AI, Judges and Judgment: Setting the Scene

Abstract

Artificial intelligence (AI) in the administration of justice is growing at rapid pace. This is driven by widespread recognition of AI justice’s undeniable advantages, despite the risks it presents to the integrity of legal systems.

AI justice may, for example, lower the administrative burden of cases. The Crown Courts in England and Wales ended 2022 with a near-record load of over 60,000 outstanding cases. AI can dramatically increase court efficiency and reduce backlogs, providing standardised outcomes faster and at lower cost. After all, AI judges do not need to rest. At the same time, AI-driven judicial decision-making could make justice more accessible to the large segments of society that cannot afford human lawyers.

Proponents also argue algorithms could improve the fairness of judgements because “AI judges strictly follow precedents, restrict improper judicial discretion, prevent personal biases and preferences of individual judges, handle large amounts of information, complete complicated calculative balances, and discover statistical representations of variations of fact patterns and legal factors”. Even where AI tools assist human judges, these tools can push relevant legal provisions through comprehensive data retrieval. This in turn can improve judges’ understanding of cases, helping them avoid one-sided access to data and information.

At this point, it is important to clarify the different ways in which AI is being deployed in the courtroom. At a foundation level, AI may be used for auxiliary administrative functions. This includes communication between judicial personnel, allocation of resources and cases, and ensuring the anonymisation of judicial decisions, documents, or data. These activities may ostensibly appear separate from the core of judicial decision-making but carry subtler implications. For instance, the allocation of a case to a specific judge, given their unique expertise or biases, could indirectly influence the outcome. These nuances notwithstanding, the primary objective of these AI-driven tasks remain administrative in nature, aiming to streamline the judicial process rather than directly determine case outcomes.