sweet! NHTSA approval is a major milestone in this becoming a reality.

Re:Longer commute, here I come

[Anonymous Coward]

In the Information Age why is it exactly we are physically commuting to work on a daily basis? Talk about fuel waste! The time lost to physically commuting will not be regained by AI. Companies need to embrace telecommuting rather than putting up obstacles to it. Doing so increases their ability to remain competitive in that their available workforce pool increases from those withing physical commuting distance to those with sufficient bandwidth to support telecommuting. Save driving for pleasurable trips on weekends instead of endless windshield time on the same route to/from work on a daily basis.

Re:Longer commute, here I come

[DogDude]

In the Information Age why is it exactly we are physically commuting to work on a daily basis?

To DO things, perhaps? Your idea that (all? most? a lot?) work can be done with a computer and Internet connection is bizarre.

Re:Longer commute, here I come

[Dog-Cow]

Context matters. The AC you responded to was responding to someone claiming that his commute would be productive time if someone else was driving. That suggests work amenable to telecommuting.

Re:Longer commute, here I come

[tnk1]

Well... if it was a salesperson using the car time for taking and making calls, they might still have to travel to their client sites for face-to-face meetings and demos.

The context does sound like an office worker, but it doesn't have to be.

Re:Longer commute, here I come

[Anonymous Coward]

Just be patient. The primates will catch on eventually. They're just a little slow on the uptake. I have to go now: I need to Skype my boss.

Re:Longer commute, here I come

[Anonymous Coward]

The AC you responded to was responding to someone claiming that his commute would be productive time if someone else was driving.

No, he said that it would make living in the country more feasible for him. He didn't say what he'd be doing with that commute time. It could just as easily be that he'd spend that time relaxing.

But even if we assume he was talking about using that time for work...

That suggests work amenable to telecommuting.

Not really, because lots of jobs are only "semi-remotable" due to a mix of duties. Since the OP didn't tell you anything about his job duties or whether they can be performed remotely, you have no business making such assumptions about him.

Context does indeed matter, which is why you should have paid attention to it.

Re:Longer commute, here I come

[LynnwoodRooster]

Some of his work may be amenable to telecommuting, but the rest of it may not be. For example, answering e-mails or writing status reports would work for me, in a telecommute-style role. But making measurements in the acoustics lab (with it's anechoic chamber) is kind of hard to do unless I'm present. I surmise most people here on /. fall in the same category, where there still is a demand for a good portion of your job to be done on-site.

Re:Longer commute, here I come

[Coren22]

Productive could also be "get all your online shopping done before you get to work". It doesn't necessarily mean productive for your employer.

Re:Longer commute, here I come

[ShanghaiBill]

Companies need to embrace telecommuting rather than putting up obstacles to it.

1. Be careful what you wish for. If a job can be done remotely from your house, then it can also be done remotely from Bangalore.

2. Have you ever worked for a company with many employees telecommuting? They tend to be dysfunctional, with many workers out-of-the-loop, and poor coordination. It is surprising how much companies rely on informal communication around the water cooler, or chance meetings in the break room.

Re:Longer commute, here I come

[Ichijo]

You won't even have to pay for parking, because it will drop you off at the entrance and then circle around until you're ready to leave. Or if you work where there's free 1-hour parking, just put your car into "musical chairs" mode. Take that, meter maids!

Re:Longer commute, here I come

[Sir_Eptishous]

Have you ever worked for a company with many employees telecommuting? They tend to be dysfunctional, with many workers out-of-the-loop, and poor coordination. It is surprising how much companies rely on informal communication around the water cooler, or chance meetings in the break room.

Correct you are.
The real key is communication skills. Most people don't have them, or have them just enough to get by.

Face to face communication is by far the best way to communicate critical or time sensitive info. Email, phone calls, etc will work, but they just add time and inconvenience.

Re:Longer commute, here I come

[Sir_Eptishous]

Productive means Civ 5!

Re:Longer commute, here I come

[Lodlaiden]

I was referring to being able focus on tasks that are entirely tele-commutable, but are normally activities that extend the overall day, such as document review, email, "hr tasks". If the time that I currently spend commuting could be used for those activities, I would entertain a longer commute, because the total work related time would go down.

Re:Longer commute, here I come

[delt0r]

Well i tried to go swimming by loading up a few jpgs of beaches. It didn't work out that well.

Insurance?

[U2xhc2hkb3QgU3Vja3M]

So if the A.I. is the driver under the law, who needs to buy the insurance? Me or the company previously known as Google?

Re:Insurance?

[Anonymous Coward]

So if the A.I. is the driver under the law, who needs to buy the insurance? Me or the company previously known as Google?

Google but you wind up paying for it in the vehicle purchase price. Like, duh.

Re:Insurance?

[U2xhc2hkb3QgU3Vja3M]

They'll include 20 years of insurance?

Re:Insurance?

[Anonymous Coward]

So if the A.I. is the driver under the law, who needs to buy the insurance? Me or the company previously known as Google?

You will. You purchased the vehicle in the first place. And your insurance premium will probably be higher too. This will never catch on.

Re:Insurance?

[Anonymous Coward]

They'll include 20 years of insurance?

Why not? It's going to be a luxury item at first so they can easily fold it into the price. And Google is big enough to self-insure, making things much cheaper for them.

Re:Insurance?

[Khashishi]

I think you don't need insurance if you are able to assume the costs of any potential liability. Most people can't do this, but it's no problem for Google.

A.I. for President

[Anonymous Coward]

If it can drive a car, it can lead the country.

Re:A.I. for President

[tnk1]

Let's wait until we see if the cars develop sentience and start gunning for pedestrians before we think about giving AI's the ability to order nuclear weapons launches.

Re:A.I. for President

[Lodlaiden]

I'm willing to double down on both. People that don't look where they are going will be gone and any country acting an appropriate level of stupid will now be film locations for reality tv shows.

this is not AI

[Anonymous Coward]

This is just a piece of software. Nothing intelligent about it.

I thought this was supposed to be a tech site?

Re:this is not AI

[Anonymous Coward]

It is software and hardware... no different than you except probably more intelligent.

Re:this is not AI

[SpeedBump0619]

Well, there are *lots* of competing definitions of AI, so I guess you'd have to provide your definition before we can discuss.

One definition that doesn't agree with you: The ability of a computer or other machine to perform actions thought to require intelligence.

If you had asked someone in the 1970's whether driving a car (correctly) require intelligence, I'm betting they would have said yes. Thus, the self driving car is an artificial intelligence by that definition. From another angle, this is an expert system, which Computer Science has taught in Artificial Intelligence classes since well before this century.

I think you are talking about something now called AGI: Artificial General Intelligence. This program is definitely not AGI, by any definition.

Driver picks the tunes.

[Anonymous Coward]

So does this mean I have to spend the whole ride listening to whatever shitty computer music for AIs the "driver" selects?

Re:Driver picks the tunes.

[dwillden]

Mr. Roboto on an endless loop!

AI as an legal person

[Anonymous Coward]

Is this the first time when AI is recognized as a responsible "person" before the law, or has it happened before? Its getting weird...

Ouch

[ledow]

Wonder how much Google public liability insurance premium just increased by.

Because, sorry, but the "AI" is really just a set of rules still. A set of rules that can't take account of every situation. Sure, it can drive more carefully than a human driver, but it can also make just the same kind of dumb mistakes as a human driver too.

But with the consequence that the first accident of note will result in all kinds of problems for EVERY instance of that model running in EVERY model of that self-driving car, rather than just a single driver being an idiot.

And if "the car" is the driver, then driving points and bans means almost nothing OR they mean the end of a self-driving car when they happen.

Aren't we still at the stage where a self-driving car knows the speed limit only by traffic-sign recognition and/or GPS lookup of a database of streets? One slightly muddy sign on a back road, and the car gets a ticket that Google will pay, or sweep under the rug until someone notices thousands of tickets issued to Google-cars over the years that are conveniently just being paid off rather than resulting in more permanent consequences as they would with a human driver.

Re:Ouch

[Nemyst]

But with the consequence that the first accident of note will result in all kinds of problems for EVERY instance of that model running in EVERY model of that self-driving car, rather than just a single driver being an idiot.

Assuming that accident is the fault of the AI, then you can reasonably expect a patch within a week, a month if the issue is extremely complicated. Good luck fixing human drivers this efficiently.

Re:Ouch

[PPH]

Wonder how much Google public liability insurance premium just increased by.

Nothing. It's called Alphabet now. The self driving car subsidiary has probably been left holding zero assets to handle the possibility that a horrible accident occurs and the victims try suing the company.

Re:Ouch

[Anonymous Coward]

But with the consequence that the first accident of note will result in all kinds of problems for EVERY instance of that model running in EVERY model of that self-driving car, rather than just a single driver being an idiot.

Assuming that accident is the fault of the AI, then you can reasonably expect a patch within a week, a month if the issue is extremely complicated. Good luck fixing human drivers this efficiently.

Which will be impossible to prove. You can blame a driver, but no car company is going to let you blame the AI they sell. The legal ramifications will be off the chart. So who coughs up when the AI has a bug that ends in someone's death? The source is closed, it will not be available, and as vehicles are networked in the wanky world of IoT, they'll all claim your car was hacked and wash their hands of the matter.

Re:Ouch

[Dutch Gun]

Because, sorry, but the "AI" is really just a set of rules still. A set of rules that can't take account of every situation. Sure, it can drive more carefully than a human driver, but it can also make just the same kind of dumb mistakes as a human driver too.

Yes, but at the heart of the algorithm is a big overriding rule of "don't drive off the road or hit anything". That's pretty cut and dried as far as rules go. The car's hardware can literally see in every direction and track everything around it, static or moving. It will react to danger and determine the best course of action even before most humans even recognize there's a problem. Unless there's a really serious flaw in the system, that means at worst the car is going to come to a stop or simply avoid all obstacles when it sees anything dangerous or that it doesn't understand.

Honestly, far more important is this: It won't ever get distracted. It won't drive angry, or intoxicated, tired, on medication, or while putting on lipstick or eating a sandwich. It won't freak out if a wasp gets into the car. It won't turn it's head and yell at the kids to be quiet and stop bouncing around in the back seat. It won't drive recklessly in an effort to impress it's girlfriend.

My prediction: Even the first generation of self-driving cars will be statistically 20 times safer or better than an average human driver (at least in terms of accident fault), and it will rapidly improve as incidents occur and the black boxes are analyzed to determine how said incidents could have possibly been prevented. Eventually, car-related deaths will be relegated to freak accidents, like when a tree falls on a car or an overpass collapses, etc.

Nhtsa!

[Anonymous Coward]

As General George S. Patton said to the krauts!

Re:Nhtsa!

[tsqr]

As General George S. Patton said to the krauts!

That was General Anthony McAuliffe, not Patton.

I won't ever trust an AI driver

[NEDHead]

Until they can flip a bird and show unambiguous road rage

Re:I won't ever trust an AI driver

[delt0r]

Old tech: https://www.youtube.com/watch?... @ 1:37

Re: The Republicans have now killed self-driving c

[Anonymous Coward]

They hate technology.

a way to do this "safely"

[laurencetux]

Phase 1: limit AI driven cars to say 35mph or under "network" control (in either case Hazard Lights GO ON)

Phase 2: increase speed by 10mph and put laws in place that a car in AI mode is exempted from DWI (as long as the car is driving directly HOME or to the nearest medical facility)

Phase 3: increase speed by another 10 MPH (or to current speed limit)

Phase 4: AI cars allowed to not have Hazard lights ON unless otherwise needed ....

Phase N: AI cars allowed to travel without somebody in the car and to pickup children (note we had better have KITT level AIs in cars at this point)

Re:a way to do this "safely"

[Anonymous Coward]

Phase 2: increase speed by 10mph and put laws in place that a car in AI mode is exempted from DWI (as long as the car is driving directly HOME or to the nearest medical facility)

Why? Is my robot car somehow more dangerous when it's driving my drunk ass to Taco Bell instead of home?

Re:a way to do this "safely"

[tnk1]

Phase 2: increase speed by 10mph and put laws in place that a car in AI mode is exempted from DWI (as long as the car is driving directly HOME or to the nearest medical facility)

I don't think that the restriction is required. The problem with DWI isn't your location, it's your ability to control the vehicle safely. If drunk guy wants his car to drive itself to the middle of nowhere, and it is under the control of the AI the entire way, I don't see why he'd be busted for operating a vehicle under the influence. He might wonder why he was at the county landfill in the morning when he sobered up, but he would have gotten there safely.

Re:a way to do this "safely"

[drinkypoo]

Phase 1: limit AI driven cars to say 35mph or under "network" control (in either case Hazard Lights GO ON)

No. You are going to need a new signal. Hazard lights already have a meaning, and that meaning is "I am stopped or otherwise below the minimum speed for this roadway which I am in, and thus obstructing it — go around me when/if it is safe".

Re:a way to do this "safely"

[laurencetux]

trying not to add "Yet Another Light" for folks to have on the car but your point is valid.

maybe lights on either side of the Third Brake light?? (flash yellow for automode flash red for driver "offline" call EMS)

Re:a way to do this "safely"

[Anonymous Coward]

Phase 1: limit AI driven cars to say 35mph or under "network" control (in either case Hazard Lights GO ON)

Only an idiot drives with hazard lights on. In many states it is illegal to do so. I was recently driving in heavy rain and there were idiot who didn't realize with their hazard lights on, no one can see their turn signals!

Re:a way to do this "safely"

[dave420]

No, the lights simply mean "hazard". Hence them being called "hazard lights". The clue is right there. They can be used for the situations you mention, but also to warn drivers of hazards ahead of the car with hazards.

Re:a way to do this "safely"

[delt0r]

Even a shitty AI car is going to be a better driver than you. Do you go around with your hazard lights on? WTF is wrong with tech geeks not understanding tech?

Re:The Republicans have now killed self-driving ca

[ledow]

In the UK, most tiny karate clubs have a GBP 1m public liability insurance, and it costs a pittance each year.

The fact of the number makes no difference, it's what's covered. I imagine they have to cover a lot more, but even the WORST of these may be better than human drivers on average, so it will quickly re-balance once the risk statistics are apparent, even if companies only pay at first for their testing cars.

Honestly, $100k+ liability insurance is pretty low. Even a school will have GBP 5-10 million and it get claimed on all the time and they handle care of children, including activities, trips, sports, staffing, etc.

Just Wait...

[sycodon]

...until the first time AI kills someone.

1. AI Manufacturer pays millions and millions in damages?
2. AI Manufacturer finds a way to pawn off responsibility on to the owner.
3. AI Manufacturer passes a law capping damages and maybe even some kind of limited indemnity for the AI Manufacturers

Re:Just Wait...

[tnk1]

They'll still find a way to make a human responsible for this. The car makers or Google or whoever, will have more responsibility, but unless it was a completely maliciously covered up bug, then I imagine that AI failure would be treated like a brake failure where the manufacturer will be responsible, but *only* if the AI was being maintained and patched according to the stated guidelines.

Re:Just Wait...

[ShanghaiBill]

...until the first time AI kills someone.

There have already been many deaths caused by software bugs. There is plenty of legal precedent. This is nothing new.

In most past instances, the manufacturer has been held responsible. The owner of the device may be held partially or fully responsible if they were using their device irresponsibly, or had modified it in a way that caused or contributed to the failure.

This will end well.

[Anonymous Coward]

As stupid as humans can be, machines programmed by stupid humans are much worse.

Re:This will end well.

[Anonymous Coward]

It's impossible to predict. The closest analog is asking whether airplanes are safer or less safe with auto-pilots. My guess is that the answer is safer.

Re: The Republicans have now killed self-driving c

[Anonymous Coward]

Wow, so those Republicans made these cars even deader than I first thought.

Re: The Republicans have now killed self-driving

[Anonymous Coward]

This. Same reason they require bicycle helmets. Yes, they do decrease the number of brain injuries, but they more than offset that in profit for orthopedics because of the increase in broken bones.

legal considerations .

[swell]

Do I still pay for insurance or can the cost of incidents be charged to the software maker?

Will the cost of my insurance vary depending upon the safety record of the software provider?

Will I be unable to use my vehicle if flaws are discovered in the software? This assumes that Big Brother can disable any vehicle or class of vehicles from a central control location. Which also assumes that Small Hacker can also disable vehicles. Which also assumes that forced updates will be required and that end user modifications will be illegal and detectable.

We may soon have adequate technology to make self driving cars, but the legislative and legal ramifications will take decades to work out. What software provider is big enough to survive the lawsuits that will grow from a deadly flaw? How will the software flaw be argued and proven/dis-proven before jurors in a courtroom? How many lawyers will get rich?

Re: The Republicans have now killed self-driving

[Anonymous Coward]

This. My wife is an ER nurse, and it's sad hearing about all of the kids with broken bones since they rode less carefully because they thought they were invincible because they were wearing their legally required helmet. The Republicans, as always, are hurting children.

Re: The Republicans have now killed self-driving

[Anonymous Coward]

At this point since the pukianz require us to buy health insurance even if we don't need or want it, it doesn't really matter if they make more profit by outlawing self-driving cars. They still make money!

Re:The Republicans have now killed self-driving ca

[tnk1]

Huh? If you lease any car you have to carry a $100,000 PIP. It's certainly more expensive, but it's hardly out of reach for a normal driver. I was carrying that level of policy when I was 22 years old.

Other liabilities

[Anonymous Coward]

Who has the responsibility/liability of repairing these cars, and who has the liability if something goes wrong as a result?

And for that matter, who has the responsibility if the car's software is hacked (in either sense of the word)?

One step closer to personhood

[grilled-cheese]

If companies can be granted quasi-personhood status, why not a car AI? Are we ready to deal with the implications of car AI rights and car AI voting?

The Great Race !

[swell]

'The Great Race' was a 1965 movie and also an American tradition. Competitors race from one side of the country to the other in various vehicles with various rules.

Self driving cars will surely do the same. They will be judged on safety and speed and technicalities like choosing the best route and handling obstacles. Car buyers will want this information and car makers will struggle to optimize their software to win the next race.

The Matrix Has You...

[evolutionary]

Imagine driving in a vehicle that can literally "take you for a ride". Hackers could have a field day with this. Wonder if law makers/enforcement agencies will allow a manual override. If people can get reveal private/classified information from federal agencies like the CIA and FBI, is a car (or car server network) going to be a tougher challenge? Hmm...

Google is on crack

[kheldan]

From TFA:

Google "expresses concern that providing human occupants of the vehicle with mechanisms to control things like steering, acceleration, braking... could be detrimental to safety because the human occupants could attempt to override the (self-driving system's) decisions," the NHTSA letter stated.

Bullshit. Vehicles must have a full set of manual controls available to the human operator at all times, and furhermore they must be fully educated, trained, licensed, and insured, just like always. To do otherwise is what will put people's lives at risk. Google is smoking crack and needs to be put in their place.

Re:Google is on crack

[Anonymous Coward]

Maybe at first, but in the long term humans will not even know what it means to drive a vehicle. Ultimately it will come down to safety... once they get all this figured out it will just be too risky to have humans manually driving. Software can be buggy but humans are reckless. Software bugs can be fixed but we have not figured out an effective way to keep people from being reckless.

Re:Google is on crack

[Anonymous Coward]

While I have no data to support this, I'm quite sure the average driver in a self-driving car would be unable to keep up the requisite spatial awareness and concentration if he didn't have to do any driving.

Re:Google is on crack

[Anonymous Coward]

No, no, no. That's all bullshit too. That sort of thinking is going to lead to an entire race of people who aren't capable of doing anything for themselves, they'll have to have machines to do everything for them. Where does it end? You'll eventually have machines creating machines and machines programming other machines. Except for a very tiny fraction of a percent nobody will have any motivation to learn how to do anything for themselves. You can laugh all you want at dystopian future cautionary tales as being too fantastic to ever be real, but we're seeing the beginnings of this happening right now before our eyes.

Software can be buggy but humans are reckless.

Speak for yourself, asshole. Just because YOU aren't a competent driver, or just because ONE bad driver out of tens of thousands stick out in your memory, does NOT mean that ALL drivers are bad and that NO ONE should be allowed to operate a motor vehicle! Your entire mode of thinking on the subject is wrong, wrong, wrong. We do not need 'self-driving cars' AT ALL, what we NEED is reforms in driver education, training, and testing. Once highschools stopped offering driver training is when things really started going bad. We need to FIX that, not force people to pay for a bunch of sketchy technology and give them no way to control the vehicle in an emergency.

Doesn't matter what idiots like you have to say on the subject, smarter heads will prevail, and ALL vehicles will continue to have a full set of manual controls for the forseeable future, you're all still going to have to be properly educated, trained, tested, and insured before you'll be allowed to operate a motor vehicle, REGARDLESS of whether it's so-called 'self driving' or not, and whether you like it or not is irrelevant. You're not only living in a fantasy world if you think you're going to sit down in a box on wheels with nothing but a seat and no controls and take a nap while it takes you wherever, and furthermore there is something wrong with your brain if you want such a thing! Seriously, do you people have a death-wish or something?

Hurr I want some faceless soulless machine to have the ultimate power of life and death over me!

That's what you people sound like every time you open your mouths on the subject. IT IS NOT NORMAL. You need psychiatric help.

Now STFU and go take some driver training classes. You're going to need them THE REST OF YOUR LIFE.

Re:Google is on crack

[Garybaldy]

Sure. We can let you have some control. All liability falls on you whenever you take control of part of the driving.

Re:Google is on crack

[Garybaldy]

I assume it will only work if the driver has 100 percent or the AI has 100 percent. Yes you can have your controls. But its either you or the AI not a mix.

Re:Google is on crack

[Gilgaron]

I would expect the biggest and most lucrative application for self driving vehicles would be for freight, not commuters.

Re:Google is on crack

[kheldan]

Sure. We can let you have some control. All liability falls on you whenever you take control of part of the driving.

Not a problem, just like it hasnt' been a problem for the last 35 years I've been driving, during which by the way there has been exactly ONE accident I was even remotely responsible for, and even then there were mitigating factors involved that influenced the outcome.

I assume it will only work if the driver has 100 percent or the AI has 100 percent. Yes you can have your controls. But its either you or the AI not a mix.

I'm guessing you don't design systems of any kind for a living. Yes, of course, it couldn't work any other way -- with the exception, perhaps, of the 'emergency braking systems' I've been hearing about on some new cars. But that's just braking, not navigating. Otherwise to have two different systems (one silicon, the other carbon-based) fighting for control constantly? Would never work. Has to be 'autopilot ON' or 'autopilot OFF', nothign in between.

Re:Google is on crack

[shawn2772]

From TFA:

Google "expresses concern that providing human occupants of the vehicle with mechanisms to control things like steering, acceleration, braking... could be detrimental to safety because the human occupants could attempt to override the (self-driving system's) decisions," the NHTSA letter stated.

Bullshit. Vehicles must have a full set of manual controls available to the human operator at all times, and furhermore they must be fully educated, trained, licensed, and insured, just like always. To do otherwise is what will put people's lives at risk. Google is smoking crack and needs to be put in their place.

Nice assertions. Got any arguments to back them up?

Re:Google is on crack

[Anonymous Coward]

Yeah! just like how passengers are allowed to assume control over taxis, buses, trains, and airplanes. Oh wait... No they're not!

STFU and go sit in the corner! You're fired.

Robocop gets a "killswitch", not an xbox controller. Fucking dumbass.

Re:Google is on crack

[Toshito]

Maybe at first, but in the long term humans will not even know what it means to drive a vehicle. Ultimately it will come down to safety... once they get all this figured out it will just be too risky to have humans manually driving.

You're right, but with a lot of emphasis on "long term". Very long term.

We're talking 30 to 40 years for having no human driver on the road.

Replacing all the cars currently on the road, considering the fact that a lot of them are used for 15 to 20 years before being scraped, and considering that for at least the next 15 years only some cars for sale will be autonomous, and you get at least 30 years for a complete turnover.

Would you force someone who took great care of their cars, which is still perfercly fonctionnal, to shell out tens of thousands of dollars to buy a new autonomous car? And what about motorcycles? Antique and collector's cars?

What about winter, bad weather, trails, unpaved roads, camping trailers, boats trailers... It will take years before the AI can handle those border cases. You know how it goes, 80% of the cases are probably currently covered by the Google car, but the last 20% will be a bitch, just like most projects.

I think the right thing to do is to have self driving cars for those who can't, shouldn't or don't want to drive, and let those who can drive for themselves.

But I would make it a lot tougher to get a driver's license, and to keep it. Better education, tougher tests, regular retesting, medical examinations, more like a pilot's license. It would be a privilege, and those human drivers would be much safer.

Re:Google is on crack

[Anonymous Coward]

Vehicles must have a full set of manual controls available to the human operator at all times, and furhermore they must be fully educated, trained, licensed, and insured, just like always.

Many of the human operators around here seem pretty well trained to focus on their cell phones.

I think I'd trust Google's AI more.

Re:Google is on crack

[Anonymous Coward]

No, it doesn't. 90% plus of accidents come down to driver fuck-up.

Putting the most unreliable device back into the system is a horrible idea. While I think having some sort of manual control is probably important, I think it should look more like the "I Robot" movie cars. cars that are basically fully automated unless you take a number of actions to get manual control. And that mostly for really unusual situations, NOT emergencies.

Re:Google is on crack

[delt0r]

You put peoples lives at risk everytime you drive. A computer will be a better and safer driver than you will ever be. Humans are not fucking magic.

Tickets / civil liability / criminal liability?

[Joe_Dragon]

Tickets / civil liability / criminal liability?

And with tickets you have

parking ticket is issued by a private, non-governmental parking authority patrolling an office parking lot or shopping area

private security guards issuing live speeding / moving tickets (non state tickets) (some HOA's and some parking lots or shopping area)

moving tickets from a real cop

parking tickets

red light tickets (parking like)

red light tickets (moving like)

speed camera tickets (moving like)

speed camera tickets (parking like)

toll violations (I can see an auto driver car with bad DB info getting one in some settings)

Re:Tickets / civil liability / criminal liability?

[SecurityGuy]

I'm curious where you live that an office parking lot or shopping area can give you a ticket. Private property (here at least) doesn't grant you the authority to fine people. It just grants you the right to ask them to leave, have their car towed, etc.

Re:Tickets / civil liability / criminal liability?

[Anonymous Coward]

The entire UK operates like this now. All shop carparks are private land, and they have cameras on the entrance and exits that read your licence plates. If you stay over X, they issue you with a parking notice. This isn't the same as parking on a double yellows, it's from a private company. It used to be crap, but the companies have since bought enough politicians and judges to make it law. By entering their carparks, you are agreeing to any terms and conditions. Overstay by 3 minutes, £80 fine. Don't pay? You'll be taken to court.

It's not the shops doing it, it's those that own the land sharing the revenue with a small number of quangos set up to exploit the situation. Look up parking eye and tossers like that if you want more info.

Re:Tickets / civil liability / criminal liability?

[Joe_Dragon]

not near me but there are places where they try bs like that. But lets say goolge owns the car there may be a big up tick in $10-$40 bs tickets that Google will just pay vs fighting them.

Re:Tickets / civil liability / criminal liability?

[fizzup]

First, imagine this: the price of a self-driving car will be too high to allow it to sit idle in your garage. You will never, ever own a self-driving car. You can choose to buy a car that you drive yourself, or you can choose to consume Transportation-as-a-Service (TaaS). Uber will provide it. Google will provide it. Maybe car manufacturers will provide it. You can answer all your questions yourself if you replace "self-driving car" with "taxi."

TaaS will be in competition with car ownership, so providers will have to comfort users and provide a better experience in order to compete. For what it's worth, it is not a high standard to provide a better experience than insuring, maintaining, fuelling, and driving a car. One specific way that they can provide a better experience is by dealing with all the annoyances that you mention, and more.

Re:Tickets / civil liability / criminal liability?

[Garybaldy]

Lots of places that give you the authority to park on the property with some type of pass issue tickets. Everything from movie studios to college campuses.

Re:Tickets / civil liability / criminal liability?

[Joe_Dragon]

college campuses have real cops.

Re:Tickets / civil liability / criminal liability?

[Joe_Dragon]

Then who is willing to own a auto taxi without an IC driver to pass the blame to?

Re:Tickets / civil liability / criminal liability?

[SecurityGuy]

Anybody can hand you a piece of paper. Having any authority at all to make you pay is the question.

Re:Tickets / civil liability / criminal liability?

[SecurityGuy]

Some have real cops, some don't. I got a ticket on a big college campus once. Never paid it. Nothing happened. Got one on another college campus where I happened to work. I absolutely had to pay that one, as I had a contractual relationship with them where I'd agreed to do so in advance in exchange for the right to park there (in writing). At least here, random companies can't just assess you fines absent some agreement where you allow them to.

Re:Tickets / civil liability / criminal liability?

[Garybaldy]

Everywhere i have seen this. Yes ZERO authority exist to force you to pay it. Unless you don't like HOA fines. or ever parking in school parking again or anywhere on the movie studio again. Odd thing about the movie studio one. You just get your pay docked for not paying the parking or moving violation ticket.

Re:Tickets / civil liability / criminal liability?

[SecurityGuy]

I get HOAs (if you're a member), as well as schools you attend (want your transcript? pay your fine...). The ones I questioned were office parking lot/shopping area, etc. I guess I understand office parking lot if it's the office you work in, rather than one you're doing business with as a customer. There's zero chance I'd ever pay a parking fine at a shopping area. Then again, I do boring things like park in parking spaces at shopping areas, so it's never been an issue.

Anyway, AC answered my question. Somebody actually got it passed as law in the UK. If so, it's legit (if terrible).

Re:Tickets / civil liability / criminal liability?

[Anonymous Coward]

Technology is cheap. No reason for self driving cars to be unusually expensive. That said, "TaaS" is something with a huge potential market.

jonny cab

[Joe_Dragon]

jonny cab is not responsible for injury or death! We hope you enjoy the ride and ride we are not a amusement park so there is no operator on board.

Re: The Republicans have now killed self-driving

[sims 2]

Thats too bad I'm sure they would have rather been killed than had a broken arm.

Also If you're worrying about the cost of care you have much bigger problems.

Maybe we should teach kids to bike carefully because people in cars aren't looking for them. Helmet or no helmet that part doesn't change.

Who gets the ticket?

[Anonymous Coward]

So if a self driving does something wrong, like goes through a red light, or is speeding.
Who gets the ticker? The passenger, the car, Google, the engineer, or the programmer?

So how much were they bribed?

[Etcetera]

Seriously... I can't think of any way shape or form that the "AI" behind a "self-driving car" is anywhere near ready for full legal responsibility for this.

Google (and/or other tech companies trying to get this to happen) must have placed tremendous pressure on them to make this happen.

Re:So how much were they bribed?

[koan]

"Google (and/or other tech companies trying to get this to happen) must have placed tremendous pressure on them to make this happen."

They absolutely did, in the same way they used lawyers in certain cases to defend Google Glass and driving.

My thoughts on this is that you lose your "independence" to a degree, no more travelling via car without a record of it.
No more going where you want, as the car may be GPS fenced and simple refuse to take you.

And what would the roads be like with 50% human drivers and 50% autonomous machine drivers?
It's been stated that the accidents occurring are because robot cars drive substantially different from a human, making them difficult to predict (for humans) and drive near.

Re:So how much were they bribed?

[kqs]

Seriously... I can't think of any way shape or form that the "AI" behind a "self-driving car" is anywhere near ready for full legal responsibility for this.

An AI cannot have legal liability; it is a machine. Depending on how this shakes out, either the auto seller (Google, Toyota) or the auto owner will provide insurance and have legal liability.

And since human drivers are almost universally incompetent, as long as the AI driver is more competent than the average human (a low bar), the insurance will be cheaper than insurance for human drivers.

Re:So how much were they bribed?

[Anonymous Coward]

A plebe who isn't informed on the state of the art in autonomous vehicles. There's a shock!

Were you even invited to CES?

I'll answer my own question:
"Of course you weren't! If you WERE one of the cool kids who saw the NVIDIA demos at CES you wouldn't be flapping your ignorant gums right now."

And So It Begins...

[kackle]

And so it begins... I predict that congestion will be significantly increased on average as these car struggle with the frequent, "non-standard", driving conditions that we all experience but don't think twice about (neither did the engineers, I presume). Just yesterday I saw there was a police chase that ended in a highway crash, turning 4 lanes into 1, forcing everyone onto the shoulder for obvious reasons. Would this also be obvious to such automated cars, especially after it snowed that morning?

Progress will again be lubricated by the blood of a few innocents. (Or at least the time & patience of many...)

Re:And So It Begins...

[Whorhay]

I can see this being an issue a bit at first. But as SDC's hit a critical mass it'll stop being an issue because SDC's and their correct driving behavior will be the standard. Hell, I can easily foresee a day when human drivers will only be allowed on closed tracks. Auto accidents like the one you mention will likely become a thing of the past and be a real rarity. Rush hours will likely be much less of a hassle because the SDC's will be able to handle the congestion much more efficiently and intelligently.

Re:And So It Begins...

[Terwin]

Rush hours will likely be much less of a hassle because the SDC's will be able to handle the congestion much more efficiently and intelligently.

Only up to a certain point, and as politicians would rather spend money on themselves, this will just mean less road construction until things are at least as bad as they were under human control (probably worse, as the time-cost is not as severe when you can do other things during the trip, so less pressure will be brought to bear on fixing things).

Re:And So It Begins...

[kqs]

Sadly, you're probably right.

But you put the blame in the wrong place. Don't blame politicians for not spending on infrastructure. Blame you and me, the voters. We seem to only vote for people who claim that they will lower taxes, and are then amazed when there is no money for infrastructure maintenance much less improvement.

Laugh

[koan]

Essentially a ground based drone, notice there's very little hysteria over automated cars, even though they present a far greater privacy threat than airborne cameras.

Oblig.

[Sir_Eptishous]

I strip away the old debris that hides a shining car
A brilliant red Barchetta, from a better vanished time
I fire up the willing engine, responding with a roar
Tires spitting gravel I commit my weekly crime

A stress free commute?

[CMOS4081]

I'll be bale to go back to reading, sleeping or listening to music? Sweet

Stupid and lazy people...

[Anonymous Coward]

While I see self-driving car technology as an amazing enabling technology for the elderly and differently-abeled, I find it pathetic that anyone would actively encourage the use of this technology to further coddle and appease lazy stupid people that can't be bothered to rise their heads from their cell phones to fucking drive.

If they're too lazy and stupid to drive safely, make them walk, take a bus, cab, or train and fuck off.

I'm ashamed to be the same species as some of the stupid lazy fucks that think this is such a great idea for anything besides enabling better mobility for those who can't drive!

Cure cancer, naw, we're too busy coming up with more ways to enable stupid lazy fucks to eat McDonalds in their car!

Pathetic!

No one mentions...

[NetNed]

The red herring in all this is that these cars can't handle weather or other road conditions very well. Potholes would be a major concern because avoiding them with oncoming traffic and other factors, with maybe a dash of weather, say ice on the road, etc. makes them not so much the great white hope that all to many seem to claim that they will be. Speaking in general terms, humans actually do a pretty good job of driving a car. Unfortunately, instead of using rational thinking in curbing distracted driving we see a push for something that is ripe for abuse and intrusion.


Throw in things like this and I find it odd that anyone would be "all in" on self driving cars.

Re:No one mentions...

[Anonymous Coward]

Talk is cheap. Lots of people like to claim to know what is challenging to these vehicles.

I have a pretty good understanding of what is under the hood and none of what you say is particularly difficult to solve is particularly difficult to solve. Google is willing to put their money where their mouth is.

Anyone willing to bet cold hard cash that Google Autonomous cars will be the at-fault driver at a statistically significant higher-rate? If not: it sounds like a nice cup of STFU is in order! I want my autonomous car yesterday and I'm tired of the sewing circle from the naysayers.

"Never interrupt someone doing what you said couldn’t be done. – Amelia Earhart"

Re:No one mentions...

[NetNed]

Talk is cheap and you posted as an anonymous coward? Ahhh yeahhh, a big cup of you're a fucking idiot is in order. It's been 2 years and they still can't overcome the weather factor. Money isn't going to fix it. But yes, because you "want it yesterday" means we should all goose step to Lord Google's drum beat. Guaranteed you are under 30 and think you are 10x smarter then you actually are.

3-star General Harry W.O. Kinnard is the one

[Anonymous Coward]

who said Nuts!

http://media2.s-nbcnews.com/j/...

George S. Patton said something to the effect, "their taking our bodily fluids".

I found many of the comments on this issue

[NEDHead]

to be perceptive and important.

But what I really want to know is how this will affect watching porn in the car

75% of drivers are better than average

[drew_kime]

I'd rather let the car handle the freeway time. That's the most boring kind of driving there is. And since the most reasonable objection to self-driving cars is they don't handle difficult conditions as well - construction zones, city traffic, poorly-marked secondary roads - this mode lets people handle the "hard" parts and AI handle the boring, routine parts.

Re:75% of drivers are better than average

[fluffernutter]

Yet the freeway is where the worst injuries and deaths will happen if something goes wrong. So especially there, things had better be 100% right for people to use self driving cars in that scenario.

AI Needed

[TM22721]

Thanks to clueless media the public thinks software can do anything.

Well software sux at most things. Especially the myriad of chaotic situations a driver encounters.

Excuse me while I go to the store to get more popcorn.