Causal Beliefs about the Economic Effects of AI among Politicians and the Public

Governance of AI will be shaped, in part, by the causal theories people hold about the technology’s economic effects. Using parallel surveys of local Canadian politicians (n=1,100) and their constituents (n=5,510), coupled with latent class analysis, we identify distinct causal theories and document systematic differences in how elites and citizens theorize about AI’s effects on firms, consumers, and workers. We further show that these differences translate into distinct policy preferences. Politicians most often perceive technology as complementary to workers and favour long-term policy responses, while citizens are more likely to view technology as substitutionary and prefer protections against short-term harm. These gaps persist after adjusting for key compositional differences between groups. Finally, we also find the divide is context-dependent: in communities marked by high levels of exposure to earlier technology (i.e., automation), politicians are more skeptical of technology’s economic benefits, suggesting that heightened visibility of technological change could diminish political support for AI.