Judicial Decisions and Artificial Intelligence

Abstract

In order to start a discussion concerning the possibility of applying AI models to judicial decisions, a necessary preliminary step is to refer to some theoretical frameworks concerning the practice of judicial decision-making and the reasonings made by judges. Of course this is not a matter of free choice or of individual preferences: what is needed is to determine which theoretical approach may produce the relatively best approximation to the empirical phenomenon of decisions made in the context of the administration of justice. This is an extremely complex and difficult problem. On the one hand, the experience of the various judicial systems (and even within a given single system) shows that judicial decision-making includes an almost infinite range of variations. The factors influencing the ways in which judicial decisions are made are numerous and include for instance the format and size of the court (single judge or panels, and so on), the composition of the court (professional and/or lay judges), several procedural rules, the factual circumstances of cases, the form and content of the substantive rules governing the case, the evidence available and the methods and standards used to decide on facts according to the proofs and to solve legal issues according to the relevant rules and principles.