What If You See It? Workers’ Perceptions of and Reactions to LLMs

• In Event: Workers, Robots, and International Political Economics

Abstract

As Large Language Models (LLMs) increasingly demonstrate their capability to perform diverse cognitive tasks, they have sparked both excitement and concern regarding their economic potential and the possibility of substituting white-collar labor. This study investigates two critical questions: How do workers perceive LLMs – as a career opportunity or a threat? And do these perceptions influence their policy preferences? Prior research on workers displaced by routine automation reveals a general lack of awareness about the sources of labor market risks. Responses to job displacement have varied, ranging from political disengagement to demands for economic redistribution, and even resistance to immigration and trade. However, the extent to which these findings apply to workers potentially impacted by LLMs remains uncertain. This group, often highly educated, might exhibit greater awareness and connect labor market challenges to policy solutions more effectively. Yet, they may also have a strong incentive to safeguard their investment in specialized occupational skills. To explore these dynamics, our study introduces an innovative informational treatment. We inform workers across 176 occupations about the capabilities of LLMs to perform the most frequent tasks in their respective fields. This treatment is embedded into a survey experiment conducted across three countries, enabling us to identify which workers predominantly view LLMs as either an opportunity or a risk. Then, we analyze how these perceptions shape a range of policy preferences. This research offers valuable insights into the evolving relationship between advanced AI technologies and the labor market, highlighting the nuanced perspectives of workers in an era of rapid technological change.

Authors

• Aina Gallego, UB and IBEI

• Italo Colantone, Bocconi University

• Massimo Anelli, Bocconi University

• Piero Stanig, Bocconi University