Accountable AI in Public Administration: A Bibliometric Review of Governance, Transparency, and Public Value

As AI is being more deeply embedded in public administration, the very nature of public authority is being reshaped from a reliance on human discretion to a reliance on classification systems based on data. While there are many efficiencies to be gained from such a system, there is a fundamental governance challenge to be addressed: how to leverage AI systems without undermining due process, democracy, or the legitimacy of public decisions. To begin to address these questions, the present study undertakes a bibliometric study of 127 Web of Science records to explore the intellectual structure of a field that is rapidly taking shape. From the results of the study, the research suggests that the field is being developed on the foundation of three core pillars: the development of institutional capabilities for AI governance, the maintenance of the principle of “good administration” through the redesign of discretion, and the leveraging of transparency to generate public value. Ultimately, the study suggests that the effective governance of AI systems in the public sector is not merely a matter of abstract principle, but of operational reality.