Should AI Decide Your Court Case? This Judge Says Yes

Geoffrey Vos, who days from retirement heads England's civil justice system as Master of the Rolls, is deliberately provoking a debate as he exits: whether artificial intelligence should one day adjudicate civil cases itself, and where the boundary should sit. In an October interview at the Royal Courts of Justice in London, he argued that while complicated commercial disputes and sensitive family matters may need a human judge, routine categories could plausibly be handled by AI — personal injury claims, low-value professional negligence cases, and similar matters. "What most people desperately want in civil disputes is resolution," Vos said. "The question is, where is the line to be drawn?"

Current official guidance in England takes a narrower view, advising judges to restrict AI to administrative work — and Vos says he stays within that line personally. He reports using Microsoft Copilot daily to surface documents created years ago and summarize large volumes of text, claiming it finds relevant material in seconds and helps him manage 11 meetings a day, though he does not use it to draft judgments. His push for a broader role comes partly from a recent visit to a court in Shenzhen, China, where civil disputes are already being decided by judges working with heavy AI assistance — a model he says impressed him and that courts globally should consider. Vos framed the tension as one of temperament, describing lawyers as "conservative with a lowercase c": cautious adopters rather than resisters, arguing that an inaccessible-to-technology justice system cannot deliver access to justice.

The track record so far gives that caution substance. Multiple lawyers have submitted case law hallucinated by AI, and in a UK first, someone recently lost their license to practice in part over misuse of the technology. At the same time, chatbots acting as de facto legal counsel are enabling litigants to flood courts with an unprecedented volume of claims, some meritless — a cost and docket-pressure problem, not just an accuracy one. Vos, who recalled starting his career amid foolscap paper, typists, and Tipp-Ex before word processors and faxes, framed the transition as generational, saying it's time to hand leadership to new judges as the system adapts to serve a cohort raised alongside the technology.

The open questions are operational as much as ethical. Which case categories qualify as "low-value" and mechanical enough to automate without a human in the loop remains undefined, and no framework in the piece specifies appeal rights, error-handling, or liability for AI adjudication errors. For builders, the Shenzhen model signals that AI-in-the-courtroom is already deployed somewhere at scale, but the UK's trajectory points to gradual, category-by-category expansion under human oversight rather than a wholesale handover. Watch whether England's guidance evolves beyond administrative uses, and whether the sanctions and hallucination problem pushes tooling toward verifiable citations rather than fluent summaries.

],