Americans’ views of artificial intelligence: identifying and measuring aversion

Abstract This study explores the phenomenon of artificial intelligence (AI) aversion within the context of public policy, building on prior research on algorithmic aversion. I aim to establish a clear conceptual distinction between algorithms and AI in the public’s perception and develop a robust metric for assessing AI aversion. Utilizing a national survey, I employed affective imagery testing to compare Americans emotional responses towards AI, algorithms, and advanced technology. The findings reveal that AI elicits significantly more negative emotional responses than the other two, indicating its unique position in public perception. I then construct the Artificial Intelligence Aversion Index (AIAI) based on responses to policy-related vignettes. Regression analyses showed a strong negative relationship between the AIAI and public support for both current and future AI applications within public policy, with aversion more pronounced towards potential future uses. These insights underscore the importance of understanding public sentiment towards AI to inform policymaking as well as helping to establish a framework by which to evaluate aversion levels.