LO
Judge / magistrate — tasks, one by one
The unit of analysis is the task, not the job title. Each one below carries its direction, whether the judgement rests on evidence or on platform inference, the reasoning, and what it does not establish.
Every task on this page#
Hearing cases and running proceedings
Still human-led✓ Evidence-backedPresiding in court, hearing evidence and argument, and keeping proceedings fair and orderly.
The person in the chair is fixed by the institution, and the guidance keeps the evidence in the judge's own hands. England and Wales tell judges they must always read the underlying documents and that AI cannot stand in for direct judicial engagement with evidence. The US statistics bureau projects judge and hearing officer employment to grow 2 percent from 2025 to 2035, limited by public budgets, without mentioning AI.
Guidance from one judiciary and a projection for one country; no record tests AI conducting hearings.
Researching the law
Being augmented✓ Evidence-backedFinding and reading the statutes, precedents and authorities a case turns on.
Research is where the tools are weakest and the guidance most cautious. A Stanford study found leading AI legal research tools hallucinated between 17% and 33% of the time, and an earlier study found general models wrong about federal cases between 58% and 88% of the time. England and Wales call AI a poor way of conducting research to find new information you cannot verify, and New Zealand says using chatbots for legal analysis is not recommended.
Tests of research tools, not of judges' work, with models that have since changed; guidance says what is advised, not what judges do.
Drafting judgments and orders
Being augmented✓ Evidence-backedWriting reasons, orders and summaries that the parties and appeal courts will read.
Drafting help is being provided, with the judge answerable for every word. England and Wales made Microsoft's Copilot Chat available on judicial office holders' devices and say judges are personally responsible for material produced in their name. China's Supreme People's Court supports AI for assisted generation and review of legal documents.
Tools made available and policies supporting them; no record measures how many judgments are drafted with AI.
Deciding the case
Still human-led✓ Evidence-backedDeciding who wins, what the law requires and what sentence to impose.
Every judiciary on this page draws the line here. China's Supreme People's Court says that however far technology develops, AI must not take the place of a judge's ruling and its output is only a reference. US federal courts have been cautioned not to delegate core judicial functions to AI, including decision-making. The EU classifies AI that assists judges in researching and applying the law as high-risk.
Rules and guidance, not evidence of what happens in individual cases; the EU obligations for such systems apply only from December 2027.
Case management and administration
Being augmented✓ Evidence-backedScheduling, triaging cases, managing files and the records of proceedings.
Administration is where courts most openly invite automation. China's Supreme People's Court supports automatic classification of files, case triage between simple and complex matters, and automatic generation of hearing records. California requires every court that does not prohibit generative AI to adopt a use policy for staff and judicial officers.
Policy support and a rule requiring policies; neither shows how much administration is automated.