Master of the Rolls: people will probably accept machine-decided small claims

Cost and delay may do more to normalise AI in the courts than any policy decision. Sir Geoffrey Vos, the Master of the Rolls, has suggested people will probably accept machine-enabled resolution of small disputes at first instance simply because it is quicker and cheaper.

He was speaking on 21 August 2026 at the John Lehane memorial lecture, delivered to the Supreme Court of New South Wales, where he argued the profession needs an urgent debate about where the boundary between machine-assisted and machine-made decisions should sit.

Pressure to let software advise the bench

Sir Geoffrey called it entirely predictable that court systems will come under pressure to let machines at least advise judges. Judicial guidance in many jurisdictions already insists the judge remains personally accountable for the decision, but he cautioned that some judges may reach for large language models outside those rules anyway.

Using AI to advise judges on case outcomes, he said, is "likely to be some way ahead" in common law systems. He also expected an uneven curve: "Human caution and circumspection is likely to mean that changes will be slow at first, but when they come, they may accelerate."

The arbitration market has already moved

He pointed to something already live rather than hypothetical. In December the American Arbitration Association's International Centre for Dispute Resolution launched an AI arbitrator for two-party, documents-only construction disputes below a set value.

A human stays in the loop by design: the system assesses the claims and drafts the award, and a human arbitrator checks the reasoning and the outcome to protect trust, transparency and due process.

"There is seemingly no reason of principle to prevent competent commercial parties from agreeing to such a dispute resolution method," Sir Geoffrey said, adding that the market will soon show how popular the format proves.

Where the line is hard to draw

Distinguishing a machine-assisted decision from a machine-made one is, on his account, genuinely difficult. The technology is improving faster than any judge can realistically verify its output.

As confidence in tools such as Harvey and Legora grows, he asked how much judicial attention will actually go into checking what the machine suggests. Hand a judge 1,000 draft decisions and, he warned, "economic and time pressures will inevitably lead to judicial approval of the machine's recommendations."

Legal research assistance raises no obvious rule-of-law difficulty, he accepted. The complication is that the same systems can answer the ultimate question in a dispute, not merely supply background law.

Principles needed before practice hardens

Sir Geoffrey praised the Standing International Forum of Commercial Courts for what he called ground-breaking work on principles that could command broad agreement about acceptable use of AI in commercial and arbitration decision-making.

He wants an informed debate about drawing those lines so that the right to an independent, impartial human decision survives, other than where parties have given free and fully informed consent to a machine deciding.

Public trust underpins the whole system, he added, and it would suffer if legal and dispute resolution providers looked protectionist or self-interested in the face of increasingly capable AI that can offer reasonable, if unorthodox, help to lawyers and judges.

Sir Geoffrey retires this autumn, having held the office since 2021.