The Brain of a Four Year Old
Recently, I recreated the classic trolley problem using a wooden Brio train set. A train is hurtling down a single track. Up ahead, I placed five Playmobil people on the track. Between the train and the five figures is a junction. The junction leads to a section of track with plastic giraffe on it (we had run out of figures).
The question I posed to my pre-school tyke was what will you do? Will you put the train down the first track and hurt five people, or... LISTEN, LISTEN! Or, will you put it down the second track and hurt the giraffe?
You can play along too and decide for yourself which of the tracks you would send the train down.
My genius son knew immediately what he would do. He picked up the entire track and shook it until it fell apart, thus rejecting the problem entirely. A pair of black sunshades appeared from nowhere on his face and everyone clapped.
Why am I telling you this? Apart from boasting about the little rascal's genius-level intellect, it's because I can't drive I am interested in self-driving vehicles.
Mo' Trolleys, Mo' Problems
The type of Trolley Problem often associated with self-driving cars is one called The Tunnel Problem. What's unique about this problem is that the choice is not between you murdering more or fewer people, but between your own self-driving car murdering a stranger, or murdering you.
The car is approaching a tunnel. Suddenly, a mother with a pram steps in front of the vehicle in order to cross the road in front of the tunnel. Should the car protect the passenger (you), by ploughing into this innocent mother+baby combo, or should the car drive directly into the tunnel wall, killing the passenger inside (you)?
Most people take a utilitarian view of the first problem. Five lives are "worth" more than one, so flip the switch, and kill the one person. Job done.
For the second problem, most people abandon the sober, rational calculation, take the selfish view and say push that baby under this car and let me ride my self-driving chariot to safety town.
It shouldn't really be surprising that when it comes down to it, we abandon utilitarianism for our own survival. What is surprising is that when we discuss self-driving cars, Trolley Problems come up at all.
Let me explain
Trolley Problems are the biggest red herring in any discussion about autonomous vehicles. 96% of all road accidents are classed as preventable. This means that if the driver had kept to the speed limit, stopped the car before fiddling with their phone, or taken a taxi after having had too much to drink, then they would have gotten home safely instead of into an accident.
The remaining four percent of road accidents are divided into causes such as mechanical failure, adverse weather conditions, the physical condition of the road, or, yes, people walking in front of a tunnel.
The fact is, Trolley Problems are vanishingly rare in real life. Autonomous vehicles don't get tired or drunk. They don't get distracted or drive over the speed-limit. A self-driving car will never be the cause of these types of accidents, not because they have some moral reasoning module or follow some prime directive of protecting the passenger at all costs, but because they simply don't get into the situation where these accidents will occur.
Accidents caused by other road-users appearing from "nowhere" can also be reduced. A human can't look through a poorly-maintained, over-grown shrub to see a hidden cyclist, or round a corner obscured by a wall, to find a semi-trailer, but an autonomous car can use various technologies to do just that. Self-driving cars can prevent accidents from happening that for a human driver would be unavoidable.
Trolley Problems can't occur if we are able to mitigate them. Ethics and morality modules aren't needed if we avoid situations requiring them.
You can't solve this problem either, driver
There will still be cases where technology cannot be employed to circumvent a potential accident, such as the time a deer jumped straight off a cliff and onto the bonnet of my friend's car.
In the title of this post, I boasted that you can't solve the trolley problem. But I just did! I hear you cry, you walked us through the scenarios and I solved it every time! Well, you certainly solved it sitting down at a desk or kitchen table with your head shoved up your phone, but would you be able to solve it while driving a car at speed?
40% of drivers don't hit the breaks during a car accident. People just have a tendency to tense up and do nothing. Hitting the breaks is the simplest, easiest thing to reduce the horror of an accident, and yet a little under half of us can't manage to do it. In fact, in some accidents the driver actually presses sharply on the accelerator, increasing the chance of collision. If a car simply applies the breaks to avoid collision, the number of accidents can be reduced by over 40%. When Tesla introduced auto-pilot, similar results were also found.
Self-driving cars don't have to be perfect and they don't have to solve myriad trolley problems. They just have to be a better driver than you, and at least in my case, they already are.
Top comments (0)