One of England’s most senior judges believes artificial intelligence could eventually decide certain civil disputes, offering faster and cheaper access to justice. But as AI becomes more involved in legal proceedings, questions about accuracy, accountability and public trust are growing.
Geoffrey Vos, who is preparing to retire as Master of the Rolls, England’s most senior civil justice role, wants the legal profession to have an open debate about how far AI should go in the courtroom.
Currently, judicial guidance in England recommends limiting AI use largely to administrative work. Vos, however, already relies on Microsoft Copilot to locate documents and summarize large amounts of information. After visiting a court in Shenzhen, China, where AI plays a significant role in resolving civil disputes, he came away convinced that other legal systems should consider similar approaches.
He believes AI could eventually handle relatively straightforward cases, including personal injury claims and lower-value professional negligence disputes. Complex commercial litigation and sensitive family matters, he argues, would still require human judgment.
“The priority in civil disputes is resolution,” Vos suggests. The challenge is determining which cases can safely be entrusted to machines.
AI Is Already Changing the Legal System
The growing use of AI is transforming how people approach legal problems. According to Vos, the number of cases being initiated in England’s county courts has been rising by approximately 10 percent annually, and he believes AI tools could contribute to further growth.
People increasingly turn to chatbots such as ChatGPT, Google Gemini and Claude for help understanding their legal rights and preparing court documents. These systems can generate claim forms and legal arguments, making it easier for individuals to pursue disputes without hiring a lawyer.
Vos sees this development as potentially positive. Historically, access to justice has often depended on whether someone could afford legal representation. AI could help reduce that financial barrier and give more people an opportunity to raise legitimate grievances.
However, there is still little data showing how many AI-assisted claims have sound legal foundations. Some people may be encouraged to pursue cases that are unlikely to succeed, while others may simply be finding a way to address problems they previously could not afford to take to court.
For Vos, the fact that someone used AI to prepare a claim does not automatically mean the case is frivolous. He believes the legal system should examine the substance of each dispute rather than dismissing it because of the technology used to initiate it.
Hallucinations and Fake Evidence Remain Serious Concerns
AI’s growing presence in legal proceedings has already created difficulties. Lawyers have been caught submitting fabricated case citations generated by AI, and the misuse of the technology has contributed to a UK lawyer losing their right to practise.
There are also concerns that AI-generated images, videos and audio recordings could make it harder to determine whether evidence presented in court is genuine.
Vos, however, believes the problem is not fundamentally new. Courts have always had to assess the credibility of witnesses and investigate forged documents, including fraudulent wills. In his view, deepfakes represent another development in the long-standing challenge of distinguishing truth from deception.
He argues that people who deliberately submit fabricated evidence should face the same consequences as those who forge traditional documents, including potential imprisonment for perjury.
Although he acknowledges the need for caution, Vos is optimistic that the justice system can adapt to these challenges rather than treating AI as an unavoidable disaster.
Can People Trust a Decision Made by a Machine?
One of the biggest obstacles to AI-powered justice may be public acceptance.
Human judges can explain their decisions, and even when people disagree with an outcome, they may eventually accept it because it came through a familiar legal process. Decisions made by algorithms could be more difficult to trust, particularly when people feel they have been treated unfairly.
Vos believes society must reach a consensus on when algorithmic decision-making is appropriate. People might agree to have a particular dispute resolved by AI, or governments could establish the framework through legislation.
He also questions whether the inability to see precisely how an AI model reaches its conclusions should automatically prevent its use. Although the internal workings of a model may be difficult to understand, he argues that its written reasoning can still be examined for plausibility.
That position raises a central question for the justice system: is a convincing explanation enough, or must courts also be able to establish how a machine reached its conclusion?
Where Should the Line Be Drawn?
Vos does not believe AI will eliminate human judges. He argues that people have a deep-rooted preference for being judged by another person, particularly when the stakes are high.
Family and custody disputes are among the areas where human judgment is likely to remain essential. Complex commercial cases may also continue to require human decision-makers, even as AI takes on more work involving document analysis, evidence organization and other preparatory tasks.
Less complicated civil disputes could be different. If AI can resolve them accurately, quickly and at a lower cost, Vos believes it could expand access to justice while easing pressure on the courts.
He stresses, however, that judges themselves cannot decide the boundaries alone. Politicians, lawmakers and the wider public must help determine which decisions can be automated and what safeguards should apply.
The Future of AI in Court
Vos’s position reflects a broader challenge facing the legal profession: adapting to technology without undermining the principles that give the justice system legitimacy.
AI could make legal services more affordable, speed up proceedings and help courts manage growing caseloads. But fabricated citations, questionable evidence and concerns about transparency demonstrate why its use requires careful oversight.
As Vos prepares to leave his position, he is calling for an international discussion about the future of dispute resolution. The goal, in his view, should not be to replace human judges indiscriminately, but to identify where technology can deliver fair outcomes more efficiently.
The crucial question is no longer simply whether AI can participate in the legal process. It is whether society will trust it with the power to decide who wins and who loses.