Integrating Civic and Artificial Intelligence in Policymaking: Experimental Insights on Public Perceptions of Policy Proposals

The integration of artificial intelligence (AI) into policymaking raises pressing questions about whether it can reconcile technocratic efficiency with democratic legitimacy. While existing literature offers mixed assessments, few studies systematically examine how AI, especially when paired with democratic processes, shapes public perceptions of policymaking. To address this gap, we apply the output–input legitimacy framework to assess how different AI deployment conditions influence public attitudes toward policy proposals. Using a vignette-based online experiment with a US representative sample (N = 3055), we combine descriptive statistics, causal inference, and dimensionality-reduction techniques. Our results offer a multidimensional perspective on how public perceptions are shaped by the interaction of individual characteristics, technological conditions, and procedural design. While AI treatments generally yield directionally more favorable responses, these effects are filtered through attitudinal and identity-based lenses, whose salience varies across contexts. This perspective enriches theoretical debates and highlights the complexity of building legitimacy in an era of algorithmic governance. It also challenges binary narratives that portray AI solely as a technocratic solution or democratic threat. These insights also offer practical guidance for the democratic integration of AI in public decision-making.