Artificial Intelligence and Public Administration: Reimagining Governance in the Digital Age. State of the Art and Directions for Future Research

The topic of artificial intelligence has evolved, in the span of a few years, to become one of the most active fronts within most scientific fields. Public administration and management research has not escaped this trend. Bibliometric reviews of the field document a rapid acceleration of publications from 2019 onward, with output peaking around 2024-2025 as generative AI tools diffused into government practice (Criado et al., 2025). This growth has not yet produced a single, settled paradigm. Instead, the literature has organized itself around a recurring set of research interests, i.e. what AI is being used for in government? (efficiency gains, service personalization, predictive and decision-support applications); what conditions enable or obstruct its adoption? (organizational capacity, digital infrastructure, regulatory pressure, professional competencies); and what happens to core public values, such as accountability, transparency, fairness, trust, once algorithmic systems enter administrative decision-making? (Paglieri et al., 2025). Recent framework-building efforts have tried to hold these strands together. Scholars have proposed reading the field through the triad of administration “with, of, and through” AI, distinguishing the use of AI as an operational tool, AI itself as an object of regulation and governance, and AI as a method that reshapes how public administration research is conducted. Others emphasize that AI’s diffusion across the public administration system reconfigures the relationship between public values, institutions, and the bureaucrats who mediate between the two (Zheng, 2026), while cross-national policy reviews document how unevenly governments have translated ethical principles into operational safeguards: as of 2026, fewer than a third of OECD countries report having in place either a transparency standard or an algorithm register for public-sector AI systems (OECD, 2026).