AI in the Public Sector: Cross-National Evidence on Bureaucracy, Procurement, and Organizational Change

Session Submission Type: Panel

Abstract

As artificial intelligence (AI) is increasingly adopted across the public sector, it raises pressing questions about its impact on public servants, organizational behavior, and institutional governance. This panel brings together four empirical studies that examine how AI is reshaping public administration. Organized into two thematic pairs, the panel spans cross-national cases and multi-level institutional settings, ranging from the lived experiences of public employees to the contracts and organizational systems through which AI is implemented.

The first pair of papers explores how AI affects public sector workforces in two national contexts. William Resh and Yi Ming analyze the U.S. federal workforce, using large language models and occupational data to project how generative AI will change required competencies and expose certain roles to disruption related to automation. In parallel, Andong Zhuge, Shuping Wang, and Jiaheng Ling examine smart city reforms in China using a staggered difference-in-differences design. They find that AI adoption significantly reduces the well-being of public servants, particularly among frontline workers, due to increased workloads and anxiety related to automation.

The second pair turns to how institutions and organizational contexts shape the design and reception of AI systems. Aniket Kesari and Jae Yeon Kim examine institutional variation in AI procurement by conducting computational text analysis of nearly 700 U.S. state-level contracts using large language models. Their findings reveal that while contracts often prioritize predictive performance, they rarely include enforceable provisions for fairness, transparency, or privacy, suggesting that current procurement practices fall short of safeguarding equity in public AI use, despite their potential to do so. Kyoung-cheol (Casey) Kim investigates how organizational context influences individual attitudes toward AI using a survey experiment with employees from both public and private institutions in Taiwan. His results show that exposure to AI increases support for its organizational use, with stronger effects among public-sector employees, highlighting how institutions shape behavioral responses to AI.

Spanning China, Taiwan, and the United States, this panel brings together diverse national and institutional contexts to examine how governments adopt and govern AI. In the U.S. alone, the papers cover federal, state, and local levels, highlighting the layered architecture of implementation. Methodologically, the panel features computational modeling using large language models (LLMs), contract analysis with LLMs, causal inference through difference-in-differences, and survey experiments. Substantively, the papers explore AI’s effects on frontline well-being, organizational attitudes, procurement practices, and workforce planning. Together, they offer a multidimensional and comparative perspective on algorithmic governance, with both theoretical insights and practical implications for scholars, policymakers, and public administrators.

Policy Area

• Public and Nonprofit Management and Finance

Secondary Policy Area

• Innovations in Science and Technology

Chair

• Daniel S. Schiff, Purdue University

Discussant

• Yi Ming, Georgia State University

Organizer

• Jae Yeon Kim, Harvard University

Individual Presentations

• A Task- and Competency-Based Analysis of Generative AI's Impacts on the United States Federal Workforce - Presenting Author: William G Resh, Georgia State University; Non-Presenting Co-Author: Yi Ming, Georgia State University

• AI Paradox: The Negative Impact of AI Usage on Public Servants' Well-being - Presenting Author: Shuping Wang, Syracuse University; Non-Presenting Co-Author: Andong Zhuge, Fudan University

• Contracting for Fair AI: A Computational Text Analysis of State AI Procurement Contracts - Presenting Author: Jae Yeon Kim, Harvard University

• Cognitive Impacts of AI on Administration - Presenting Author: Kyoung-cheol (Casey) Kim, University of Central Florida