Law firms risk automating away their next generation of lawyers, Supreme Court judge warns
Lord Sales warns AI gains could come at the cost of future legal talent.

Supreme Court judge Lord Sales has warned that law firms risk undermining their future talent pipeline as AI takes over work traditionally performed by trainees and junior lawyers.
He warned that lawyers and judges could see their critical faculties atrophy as they increasingly rely on AI, losing the ability to recognise when its conclusions are wrong.
Supreme Court judge Lord Sales says the rapid adoption of AI risks leaving firms without enough experienced lawyers capable of exercising the independent judgement needed to supervise AI-generated work.
Speaking at a conference held at DAC Beachcroft’s London office this week, he described the potential erosion of legal expertise as one of the most significant risks facing the profession.
By automating the work through which junior lawyers develop their expertise, law firms could be creating a long-term talent problem, he warned.
While automating research, drafting and document review offers immediate productivity gains, Sales argued that firms risk overlooking the role these activities play in developing professional judgement.
"The less efficient human process has been producing two different things at once: today's legal work and tomorrow's lawyer,” he said.
An erosion of expertise
Sales questioned what happens to the profession when junior lawyers no longer undertake the work that has traditionally equipped them to become senior practitioners.
"Where does the experienced senior lawyer come from? Who will be the lawyer with the critical faculties required to examine the product which AI generates?"
He warned that firms could improve their short-term economics while undermining their ability to develop senior talent over the next decade.
The 'human in the loop' can gradually become the human rubber stamp.
"A law firm which eliminates expensive junior work may improve its productivity this year while quietly failing to generate and develop its own human capital for ten years to come,” he said.
The risk extends beyond law firms to the judiciary, where Sales warned that critical faculties could atrophy as judges increasingly delegate research, analysis and drafting to AI.
When AI can produce polished legal work in seconds, there's a risk that lawyers and judges will simply accept its conclusions rather than scrutinise them independently.
"The 'human in the loop' can gradually become the human rubber stamp," he said.
Training the next gen
Sales suggested some lawyers may need protected periods of training without AI, during which they are required to research, analyse and draft independently.
Law schools are already experimenting with ways to preserve those skills. The University of Chicago Law School has banned laptops and phones from core first-year classes, encouraging students to develop their reasoning skills without digital assistance.
UCL's law faculty requires at least half of the assessment in each module to be conducted under conditions that prevent AI use, including in-person written exams and oral assessments.
For law firms, Sales suggested they may need to invest more heavily in a smaller group of lawyers capable of exercising the sophisticated judgement required to oversee AI systems.
"The number of lawyers required to produce the same quantity of legal work may fall sharply," Sales said.
The result could be a profession employing fewer lawyers overall but spending considerably more on developing them, he says.
AI in the courtroom
Sales also appeared open to the possibility of AI eventually deciding certain categories of legal disputes, particularly lower-value commercial cases where the cost and delay of traditional litigation can be disproportionate.
He argued that the European Convention on Human Rights does not necessarily require every tribunal to consist of a human being, suggesting that AI could make initial decisions subject to human judicial review.
While stressing that more serious cases would require a different approach, Sales suggested the future may involve dividing judicial functions between humans and AI rather than insisting that all decisions remain exclusively human.
His broader concern, however, is whether lawyers and judges will retain the expertise needed to scrutinise AI-generated decisions.
"The danger is not simply that the machine gets the answer wrong. The greater long-term danger is that human beings gradually lose the ability to know whether it has got the answer wrong."
Join 10,000+ City law professionals who start their day with our newsletter.
The essential read for commercially aware lawyers.




