Self-driving cars should spare the young over the old
Ethics: if a self-driving car has to take a decision about knocking down a pedestrian, should it spare the life of a young man or an old woman?
Given the choice of programming a self-driving car, people would spare human lives over animals, would opt to save people rather than kill them, and to spare young lives over old. These moral dilemmas were outlined in an exploratory survey conducted by Nature magazine.
Once driverless cars actually take to the roads, these grisly dilemmas will become all too real. Imagine that a self-driving car is not able to save the lives of every potential accident victim: who should it save? An old lady or a young man? According to this research, automobile manufacturers and policy makers are all wrestling with the question: how do you solve these moral dilemmas?
We don’t know how often such moral dilemmas arise in traffic. Motorists who may have faced such a situation don’t always recognise it as a dilemma. Or didn’t live to tell the tale. But even if we seldom face such a dilemma, you do need to know in advance what you would want to happen in such a situation. And define what the general public would want to happen. Even if you don’t want to leave these decisions to them, there has to be a consensus.
Which is why the research team developed the Moral Machine, a free web-based computer game. Players see an image of a self-driving car transporting some or no passengers. It is driving towards a number of people on a zebra crossing. The pedestrian signal is either red or green. The players have to choose: should the car kill everyone on the zebra crossing or smash into a barrier and kill all the passengers in the car?
Whose lives should a driverless car spare in an accident that cannot be avoided? People are most comfortable sacrificing a cat, and would rather see an adult die than a child.
NRC 251018 / DM / Source: Nature
The study analysed almost 40 million decisions from people in 233 countries. It revealed three main geographical clusters (Western, Eastern and Southern) with some preference nuances. Eastern countries, for instance, showed a slightly lower relative preference to spare young lives.
The survey is very much a first attempt. Over two-thirds of respondents were male, most of them under 25. The scenarios were limited: only two options (drive on or take avoiding action). That some will live and some will die (again only two options) is a certainty, but all the characters are anonymous and allocated to a limited number of stereotypical categories. There are no indications of skin colour or other physical features. If the research had included these factors it would no doubt have generated moral objections. On the other hand: it already mentions the fat, the old and the sporty people. It’s a pretty crude survey, no two ways about it.