This week was the very first recorded incident of a fully-autonomous “self-driving” car driving into a human being and killing them as a result. Of course this is a horrible thing especially for the victim, Elaine Herzberg, her children, other members of her family, and her friends. In writing this article I don’t mean to diminish the repugnance of this event or the sadness that it has caused to many. I do think it was inevitable that someone would get killed by a self-driving car at some point. It was just a question of when this would occur. Logically we all know this must be true. Autonomous vehicles are not going to be perfect. Errors will occur — errors made by humans, and errors made by autonomous computers — and sometimes death will result.
There was a human driver (alone, no other passengers) in the autonomous vehicle as well, but the car was in fully autonomous mode when the accident occurred. The car had a plethora of cameras and one was pointing at the human occupant. The look of shock on that person’s face at the moment just before impact was horrible. Although I believe the driver was not physically injured by the accident, I expect it will have left permanent emotional scars.
The optimistic side of this is that autonomous vehicles will likely kill fewer people on average than human drivers will kill with their cars over a given number of miles or by whatever measure you use. Accepting this will be no comfort to those who die at the hands of these autonomous vehicles. I was having a conversation with Dr. Joel Zivot about this and he rightly pointed out that autonomous vehicles may well kill different people than human drivers would kill. That is, even if they kill fewer people overall, they may kill people that human drivers might not kill in the same circumstances. That conversation naturally led us to the Trolley Problem. How would an autonomous car handle such a dilemma?
Joel is much more familiar with death than I am. As a critical care physician he experiences the deaths of hospital patients several times each month. Furthermore, Joel is a bioethicist who has brought a new challenge to the “medical model” of prisoner executions in the United States based on drug shortages negatively impacting medical care. Joel went to Missouri this week. Russell Bucklew, an inmate who is sentenced to death, invited Joel to attend his execution. However, just hours before the execution was to occur, the U.S. Supreme Court stayed that execution.
It is difficult for me to even begin to imagine what it must be like to live as a condemned inmate, or to work as a medical professional in critical care, or even to be someone looking through a car window and realizing the vehicle is going to kill someone in the next second or two and there is nothing you can do to prevent that. Just thinking about each of these people and imagining how they feel is intensely emotional for me. How do these stories make you feel?
Maybe this is what is at the heart of the widespread discomfort about autonomous vehicles. We know they will kill people, but we also know they won’t feel anything when they do. At the same time, they will be saving lives, but they won’t feel anything about that either.
Although these are still very early days in the development of autonomous driving technology, there are already a lot of these vehicles on the roads around the world. Here in the state of California 52 corporations hold autonomous vehicle permits. In San Jose, California where I live, I have many times seen autonomous vehicles on our streets, but they always have a human driver as well. That is still required by law here, I believe. Of course, many manufacturers already provide limited computer-controlled self-driving capabilities such as: anti-lock braking, automatic emergency braking, adaptive forms of cruise control (great for traffic jams), automated lane minding, and the most advanced I have seen to date, fully automated self-parking. So computers are helping us now, and it seems likely that in the “near future” they will just do all the driving for us completely, maybe sooner than expected.
When will autonomous cars be able to drive themselves without any human involvement? Most experts have been predicting that this is only a few years away. Once autonomous vehicles are provably much better drivers than humans (which I think is inevitable) and once humans start being more trusting of the technology (I bet that will be slow coming for many people) then they will start rapidly displacing human drivers from our roads. Some cars will start to be built without human-accessible driving controls. Autonomous vehicles will be able to drive faster safely, form tight ad-hoc convoys or plattoons at speed, and do other things humans cannot do. Eventually those anachronistic few humans that cling to their drivers licenses will come to be viewed as wild and dangerous people. Hopefully they will have to pay high license (and insurance) fees to have the right to put the rest of us at risk. Eventually there will be autonomous-only roads. Today you are not allowed to ride your bicycle (or drive your 1927 model A Ford with a top speed fo 25 mph) on our freeways. In future you will likely be allowed to drive yourself, but only along certain roads.
If you currently have a career driving something (taxi, limo, uber, lyft, bus, long haul truck, delivery van, etc.) you might want to start considering a backup plan for your career.
I am keenly interested in this subject because I am a parent of a 13 year old male. I feel like autonomous driving technology is racing against my son’s age. I am very eager for the technology to win so that he will not have to drive a vehicle in his teens. I was a teenage male driver once, a long time ago. Looking back on it, I was pretty lucky to have survived through that (as were my passengers, and all other people on the roads near me at the time). In case you think I am over-reacting, or generalizing from my sample size of one, the U.S. CDC statistics clearly show motor-vehicle accidents are the number one cause of death for teens. I want to spare him from those risks so he can more energetically indulge in all the other risky teenage craziness.
So what do you think about self-driving cars? Are you eager to be an early-adopter and hop on board for a ride? Or do you want to wait for clear evidence that they are better than us at driving before you take the plunge? Or perhaps you just love driving and never want to give it up? Please leave a comment below to share your views and continue the conversation.
I’m very curious and always adaptive for any new technologies. As to the autonomous vehicles, I would certainly like to give it a try as soon as there are more clearer evidence to show they can drive better than us.
I think autonomous vehicles make sense as baby boomers hit 80 years old. I’ll need mine in 20 years. Also maybe smaller more lightweight vehicles would do less damage. Think mobility scooter in full auto.
We teach our kids to look both ways when crossing roads on foot. At night, with its headlights on, I would have thought that car would be very visible from quite a distance. Unless there is something about the curvature of the road that prevented visibility, I’m going to suggest the pedestrian was at fault. As a pedestrian and bicycle user, I don’t expect cars to do anything other than obey traffic signals and not ride onto sidewalks/bike paths; I don’t expect them to see me as anything other than part of the stationary background.