Abstract
<p>Political disagreement over artificial intelligence may reflect different expectations about its economic consequences or judgments about appropriate government responses. Using parallel surveys of local Canadian politicians (n=1,100) and the public (n=5,510), and latent class analysis, we identify four causal theories of AI’s economic effects. Politicians are much more likely to view AI as complementing workers, while the public more often sees it a substitutionary, even after compositional adjustment. Causal theories further organize policy preferences: those with substitutionary beliefs favour policies restricting AI, while those with complementary beliefs favour adaptation. Yet politicians and citizens with similar causal theories also emphasize different policies, and these within-class differences account for a substantial share of policy gaps. Elite-public policy disagreement therefore reflects both different economic diagnoses and responses to shared diagnoses. Finally, politicians in communities shaped by past economic shocks hold more substitutionary views, suggesting political responses may change as AI’s distributional consequences become visible.</p>