Can AI agents emulate human bureaucratic motivations and loyalties?

Determine whether artificial intelligence agents deployed within government agencies can reliably emulate the motivations, loyalties, and role-based commitments historically exhibited by human bureaucrats, in order to assess their suitability for exercising discretion and sustaining accountability in public administration.

Background

Within the paper’s analysis of AGI-driven administrative decision-making, the authors note that AI agents could bring substantial improvements in scalability, cost, and output quality at the micro level of bureaucratic tasks. However, they emphasize enduring concerns central to liberal democratic governance—namely transparency, accountability, and responsiveness—when substituting AI agents for human officials.

A key unresolved issue arises from the distinctive non-technical qualities humans bring to public service roles, such as motivations, loyalties, and role-based commitments. The authors explicitly state uncertainty about whether AI agents can replicate these human commitments, which bears directly on the legitimacy and trustworthiness of delegating high-discretion governmental decisions to AI systems.