Abstract

Earlier technological shocks displaced low-skill or routine workers and complemented cognitive workers, generating a deep political realignment in response to the rise of the knowledge economy. Generative artificial intelligence (GenAI) now affects skilled cognitive occupations, potentially generating different economic winners and losers of technological change. To examine how AI-exposed workers may react politically to realizing the potential of this technology in their job, we conduct a pre-registered survey experiment on approximately 6,000 workers in 98 AI-exposed occupations in Germany, Italy, and the United States. Treated respondents watch an occupation-specific video of ChatGPT performing their most frequent core task---a non-directional stimulus that can shift perceptions in either direction. Using principal stratification, we show that the treatment realigns workers: those pushed to feel more threatened increase support for AI regulation, automation taxes, codetermination, and trade protectionism, while those pushed to feel more optimistic move in the opposite direction. Effects concentrate on AI-specific ``new policies’’ rather than traditional welfare state instruments. Threatened workers also warm toward backlash parties. We tentatively conclude that GenAI exposure may not simply feed into the political cleavages of the knowledge economy but create demand for new technoloty policies.

Reference

Aina Gallego (Universitat de Barcelona, Spain), “What if you see it? Workers’ perceptions of and reactions to Generative AI”, IAST General Seminar, Toulouse: IAST, April 14, 2026, 11:30–12:30, room Auditorium 4 (First floor - TSE Building).