Managers as Gatekeepers in the Age of AI

• In Event: Labor Impacts of AI

Abstract

Despite widespread expectations that artificial intelligence (AI) will transform productivity and displace workers, organizational adoption remains uneven and aggregate impacts are mixed. While individual-level experiments document substantial productivity gains, economy-wide evidence is more modest. We argue that middle managers—who hold day-to-day authority over technology deployment and staffing—serve as key gatekeepers in this disconnect. 
We test this mechanism using a pre-registered survey experiment with 2,000 managers across the United States and United Kingdom. Respondents were randomly assigned to view information about AI’s productivity benefits, its labor-displacing potential, or a neutral control. Managers exposed to labor-displacement information substantially reduce their intentions to adopt and advocate for AI (by 0.4–0.5 standard deviations) and moderately reduce staffing intentions (by 0.2 standard deviations). Mediation analysis indicates these effects operate through both decreased perceived benefits and increased perceived risks of AI, suggesting that displacement narratives shift fundamental beliefs about the technology. 
In contrast, information about productivity benefits has no meaningful average effect on AI adoption or staffing, consistent with managers already holding optimistic priors. However, such information does increase AI advocacy among managers with lower prior familiarity, indicating that responsiveness depends on baseline exposure. At baseline, managers report high rates of AI use, broadly positive attitudes, and expectations of continued expansion, limiting the scope for further upward revision. Importantly, displacement information does not trigger substitution from labor to AI. Instead, managers retrench on both AI adoption and staffing simultaneously, suggesting a broader pullback driven by uncertainty or implementation concerns. These findings highlight the central role of the information environment—particularly the growing salience of displacement narratives—in shaping managerial decision-making.To our knowledge, this is the first cross-country randomized experiment isolating how information causally affects managerial AI adoption and staffing intentions, providing new evidence on how organizational gatekeepers shape the economic impact of AI.

Keywords: Artificial Intelligence, Managers, Gatekeepers

Authors

• Yong Suk Lee, University of Notre Dame
  Non-Presenting Co-Author

• Cassandra Merritt, University of Notre Dame
  Presenting Author