iGoofed: How Consumers Assign Blame in Response to AI Accidents

Artificial intelligence (AI) agents are becoming pervasive in a variety of domains, including automobiles, medicine, and housekeeping. However, AI agents are not perfect and as brands increasingly employ AI solutions, they will find themselves having to manage consumer responses to accidents involving AI agents. This research examines the differences in consumer blame of AI agents compared to humans for accidents, as well as important reputational and financial consequences for brands arising because of these differences in blame. Across seven studies—including one that compiles and analyzes a database of Twitter posts—this paper demonstrates a “blame penalty” for AI agents compared to humans, and for the brands that employ them. It further demonstrates the importance of counterfactuals in explaining these blame differences, and it provides two interventions to attenuate or reverse the AI agent blame penalty. The results suggest that brands using AI agents and wishing to avoid undesirable consequences may need to consider how they will respond in the face of an accident.