How AI Affects the Legitimate Authority of Courts

The use of AI in courts is regarded with ambivalence, and assessed by a confusing compilation of normative standards and analogies. The paper suggest that our joint reflections improve by taking as fundamental the question of how AI affects the legitimate authority of courts, for better and worse. This helps bring order to the various standards, and sheds light on when trust in the use of AI by courts is justified – and why that matters. The paper start by asking when and on what grounds do court and other adjudicatory bodies [henceforth courts] make justified claims to be legitimate authorities – i.e. that their subjects have reasons to defer to their judgments and interpretations. Following Joseph Raz’ proposal, this is centrally a question of whether the authority enables the subject to better act as it has appropriate reasons to – including self- and other-oriented interests. Courts may provide such services when they adjudicate impartially based on legal rules by judicial method, and when they offer interpretations of those legal rules. The paper then asks how AI may strengthen or challenge such justifications for the legitimate authority of courts, by influencing how, and how reliably the judges and the judiciary provide those services – and how the affected parties can be assured thereof – hence the role of trust. One general finding is that legitimacy is more at stake when AI shares or takes over the authority hitherto enjoyed by judges – particularly their discretion of various kinds which has legal implications. The paper addresses five stages/aspects of adjudication where some of these aspects of legitimacy are at stake in ways affected by AI: The legal process (case filing, allocation of judges ..), legal problem solving (identification of issues and legal rules), harmonisation (based on 2nd order considerations regarding legal problem solving); and issuing justified judgment and authoritative interpretation – and then legitimation/maintaining legitimacy by showing that trust is deserved in the particular decision and in the judicial system. The paper ask under what conditions various uses of AI can strengthen – or weaken – these requirements at the various stages.