The field of law, despite its longevity and entrenched traditional principles, stands on the threshold of profound technological transformation, notably with the advent of machine learning. This article, premised on an intrinsic affinity between legal reasoning (the derivation of rules from precedents) and the algorithmic logic of machine learning, investigates the potential of these technologies to predict judicial disputes. The primary objective is to elucidate how machine-learning algorithms—specifically the decision-tree algorithm—can be employed to assess litigation risk and assist attorneys in making more informed decisions prior to initiating litigation. The findings indicate that the deployment of such predictive tools in legal practice directly increases the predictability of judicial decisions (as a prerequisite for an efficient judiciary and legal certainty) and reduces uncertainty about case outcomes. Consequently, by enabling the early resolution of legal disputes and more accurate legal counselling, these tools substantially reduce the inflow of cases to the judiciary and materially contribute to the practical dejudicialization of disputes. Emphasizing the necessity of public access to case records and judicial opinions as an essential data substrate, this article presents the decision-tree model as an operational instrument for statistical–predictive analysis in the legal domain.
Keywords: Public law, machine learning, judicial dispute prediction, predictability of judgments, decision tree, de-judicialization.