Predictive Justice and Work

In EU Regulation No. 2024/2831 (the so-called AI Act), the AI systems “intended to be used by a judicial authority or on their behalf to assist a judicial authority in researching and interpreting facts and the law and in applying the law to a concrete set of facts” are classi ed among the “high-risk” systems. The rightful concern of European policymakers for such use of AI provides the chance to re ect on the bene ts and limits of so-called predictive justice in the labor law eld. This article advances the distinction between technologies that replace (“robot judge”) and those that merely supplement (“robot clerks”) the person of the judge in the adjudication of labor disputes. Whereas the former systems seem at odds with the due process principle, the latter might also be problematic and risky, mainly due to the natural tendency of the person of the judge to passively rely on their “automated assistant.” The article contends that the value of justice is rooted in the single case and not in aggregated data; most important, the article notes that the capacity of the overall legal order to develop through the dialogue of its components should not be sacri ced in the name of celerity and certainty, especially when the latter goals are pursued through a probability-based and rather opaque algorithmic system.