suggested citation: K. De Blasio, Comparative remarks on the regulation of predictive justice, lceonline (www.lceonline.eu), 3/2024, II/Rubriche, p. 1 ss.
key words: predictive justice, artificial intelligence, risk assessment tools, European Union, United States
abstract: The availability of artificial intelligence software is increasingly permeating the law. In particular, predictive justice has become a popular topic among legal scholars in the last few years because of the availability of artificial intelligence programs allowing users to predict the outcome of a trial or the risk of recidivism during the trial or pre- and post-trial phases. These tools raise many issues in terms of discrimination, lack of transparency and crystallisation of the law, especially because studies show how difficult it is for a “human” judge to depart from the outcome of a highly technological tool. This paper reviews some examples of the implementation of predictive justice programs in different legal systems, as well as the regulatory instruments that have been introduced in order to reduce the risks generated by their use. To this end, documents containing ethical principles have been adopted, although their effectiveness is limited by their non-binding and generic nature, which make them difficult to embed in artificial intelligence tools. Therefore, some strict statutes have been introduced by the European Union and some Member States. These statutes seem to prohibit some uses of predictive justice tools, although it is doubtful that this is the best solution, since these bans can be easily circumvented. Therefore, we argue that the preferable solution is to stimulate cooperation between technical experts and legal experts, to promote greater awareness among them of the applications and implications of artificial intelligence in judicial proceedings, a solution that is already being pursued in the United States.
