Last year, lawyers were generally in denial about the value of AI to their profession. This year, they are piling in to using Harvey, Legora, ChatGPT 5, CoPilot, Claude and Gemini and everything else they can find for every purpose under the sun. They have realised what was always obvious, namely that AI is just a tool like so many other tech tools we use every day. True, it is an important, innovative and useful tool, but, just like a chain saw, a helicopter or a slicing machine, in the right hands it can be very useful, and in the wrong hands, it can be super-dangerous.
For the avoidance of doubt, I see no reason why AI should not be used by lawyers to draft contracts and to research legal questions. Lawyers and clients should always check what it has done carefully before using it, but that is a different issue.
In my view, the use of AI for judicial decision-making raises more complex questions than the use of AI by lawyers.
AI for judges
Judges in some jurisdictions have come late to the party. In the judiciary of England and Wales, we have done two things. First, we have put CoPilot on all our judges’ computers for them to use securely. Second, we have issued clear guidance incorporating three core rules as to the judicial use of AI: first, that you need to understand what a Large Language Model (LLM) is doing before you use it; secondly, you need to avoid putting private data into a public LLM; and third, you need to check what comes out of an LLM before you use it for any purpose at all.
It is beyond doubt that AI can help judges with all manner of daily tasks in order to save time and drudgery. The summarising capability that appears on all our screens every time we open a Word document is a great labour-saving device. An AI generated summary does not mean we can avoid reading an important document, legal precedent or judgment, but it does give us easy access to the guts of it far more quickly than was ever available before.
Should AI be used for judicial decision-making?
But the big question that I want to address here is whether AI ever can or should properly be used to generate judicial decisions or judgments for judges to approve. I should say at once that it obviously can be used for those purposes. And I shall assume for the purpose of this article that it is likely that AI will become ever more capable and will reach a stage, before too long, where it will be as reliable, or perhaps even more reliable, than human judges.
The ethical issue that I want to address, however, is whether AI should be used in judicial decision-making. The answer to that question is difficult and potentially troubling for a number of reasons.
The answer is not obvious, because it is hard to see why AI should not be used to assess, for example, personal injury damages by reference to the numerous authorities found in the textbooks. That task would take AI a couple of minutes, whilst the wait for a judicial hearing and determination might be more like two years.
Undoubtedly, once people find that machine-made judicial decisions are quick and cheap, they may find economic pressures create a slippery slope, which makes it hard to go back to human judicial decisions, even for appeals. That is why it is so important, right now, to work out whether our societies and the people we judges serve, really want judicial decisions to be made by machines.
There are three reasons why they might draw the line at machine-made judicial decisions. First, judicial decisions are the last resort for everyone in our society. If the decision is wrong, at least after an appeal, nothing can be done about it in most cases – our parliaments are unlikely to change the law to reverse run-of-the-mill AI-generated judicial decisions.
Second, machines, even those sporting the much-vaunted artificial general intelligence (AGI) when it comes, will arguably never be able completely satisfactorily to mimic a human’s emotion, idiosyncracy, empathy and insight. Even AGI cannot laugh or cry.
Third, with an AI judicial decision, you will be getting something generated from the state of intelligence at a given point in time, without the application of developing human thought. Where will that leave the law in generations to come? There is a potential problem if we, as humans, become unable to second guess or even check what the machine is suggesting or deciding. It might be very difficult for human thought processes to influence the law of the future in the way that many people might think remained appropriate.
Things we need urgently to talk about
For these reasons, I think all jurisdictions require an urgent debate as to (i) what human rights people should have in the light of ever more capable AI, and (b) what humans want, as a matter of consensus, human judges rather than machines to decide in the future.
The first question asks, in Europe at least, whether a machine-made decision can ever be properly regarded as having been made by an “independent and impartial tribunal established by law” for the purposes of article 6 of the European Convention on Human Rights. Some think so, but many more think not. If so, does that protection need amending?
The second question asks what we want machines to decide about our lives in preference to human judges, and ought we to have a choice. Ought a criminal, before being sentenced, be able to say that they want to be sentenced by a machine rather than a human or vice versa? In China, judges already routinely use AI in that process. Do we want judges to feed the facts of our cases into an AI tool, to see what an AI tool, or even a range of AI tools, think the answer should be? Or would we rather stick with the more familiar human judge, whose experiences may differ one from another, and whose idiosyncrasies we cannot predict, and only an appellate court can correct?
Conclusion
The questions I have raised are very important for our lawyers and judges of the future. I believe, however, that most present-day judges are committed to two significant objectives. First, we want to see that AI is used responsibly, effectively and safely in legal systems and processes. Second, we want to see the creation of efficient, economic and expeditious Digital Justice Systems to resolve people’s many legal disputes online and out of court and, thereby, to create greater access to justice for all.
This opinion article originally appeared in Issue 52, December 2025, of the Brief magazine, produced by the Law Society.
Issue 53 is due to be published and posted to Law Society members in late June 2026.
