Should a robot decide who lives or dies?
Fri, June 24, 2016 A recent survey asked the following hypothetical question. If a driverless car is headed on a collision course with a group of pedestrians, and it is impossible for the car to stop. Should the car be programed to plough into the pedestrians, killing many of them, or swerve off the road to avoid collision and possible kill the passengers in the car?
People answering the survey no doubt thought this way: “I am not a sociopath therefore I don’t want my car to kill people, but on the other hand I don’t want it to kill me and my family, so I won’t buy one.”
But shouldn’t this question have been asked long before now when buying an SUV that is as big as a small house, and built like an armored truck. People buy these vehicles to protect themselves and their families. Protect themselves from the other driver, that is.
The fact that these larger, heavier vehicles are then a greater hazard to every other road user, pedestrians and cyclist especially, and even smaller compact cars. The question doesn’t even arise because each individual SUV buyer sees himself as a good and safe driver, it is always the other driver that is the problem.
Isn’t the whole purpose behind the driverless car to eliminate driver error? The cause of the majority of collisions ever since the automobile was invented. You notice I said “Collision” and not “Accident.” This is deliberate.
When human error is a factor it is easy to say “Whoops-a-Daisy” it was just an accident. As I previously pointed out most people are not sociopaths they don’t intend to kill people. But drive in a reckless and dangerous fashion and someone’s death is a likely outcome.
But what will happen when the robots take over and all cars are driverless? Without the human error factor you can no longer call it an accident when someone dies, either inside or outside the car. Who gets sued? Not the driver, because there isn’t one. The robotic system will have failed, so the car manufacturer will be responsible.
And can a corporation even program a computer controlled car to decide who lives or dies? I can see that one going all the way to the Supreme Court.
If cars become driverless, speeds will have to come down dramatically. A pedestrian hit at 30 mph. or less has a good chance of survival. Above that speed the odds become less, and above 50 mph. death is almost certain.
Robotics does not overcome physics. A vehicle traveling at 50 mph. still needs 125 feet to stop. That doesn’t include human driver reaction time, I am assuming a computer will react faster.
Another factor to consider. Cars may be robotic, but not pedestrians and cyclists. Will pedestrians learn that if you step out in front of an approaching driverless car, it will stop? Will cyclists realize that by riding in the middle of the lane, a car on auto pilot will not pass unless safe to do so? How will that go over on a morning commute? Following a cyclist at 10 or 15 mph.
I think fully driverless cars are a long way off. The delay will not be a technical issue, it will be one of, “Will it be accepted by the general populous, is this what people want, and will they buy it?”
A more sensible approach would be to concentrate on reliable and low cost public transport. (Possibly driverless to cut costs.) Another way robotics could come into play is to restrict speeds in heavily congested areas. The current system of everyone driving at least 5 mph. over the limit is ludicrous.
Maybe put a few slow moving driverless cars into the traffic system just to slow everyone down. You can honk and cuss at a robot as much as you like, it won’t do you any good. Speed is the culprit, if everyone was forced to drive at 20 or 25 mph. there is time to react and avoid collisions even when the other driver makes a mistake or does something stupid.
To Share click "Share Article" below
Car Culture, Comment, Opinion
Reader Comments (23)
20-25 MPH is exactly the sorts of speed that these cars are targeting, and they "drive like your grandma," as it's put.
I presume they don't mean mine, who drove like a bat out of hell with her Dodge Omni, a tin can on wheels (fortunately to nobody's harm).
I read somewhere recently that a driverless car-caused death is statistically likely sometime soon.
I believe the Google cars currently on test only go at about 25mph. One was stopped by the police for going too slowly recently.
This question is a remake of that old thought experiment: does one divert a passenger train to avoid killing all the passengers, but kill a sole boy on the tracks, or let the train go off the rails and kill everyone but save the boy? There is no right, or wrong answer…Thus the dilemma.
Why? Because humans recognize endless possibilities, because we think, and become emotionally involved in our decisions. And some are made quite quickly in extreme situations. Computers can’t even come close to that.
Hell, computers can’t even understand poetry, or irony, or allegory, or humor! Think of the Turing Project from the 90’s.
So, the driverless car will kill humans in the car or sidewalk in a most inhumane way. Since our brains work nothing like a computer, a robot cannot make humane decisions.
But we can. And I’d rather make that decision than robots. Geez, it sounds like someone doesn’t trust human ability! Or just wants to take it away.
Very thoughtful post, Dave.
For me, the overall less human death from safer robotic driving far overrides any questions about the nature/cause of those fewer human deaths (although of course, we should seek to understand them so we can minimize them even further). I think that very nearly all those deaths will be the pedestrian's fault (walking into traffic without looking)
However, the equation changes as human behavior shifts. I have never before seen your questions about behavior of pedestrians/cyclists when they learn they can count on driverless cars stopping for them, that's fascinating. Initially I think driverless cars will be given the same amount of caution as manned cars, but if everybody starts learning where the 'edge' is (how close can I step in out in front of a driverless car and it will stop for me?), will human behavior naturally push close enough to that edge (statistically speaking, according to some distribution among the population according to individual risk tolerance) to get overall human deaths at about the same levels?
I know of not one person that wants to give up driving. Driving is a human activity.
I also know of no one that wants to give up riding a bike, or surfing, or climbing, all that can be seen as dangerous. It’s what humans do.
So the message to those fawning over driver-less cars: ‘As we used to say to people in the gym that worried about becoming muscle-bound by lifting weights: “Don’t worry, it won’t happen.”’
Same with robotic cars, they aren’t going to take away our vehicles. Not in any of our lifetimes. Nor should they ever.
I live in Mountain View, or Google Land as we sometimes call it. The place is infested with self-driving Google vehicles, both the larger version and the small, 25 mph battery powered ones (which I refer to as "Google Beans").
Google is spending a lot of time (and money) trying to make a computer that thinks in zeros and ones think like a human brain (which sometimes doesn't think at all). It's been a time consuming process.
But the cars are getting better and better. Ride your bicycle up next to one of these vehicles and they'll quietly (and instantly) move out of the way. Three feet passing? More like five or six feet.
Although I doubt that today's computers will be capable of eventually being able to drive a car it's definitely in the future. Although I wish Google the best of luck I'm still sticking to my bicycle transportation model.
I used to enjoy driving when I lived in Canada. In rural New England it's still IK and parts of Europe. But in the UK it is almost unbearable. I would rather drive to Madrid from London than to Birmingham.
But driverless vehicles will be a soft change. First, large trucks - like the UPS double trailer trucks you see on interstates today - will run safely and efficiently between suburban terminals overnight.
Then, some cars will come with a "city safe" option that will operate at sub-20mph speeds (the average speed of traffic in London is currently 11mph). I can only hope it happens soon.
As far as killing pedestrians. Have you researched how many people jump in front of trains or subways? All the Google car needs to do is brake as hard as possible. That's probably 100 times better driving than the average driver right away. Get over yourselves
A simple reason driving will not be done by robots any time soon:
• Passengers will not be ready for sudden braking; you can imagine the result to those passengers.
• When the robot cannot handle the situation, it hands over control to a human, supposedly who then takes over and saves the day.
• Google said their cars will drive wherever cars go now.
And Google explaining how this takes place?
They sound like children playing with toys when doing so…I laugh every time.
The whole scam, sham and shame on Google is it promised robotic cars in five years, and people threw millions of dollars at Google to do so, and now, a few years later, Google changes its tune and says we are decades away from robotic vehicles, and maybe never on roads with humans. The most those investors will get is golf carts robotically driven on closed courses, like the city of Carlsbad, CA is promoting for its downtown, and only to certain destinations.
The irony is the ones who could use robotic vehicles, the old and infirmed, who cannot get a license anymore, are going to be driven by robotic cars going not much faster than they go in their mobility carts, and then not where they will want to go, from where they need to leave!
You can’t just Wish for something (based on Good Intentions) and expect it to happen, science (Reality) has a nasty habit of calling you out!
Hi Steve, I'm Reuben. I want to give up driving.
Two biggest reasons a "driverless"car won't become reality (at least in the US):
It will always drive at the posted speed limit.
It will always come to a complete stop at signs and red lights.
Actually, many of the cars on the road are already essentially "driverless" with the person behind the wheel doing everything but paying attention to driving.
Progress?? The best car that I ever owned and I still own is a 1970 Porsche 911T NO POWER op windows doors air con. just basic 5 speed stick shift, roll up windows remember them? NO power steering, brakes, A drivers car. YOU HAVE CONTROL NOT the car. Lots of FUN!! I also have a Porsche Boxster that has POWER everything BLOODY BOORING!! to drive. Guess which one I drive the most? The 911T
Five years ago, Google gave the impression they had secretly developed the technology for driverless cars. Local governments put on hold, and diverted many mass transportation projects, interpreting Google’s statements as true, the result of which those cars would now be on all roads.
They are not.
Now Google claims they are “expanding” on those earlier statements, not changing those promises. In political parlance, it’s called “clarification”, “misunderstanding”, “not what they meant” or “restating”. Oh, and five years is the catchphrase for investors. If they had said decades, no one would have paid attention, or money.
Let’s get real and bring back those other plans.
Yes, driving fun is good for when you want to have that kind of fun, but what's also fun is being able to just ride in a car, get where you need to go, and read a book, do a crossword, surf the web, take a nap, ...
Good day! I'd like to share our new app for bicyclist like you all. It's called Yazda App it's newly released so we would really appreciate your feedback about it if you can. Thank you and have a good one!
I agree with Mike W. People do everything except drive while driving. They also go way over the speed limit, cut you off, don't use turn signals etc.... I could write a book about the shitty drivers in California. I think they should make getting a license to drive much like getting a pilots license. Many training hours, defensive driving etc.... I hear in Germany it's much more difficult to get a license. Here in California the written test is given in many languages. Can you imagine a police officer stopping someone say from Viet Nam for a traffic violation and getting that "blank stare" as the officer tried to explain why he's stopped the person. I'm not being racist about this it just happens to be the truth. They might as well let a gorilla or other primate drive as many here can do no better. I think if you make it more difficult to get a license ( extensive training) will make our roads safer. Not driverless cars. If you need to take a nap or stare at your phone then take public transportation. This way I will know you are not on the road not paying attention to your driving. Also, make it mandatory to get a license you must have to take the test in a manual transmission car. You have use the clutch and shift ( no paddles allowed).This way you must pay attention and you can drive any motor vehicle.
There's a lot of food for thought in this topic. I don't know the answers, but when I see probably ⅓ of drivers talking on the phone, or very obviously texting while driving (just based on my observations while commuting by bike - not any official study) I do wonder if these driverless cars might not be an improvement. People know texting while driving is dangerous -- how can they NOT know -- yet they do it anyhow. Giving control over to the car might be the best thing for them (and the rest of us).
By the way, having just returned home from two weeks in London and Paris (1 week each) and having no car the whole time, I have to say that easy access to expansive public transit was a very nice thing. It left me wishing we had such a system where I live, as it would be easy to cut our car use in half.
Increased safety is not the only reason we want driverless cars. A related reason is it frees up humans to work or recreate otherwise, rather than driving poorly in a foolish attempt to multitask.
Another important reason is to reduce the number of required vehicles, parking infrastructure, and fuel consumption. Vans picking up multiple people that a global routing algorithm has determined are making coincident trips, will reduce road congestion and fuel consumption. People using driverless cars only on demand will reduce the number of overall cars that sit idle most of the time. A well-run network of driverless cars will be a more flexible system of 'public transportation'. Most families should find it cost-effective to reduce to at most one personally owned vehicle.
Tesla: news item today - driverless car didn't see a white tractor/trailer HGV against a bright sky, so carried on driving at speed (even after it took its roof off and killed the driver). So, what will cyclists be made to wear to make sure that driverless cars can 'see' them?
And Tesla is now being investigated by the SEC for covering up that accident!
Reality punches Elon in the face...
Tesla car is a good idea. Because they do not use gas, they use electric. By the way, self-driving car, sometimes useful when you drive long way.
I'm not sure driverless car can help reducing accidents but i think that all problems are human. We need to raise people's awareness and simultaneously tighten the traffic law, maybe it's able to raise the fine or have punishment when one breaks the law. So people have to pay more attention when driving.
So, the driverless car will kill humans in the car or sidewalk in a most inhumane way. Since our brains work nothing like a computer, a robot cannot make humane decisions. I trust it
I agree with Fred E. Graw that self-driving car sometimes useful, but I support the idea that human control masterly, not the robot.l