The Effects of Generative AI on Administrative Discretion and Decision-Making Quality: A Systematic Study

Abstract

This research aims to analyze the effects of generative artificial intelligence (GAI) on administrative discretion and quality as well as decision-making trust in public administrations. It addresses a gap in the literature related to the lack of integrated modeling of the mechanism linking the use of AGI, the evolution of administrative discretion and the quality of public decisions. The research question is as follows: how does the use of AGI transform administrative discretion and what effects can this transformation produce on decision-making quality?

The study adopts a qualitative approach based on a systematic literature review in accordance with the recommendations of PRISMA 2020. The documentary corpus, made up of several scientific bases and studies selected according to explicit inclusion and exclusion criteria. This corpus mainly collects resources dated from 2019 to 2026. The literature review thus highlights that AGI could improve analytical skills, reduce information overload, and increase decision-making speed. etc., but that the overly systematic use of the recommendations produced restricts the autonomy of judgment, increases automation biases and shifts administrative responsibility. Administrative discretion thus plays a central role, making the link between the use of AGI and the quality of the decision. The results also highlight the importance of human supervision, explainability and calibrated trust in AI. On a practical level, the study recommends that administrations favor an AI that increases human judgment rather than replacing it, in order to preserve accountability, fairness, and legitimacy in decision-making.

Classification JEL : H83, D73, O33, D83  

Paper type: Theoretical Research