An Experimental Study on the Attribution of Autonomous Driving Accidents by Multiple Subjects

With the increasing adoption of artificial intelligence in public safety and urban governance, autonomous driving systems are becoming an important component of public service delivery. At the same time, accidents involving autonomous vehicles raise new challenges for responsibility attribution, particularly in complex governance settings where multiple human and technological actors are involved. Existing studies have largely focused on single responsible actors or simplified accountability scenarios, leaving limited understanding of how the public distributes blame across multiple subjects in real-world contexts. Drawing on a behavioral public administration perspective, this research uses a series of survey-based experiments to examine how the public attributes blame to different actors in autonomous driving accidents. The study explores how variations in agent type (human vs. technological), accountability chain structure, perceived algorithmic controllability, and information framing shape blame distribution patterns among multiple responsible subjects. By systematically manipulating these factors, the project seeks to identify key psychological and contextual mechanisms underlying public responsibility judgments. The expected outcomes of this research include a clearer understanding of how technological agency alters public blame attribution, how accountability structures condition responsibility judgments, and how information presentation influences attribution patterns. The findings aim to contribute to theoretical discussions on responsibility and accountability in AI-enabled public services, while also offering practical insights for the governance and regulation of autonomous driving technologies.