Home » Fantasy/SciFi » Computers can qualify as drivers

Computers can qualify as drivers

10 February 2016

Not exactly related to the book biz (other than SciFi), but interesting.

From Reuters:

U.S. vehicle safety regulators have said the artificial intelligence system piloting a self-driving Google car could be considered the driver under federal law, a major step toward ultimately winning approval for autonomous vehicles on the roads.

. . . .

Google’s self-driving car unit on Nov. 12 submitted a proposed design for a self-driving car that has “no need for a human driver,” the letter to Google from National Highway Traffic Safety Administration Chief Counsel Paul Hemmersbaugh said.

“NHTSA will interpret ‘driver’ in the context of Google’s described motor vehicle design as referring to the (self-driving system), and not to any of the vehicle occupants,” NHTSA’s letter said.

“We agree with Google its (self-driving car) will not have a ‘driver’ in the traditional sense that vehicles have had drivers during the last more than one hundred years.”

Link to the rest at Reuters


27 Comments to “Computers can qualify as drivers”

  1. So what are their insurance rates and can they be ticketed? Or maybe more important, do they have to pass an eye test?

    • Like you, I wonder if we won’t be worse off as litigants if we have to sue Google instead of the owner/driver and his/her insurance company?—or all the folks who we used to have to sue, plus the provider of the “artificial intelligence system.”

      And if Google (or worse yet a surrogate) is the “driver,” do they have the same duty of provide insurance, etc., and will they behave in the same pattern as human owners/drivers?

      • One thing Google will have on its side is a complete recording of how their car ‘saw’ things before and during any accident …

        • Except I believe most of these systems–maybe not Google’s, but I would imagine so?–use neural networks, so they’ve really got no idea what the system thinks it’s doing when it runs over a bunch of nuns leading disabled kids over a pedestrian crossing.

          With a neural network, you feed stuff in, you get stuff out, and the relationship between the two is a complex function that’s likely impossible to calculate without knowing the exact state of the system when you fed stuff in. There’s no line of code saying ‘if you see nuns leading disabled kids over a crossing, run them over’.

          • I think it is all matrices and linear algebra, not neural nets. The rules are in the data in the matrices. The problem with tracing the reasoning is a decision can be based on a large number of variables, making the rule difficult to understand. I share your distrust of neural nets.

    • Econ blogger Megan McArdle has had a couple of posts about this issue. She thinks we’d have to change liability rules in order to make driverless cars a real possibility:

      On the user side, liability is limited by the fact that most people are judgment proof, or close to it. Plaintiffs will usually sue up to the limits of your liability policy, but not beyond, because it’s not worth the hassle of trying to collect.

      -Snip –

      On the other hand, the expected value of a $10 million judgment against Ford is … $10 million. That gives lawyers more incentive to file claims, and increases the expense to the company of the claims that are filed.

      And even if the overall number of accidents drops, the number of accidents where the automaker is perceived to be at fault will approach 100 percent. After all, they’re the ones who designed or installed the software that made the decision. And while in theory, a jury should be able to say, “Well, this was a hard design problem, you can’t make everyone happy, and this is an unfortunate tragedy,” in practice, this is unlikely.

      Source here, http://www.bloombergview.com/articles/2013-09-11/to-enjoy-driverless-cars-first-kill-all-the-lawyers

      If you’re curious, she’s got several more posts on the subject. Bottom line, these cars won’t be ubiquitous until Google — and Ford, and Toyota etc. can be sure they won’t be bankrupted in the event of an accident. She suggests you’d have to adapt New Zealand-style “no fault” laws before we’d get any serious traction on driverless cars.

      Personally, I am irritated that we don’t have the flying cars we were promised. That’s what Google should really spend it’s time trying to give us 🙂

  2. Can’t be any worse than some of the carbon based driving units on the roads these days …

  3. It’s an interesting science fiction world we live in, isn’t it?

    But I don’t think the self-driving cars make a good SF story because there’s no conflict. I think the real story is what happens when the cars get hacked.

    I…um…wrote that novel back in 2012.

  4. While I am sort of looking forward to having a self-driving car in my golden years, I see real marketing issues for them. Face it, if Google – or any other corporation or organization – is driving, you are going to do the speed limit or less. Always. It will not be in your control.

    Many will not stand for that.

  5. “Why Driverless Cars Will Screech to a Halt”


  6. P.G.

    I find it interesting that ALL the road incidents/crashes Google cars have been involved in are those where human driven cars have rear-ended the Google car.

    This is a nation of lead-booted tailgaters. I suspect Google will win a whole lot of those legal cases. Being robots, they automatically keep to the speed limit while 90% of the rest of you don’t.

    The fatal accident rate in the UK, for instance with almost 25 million cars on the road is almost exactly the same as that of GA, which has about 3 million cars on the road.

    Ya’ll drive too fast, folks.


    • Most UK drivers drive fewer miles in a month than US drivers drive in a day.

      Apples. Oranges.

      • Source?

        I’ve worked and driven in both countries over periods of 5 – 10 years. I drove about the same time/distance ratios for work.

        I could drive faster in the UK – motorway speeds were 70 mph.

    • I find it interesting that ALL the road incidents/crashes Google cars have been involved in are those where human driven cars have rear-ended the Google car.

      Only because you haven’t read the recent report on the real capabilities of Google’s cars.

      If I remember correctly, they have to hand control back to the human driver about once every thousand miles, and would have crashed at least once every 40,000 miles if they couldn’t do so. This is the equivalent of an aircraft autopilot handing control back to the pilots because it doesn’t know what to do, the pilots crashing, and the crash being blamed on ‘pilot error’. Which it may well be, but the pilots only errored because the autopilot broke.

      But, yeah, most accidents I see on the roads around here are caused by tailgaters when the car in front stops and they don’t. I remember watching one time as a car stopped for a pedestrian, the car behind slid to a halt on the ice with the tires scrabbling for grip, and the car behind that slid onto the far side of the road with the ABS chattering away, because they couldn’t stop on the ice before they’d run into the second car. All because they were driving far too close for the conditions.

      At least no-one actually hit anything that time.

      The auto-braking systems on a lot of new cars that prevent you from getting too close will probably cut accidents significantly as they become the norm.

      • most accidents I see on the roads around here are caused by tailgaters when the car in front stops and they don’t.

        This is exactly the cause of virtually every accident I’ve been in. I’m at a red light, the person behind me is either distracted by her children or mistakenly doesn’t realize the light is still red.

        The auto-braking systems on a lot of new cars that prevent you from getting too close will probably cut accidents significantly as they become the norm.

        I did not know this existed (my car is old. Bought it after one smashing tailgater smashed my previous car). I hope it becomes more ubiquitous soon so my insurance can go down.

  7. I think the best thing about driverless cars is they stay calm. In London, drivers have to contend with the Congestion Charge (£11.50 a day in central London), frequent traffic jams, speed bumps, speed cameras, traffic wardens, road works, bus lanes, and difficult-to-find and expensive parking. Many drivers are cross and impatient, and hence dangerous.

    Driverless cars working 24/7 that you could summon as needed, replacing the private car, would transform London. That’s what’s happened by 2045 in my latest novel :o)

    • I think the best thing about driverless cars is they stay calm.

      [Citation needed].

      How calm will your car be when you tell it to pick you up at work at 5:00pm, and there are major construction works and a few crashes to get past?

      Someone’s going to have to program it to prioritize, and I suspect many people would rather buy a crazy car than a ‘calm’ one that will leave them waiting for half an hour when they need it. Even if they might press the ‘calm’ button once they get inside.

      • My idea is that you no longer need a car of your own, when you can summon a pod driving in a street near you from your phone (like Uber). There will be less congestion, because cars will be driving 24/7 and serving many people, so there will be fewer of them. Roads will not be narrowed by cars parked along the kerbs.

        • So, where are all those cars going to come from to carry people around during the rush hours? And what will they do in between?

          You could probably reduce the number of cars quite substantially, but there’ll still have to be enough to handle peak times, and then there’ll be far too many for them all to be driving people around during work hours.

          • When not needed, they would dock at docking stations, or visit service stations for upkeep. How many hours per week are most cars used? Not many. In London, most cars are kept parked on the roads, not in garages. Imagine the space if they weren’t…

            I’m not saying this system would be perfect. I am quite certain it would be better than the current lethal mess.

            • That’s the vision I have, as well. Drone cars, summoned by cell phone or appointment. It would make owning your own car completely redundant, at least in cities. (Still considering walking, biking, busses and rail.)

              Of course, managing rush hour would be a challenge. But maybe rush hour is obsolete when many could work in a home office and it no longer matters when work hours happen?

              Where I work (German civil service, basically), we have “core hours” when people should be there, but it doesn’t matter when we start or leave. I aim for a similar time frame every day, but it doesn’t matter if I arrive five minutes earlier or later. Not even 15 minutes. Anything longer, and my boss would like to know so she doesn’t have to worry about me, not because it creates any problems. And if I need to change work hours completely on a given day, I can just tell her and it’s fine.

              So, maybe adapting work and work hours could erase rush hour. And honestly, who doesn’t want to stop paying lease rates or debt for a car? I’d go for a drone car the moment they are available.

  8. Let me know when they have a computer that’s a decent putter.

Sorry, the comment form is closed at this time.