ARTIFICIAL INTELLIGENCE IN PUBLIC ADMINISTRATION: comparative lessons for Contemporary Public Governance

Abstract

This research investigates the sociotechnical transition driven by the incorporation of algorithmic systems into the State and its impacts on decision-making capacity, policymaking, public service delivery, and governmental oversight. The study, of paramount importance given the risk that technological efficiency might supplant constitutional guarantees and perpetuate historical asymmetries through biased automation, seeks to reconcile innovation with technological due process. Employing a qualitative methodology with a theoretical-comparative and analytical-deductive approach, the research integrates a literature review, the examination of global (OECD, UNESCO) and national regulatory frameworks, and the scrutiny of empirical international cases (such as the success of the Estonian model and the failures of Robodebt in Australia and SyRI in the Netherlands) alongside Brazilian experiences (ALICE at the CGU and ChatTCU). It concludes that technological precision must not be conflated with administrative justice, and that the adoption of Artificial Intelligence requires an augmented State action grounded in transparency, algorithmic accountability, and the continuous evaluation of its impacts (ex ante). Innovation in the public sector, structured within a cooperative federalism, does not reside in the substitution of public officials or in the unrestricted delegation of authority to black boxes, but rather in the construction of a robust governance infrastructure that preserves the centrality of human judgment, thereby protecting fundamental rights against automated exclusion.