2021 Volume 61 Issue 1 Pages 10-21
We examined how people attribute cause and responsibility for harm caused by artificial intelligence (AI). The participants read a vignette about a pedestrian fatality involving a self-driving car and allocated cause and responsibility to the car manufacturer and/or the user. We found that those who attributed cause to the AI also attributed cause to the manufacturer and user, and those who perceived autonomous machines to have mental capacities such as intention attributed cause to the manufacturer. These results indicate that people do not perceive AI as an agent that acts independently from humans, even though they think that AI is autonomous. We also found that those who attributed cause to the manufacturer and user also attributed responsibility to them. Moreover, those who attributed cause to the AI believed that the manufacturer should solve the problem. Thus, people presumed that the manufacturer was responsible for solving the problem, even though it was not responsible for causing the problem. Taken together, this study illustrates the importance of examining people’s attitudes when developing rules relating to AI.