Skip to Main content Skip to Navigation

La prédiction judiciaire par les algorithmes

Abstract : Awyers have always made predictions about the judicial treatment that could be brought to a case. These predictions were often based on experience, and therefore empirical in nature. By offering a very significant computing power, artificial intelligence provides new perspectives in the field of forensic prediction. The objective of this study is to analyze the possibility of judicial prediction and to consider the implications of its use both for the office of the judge and for legal professionals. Indeed, many detractors specify that algorithmic judicial prediction would be nonsense taking into account the unique nature of each dispute decided by the judges. For critics, the legal hazard constitutes an insurmountable obstacle to this prediction. Yet judicial, empirical or algorithmic prediction is based on this hazard. Even if the hazard is a disruptor of this prediction, the familiar notion of case allows to circumvent this obstacle which does not prove, in reality, not insurmountable. The concept of similarity exists in law and the principle of equality of citizens before the courts prescribes identical judicial treatment of legally identical or similar species. Algorithmic modeling of cases with a view to judicial prediction is therefore possible even if each computer technique existing to date has its limits. However, the most recent technique of self-learning algorithms, known as machine learning, offers promising results to date. Therefore the question of the impact of the use of judicial predictions by judges and court officials arises. The main criticisms directed against the predictive applications will be studied: the performative effect of algorithms, the profiling of magistrates, the "dehumanization" of justice and the uberization of legal services through the prism of the promotion of alternative dispute resolution methods and the consequences for the legal profession.
Document type :
Complete list of metadata
Contributor : ABES STAR :  Contact
Submitted on : Thursday, January 27, 2022 - 4:05:10 PM
Last modification on : Thursday, February 3, 2022 - 3:03:11 PM


Version validated by the jury (STAR)


  • HAL Id : tel-03545971, version 1



Anaïs Coletta. La prédiction judiciaire par les algorithmes. Droit. Université de Nîmes, 2021. Français. ⟨NNT : 2021NIME0006⟩. ⟨tel-03545971⟩



Record views


Files downloads