Causal Beliefs and the Potential for Political Backlash Against AI

Abstract Artificial intelligence is poised to reconfigure the economy and politics. Although new technologies often produce net economic gains, their costs and benefits are unequally distributed, making them susceptible to politicization. We argue that whether and how AI becomes mobilized for partisan gain will depend on the public’s causal beliefs about the winners and losers of AI. We categorize these causal beliefs into four types using a novel survey instrument fielded with approximately 6,000 Americans and Canadians. Using latent class analysis, we show that while some respondents are supportive of AI, a significant portion of the public theorizes it as a threat—harming consumers and replacing rather than complementing workers’ skills. These beliefs are aligned with political preferences, predicting support for policies that delay job loss over those that help workers adapt, and polarizing voters along existing partisan lines. We conclude that fissures in the public’s attitudes toward AI already exist and are primed for exploitation by political entrepreneurs.