IVIR

Judges across the world increasingly use various AI-tools in the administration of justice. However, litigants may be unable to contest important pieces of evidence or legal arguments when the functioning and usage of these systems remains unexplained. This dissertation discusses how a right to an explanation could enable litigant contestation of judicial AI. To that end, a comparative overview is provided of the scope, content, and restrictions of this right under due process safeguards, Data Protection Law, and AI Law in the EU, Brazil, and China. Moreover, based on four different theories of procedural justice, the normative reasons why litigant contestation of judicial AI should be enabled are also discussed. This includes analyses of utilitarian, rights-based (including a dignitarian and a Dworkinian approach), and relational approaches to procedural justice. These highlight different values of litigant contestation; it has instrumental value in error correction, and intrinsic value in respecting the dignity of litigants, either as rational autonomous agents or as socio-relational beings. However, it is technically difficult to faithfully explain the internal workings of certain opaque AI-models. Moreover, integrity and safety concerns and the potential violation of trade secrets and business interests of external developers could block the disclosure of explanations. To address these issues, this dissertation argues that judiciaries should adopt certain technical and organizational measures already during the development of judicial AI. This shows that the right to an explanation should not be conceptualized as a mere procedural right, but also as a substantive requirement that safeguards contestation by design.