The Blame Game: Double Standards Apply to Autonomous Vehicle Accidents

Abstract

Who is to blame when autonomous vehicles are involved in accidents? We report findings from an online study in which the attribution of blame and trust were measured from 206 participants who studied 18 hypothetical vignettes portraying traffic incidents under different driving environments. The focal vehicle involved in the incident was either controlled by a human driver or autonomous system. The accident severity also varied from near miss, minor accident to major accident. Participants applied double standards when assigning blame to humans and autonomous systems: an autonomous system was usually blamed more than a human driver for executing the same actions under the same circumstances with the same consequences. These findings not only have important implications to AI-related legislation, but also highlight the necessity to promote the design of robots and other automation systems which can help calibrate public perceptions and expectations of their characteristics and capabilities.