Algorithmic Justice: Safeguarding the Judicial Duty to State Reasons in the Age of Automation

The use of artificial intelligence (AI) technologies such as Machine Learning and Natural Language Processing as applied to jurisprudence analyses has been on the rise for a considerable time. This is confirmed by the increasing interest shown in the topic by different bodies, such as the CEPEJ, and the emergence of the AI Act. It is clear that the introduction of AI in the context of jurisprudence should be accompanied by appropriate safeguards to avoid violations of, inter alia, due process and the obligation of motivation. In particular, Annex III of the AI Act classifies AI systems intended to assist judicial authorities in researching and interpreting facts and the law as high-risk AI systems, which must subsequently meet strict requirements of data governance, record-keeping, transparency, human oversight and so on. In addition, the recent publications of CEPEJ provide tools, good practice guidelines and action plans to steer the use of AI in jurisprudence in the right direction. The research is situated in the broader context of how automated decision-making can assist judges. The research intends in the first place to explore what requirements are under due process and the obligation to motivate and how these can be aligned with technical standards used in AI systems. To what extent will there be a match or rather a conflict between data sets and legal requirements? The topic addresses the question of how data-driven methods to support the judiciary can be reconciled with legal requirements. Besides, both potential and limits of AI systems must be mapped out, supported by interdisciplinary research in cooperation with engineers and compute scientists. Evidently, empirical research on jurisprudence analyses is taken into account.