This paper aims to reconstruct the current state of the art regarding the application of artificial intelligence within administrative procedures. After outlining the relevant regulatory framework and the main theoretical coordinates, the analysis focuses on the distinction between bound and discretionary administrative activity, examining the legal, organisational, and accountability implications arising from the use of algorithmic systems in both contexts. The paper concludes by offering some systematic considerations, identifying transparency and the duty to provide reasons as the essential safeguards to ensure the legality and legitimacy of administrative action in the era of algorithmic governance.

A public-private company cannot participate in any tender procedure other than the one for which it was established

Article 80, paragraph 5, letter c) of the Code of Public Contracts provides for exclusion from participation of economic operators who have been convicted of serious professional misconduct affecting their integrity or reliability. The rationale for this prevision is «the need», of the public administration, «to ensure the reliability of those who propose themselves as contractors» (Council of State, decision no.1412, of April 11, 2016) through an evaluation process to determine their integrity. This provision has been the subject of a recent decision by the Regional Administrative Court for Sardinia: no. 646 of October 3, 2022. That decision focused on the discretion left by this provision – at the point where it provides for the exclusion of the economic operator for serious professional misconduct on the basis of any means considered by the Contracting Authority as adequate. The decision has clarified how this evaluation presupposes a specific reasoning based on two levels: one objective and one relative, which considers both the gravity of the professional misconduct committed by the economic operator, and the prejudice that it may cause to the specific contract in question.