L’autolimite algoritmico alla pubblica amministrazione

Abstract

The considerations carried out in this work arise from a specific scenario that generates doubts about the applicability of administrative automation in discretionary activity, and focuses on a specific question: to what extent the public administration, once it has decided to adopt artificial intelligence tools for the management of a specific procedure, can deviate from the output produced by such tools. Basic question that underlies a plurality of issues that refer to classic institutions of administrative law, but also recently formulated principles. Among the traditional concepts we find the notion of discretion and that of self-limit; with reference, however, to the most current theories, we highlight the principle of algorithmic legality, in the particular declination of the principle of non-exclusivity of algorithmic decisions and the role of the human being in automated administrative procedures.