Let’s talk about driverless cars. Why? Well, news broke that police in the US are investigating the death of a man who may have been using his Tesla’s self-drive mode. We don’t know exactly what happened yet. But this is the kind of thing we are going to hear more of and understand. Driverless cars are coming. Some people will die as a result of them. Does that make them bad?
I’d say no. Keep in mind that if someone invented the regular car today, it’d probably be considered far too dangerous. Cars kill roughly a million people a year. Driverless cars would almost certainly kill far less. But the number won’t be zero.
The ethical considerations of this are fascinating. Mostly because we have to program the car to decide what to do in extremely dangerous situations where one or more people may die.
Just because it works, doesn’t mean it’ll be popular
A runaway trolley is speeding down a set of train tracks. Five people are cowering on the tracks up ahead, certain to die – unless you pull the signals lever in your hand, diverting the trolley to a different set of tracks.
But that set of tracks isn’t empty either. Someone’s stuck there, too.
Do you do nothing, and let the trolley kill five people? Or do you pull the lever, knowing you’ll kill the other person?
In ethics, that thought experiment is known as The Trolley Problem. Until recently, it was purely theoretical – but self-driving cars are going to face problems like this every day, and for them, the dilemma might be a no-brainer.
Clearly, the public’s attitudes towards safety will play a huge part in the success or failure of self-driving cars. Engineers will have to develop their autonomous vehicles in accordance with prevailing social norms.
As of now, though, there are no social norms regarding self-driving cars. The technology is so new, and so rare, that most people don’t know how they feel about it.
Jean-François Bonnefon, a psychological scientist working at the Toulouse School of Economics, is trying to fix that. A pioneer in the field of experimental ethics, Bonnefon poses ethical dilemmas to large numbers of people and notes their responses.
Hundreds of respondents were asked to consider a number of Trolley-style scenarios. Should a self-driving car swerve into a wall, killing its passenger, if it would save the life of one or more pedestrians?
What about three pedestrians? How would you feel if it were your wife or child driving? Would that change your response?
In the main, respondents were happy for self-driving vehicles to minimise casualties. If an on-board computer could save lives, it should do so, even if that meant swerving to hit a passer-by.
This utilitarian approach was popular, but, like so many virtuous positions, people supported it as long as it didn’t impact them in any way. Bonnefon’s report notes that respondents “actually wished others to cruise in utilitarian autonomous vehicles, more than they wanted to buy utilitarian autonomous vehicles themselves”. As paper co-author Iyad Rahwan put it, “Most people want to live in in a world where cars will minimize casualties, but everybody wants their own car to protect them at all costs.”
This is a complex topic, and Bonnefon’s work represents the first timid steps towards an ethics of automated transport. The psychologist’s work was fairly simple, but there are dozens of variables: the age of the potential victims, the safety rating of the cars themselves, or the relative probabilities of different victims surviving an impact. Expect to see dozens more studies of this type – and expect them to shape the autonomous vehicle industry.
This paper is where market research, philosophy, and safety converge
If the autonomous car industry grows to the size most analysts expect it to, Bonnefon’s paper will become an important historical relic. It will likely be cited hundreds of times, engender dozens of responses and change the course of science in this field.
Data like this could be gold dust to companies looking to stamp their mark on the autonomous vehicle market – but Bonnefon’s unlikely to get rich off his research. A company the size of Google or Tesla is more likely to carry out any necessary research under its own steam. Academic institutions won’t have the financial firepower to compete. The autonomous vehicle market is projected to be worth $42bn by 2025 – you can be sure those companies won’t be outsourcing research that affects their future earnings so directly.