An experimental investigation of the relationship between AI-human workflow design and legal liability for radiologists: the erroneous-change penalty and omission bias.

Clinical workflow adjustments to incorporate AI may influence liability. Using experimental scenarios, this preprint examines how different AI-human workflow designs affect radiologists' diagnostic decisions and perceptions of legal responsibility by hypothetical jurors. The findings suggest that AI-centric workflow changes involving double-reads, documentation of decision changes, and AI error rate transparency may shape clinician behavior, highlighting the need for careful evaluation before implementation.