The human factor in AI governance: Dispositional predictors of public concern and regulatory support

This study uses a 4 × 2 factorial experiment (N = 1,640) to examine how perceptions of AI as autonomous agents versus decision-support tools, along with application context (medical diagnosis, screenwriting, law enforcement, or genetic screening), influence public concerns about AI and support for its regulation among U.S. adults. Counter to theoretical expectations based on algorithm aversion literature, neither AI agency nor application context produced significant main effects on public attitudes. In contrast, three key dispositions consistently predicted outcomes across contexts: authenticity preference, concern about disproportionate impacts on vulnerable groups, and general attitudes toward AI. These effects remained consistent across all experimental conditions, indicating that public responses to AI are fundamentally driven by pre-existing individual dispositions rather than contextual features or implementation approaches. These results are consistent with moral foundations and mind-sponge accounts of value-based judgment. The findings imply that risk-tiered regulatory frameworks, while necessary, are unlikely to be sufficient on their own; durable governance will require deliberative engagement with the authenticity and equity intuitions that drive public response, and longitudinal attention to how those orientations are themselves reshaped by sustained algorithmic exposure.