Who should a runaway AI driverless car kill?

Oct 29, 2018, 8:52 AM EDT
(Source: smoothgroover22/flickr)
(Source: smoothgroover22/flickr)

The researchers at the Massachusetts Institute of Technology conducted an online survey with two million people, presenting them with different variations of the classic “trolley problem” thought to know what moral decisions a runaway self-driving vehicle should make.

Participants, asked to choose between two options of who should be hit by an AI-powered out-of-control vehicle, showed “some striking peculiarities” with preferences varying greatly across different regions, notes The Guardian.

Although researchers noted three key areas of consensus – saving humans over other animals, larger groups over smaller ones and children and young over the elderly – responses varied across regions and cultures based on factors such as economic equality, status, the criminal record of the individuals, etc., reports Dezeen.

The findings of the survey hold are crucial for auto manufacturers seeking to design the AI software that meshes with human values and ethics.