Public Budgeting in the Age of AI: A Case Study of Maryland's Budget Process

• In Event: AI in Public Procurement, Budgeting, and Government Operations

Abstract

Public budgeting processes are critical government functions persistently undermined by three core complexities: cognitive and institutional limitations, rigidities for fiscal accountability, and political and bureaucratic gamesmanship. These theoretical challenges manifest as concrete operational bottlenecks, fragmented legacy systems requiring exhaustive manual reconciliation, fractional fiscal oversight, incremental decision-making heuristics, and information asymmetries that enable bureaucratic gamesmanship. This paper bridges theory and practice by employing a case study of the State of Maryland's budget formulation and execution processes, drawing on practitioner interviews with staff from the Department of Budget and Management, the Comptroller's Office, and the Department of General Services, alongside a systematic review of the public budgeting and AI literature.

Against this empirical backdrop, we evaluate how Artificial Intelligence can mitigate each area of complexity using the Technology-Organization-Environment (TOE) framework as our analytical lens. The TOE framework allows us to assess AI adoption not merely as a technical question but as one shaped simultaneously by the maturity of available AI tools, the organizational capacity and institutional readiness of state budget offices, and the broader environmental pressures of fiscal accountability and political context. Drawing on both interview evidence and the existing literature, we demonstrate that different layers of budgetary complexity require different levels of AI capability, and that the conditions for successful adoption vary significantly across these three dimensions.

Importantly, we also surface critical governance risks, including algorithmic administrative errors, data security vulnerabilities, and the erosion of human oversight, arguing that responsible AI deployment in public budgeting demands as much institutional design as technical implementation. This paper contributes to the emerging literature on AI in public administration by providing one of the first theoretically grounded, empirically informed analyses of AI applications across the budget cycle, especially from formulation to execution at the state government level. Our findings have direct implications for budget practitioners, public financial management reformers, and scholars examining the governance of algorithmic systems in high-stakes administrative contexts.

Author

• Salman Alam, University of Maryland, College Park
  Presenting Author