For as long as I can remember, City law firms had a remarkably effective way of training their future partners. The clients paid for it.
Junior lawyers learnt what a problematic indemnity clause looked like because they had read hundreds of them, often late at night and in files they would probably rather forget. It was repetitive, not always particularly interesting, and rarely felt like training.
But it built something important – judgement.
Hidden inside the billable hours was a cost rarely acknowledged. The cost of creating the next generation of partners. Now, AI is beginning to change the economics of this.
Clients will understandably question paying for hundreds of hours of junior review when technology can complete much of the initial work faster and at lower cost.
That should be positive. Few will miss the hours spent trawling through disclosure documents. Yet removing that work also removes something valuable.
Automating the repetition risks and eliminating the training ground where judgement was developed.
Which leaves law firms with a rather more difficult question. If clients are no longer willing to pay for junior lawyers to learn, who will?
The issue is not whether the next generation of lawyers will understand AI. They almost certainly will. The question is whether using it will give them the depth of experience they will eventually need as partners.
Knowing how to ask an AI tool to identify a change of control provision is useful. Knowing why that provision matters in a difficult transaction, what might have been missed and when something simply does not look right is another matter entirely.
That judgement is difficult to teach.
Traditionally, it was accumulated almost by accident. You read the documents. You made mistakes. Someone more experienced challenged you. You saw the same problem in slightly different forms. Eventually, patterns that were initially invisible became obvious.
As AI takes on more of that repetitive work, firms may have to recreate that experience deliberately.
That could mean structured training, simulated transactions, reviewing historical deals, shadowing partners or asking junior lawyers to challenge AI-generated work, rather than simply accepting it. Firms may, in effect, have to manufacture the experience that the traditional apprenticeship model once produced naturally.
And this is where it stops being simply a training question and becomes an economic one.
Development that was once wrapped inside client work increasingly becomes a direct cost to the firm. Junior lawyers may spend more time developing skills that cannot immediately be billed, while firms continue to compete hard and pay significant salaries for the best people.
None of this is an argument against AI. Removing thousands of hours of repetitive work should make law firms more efficient and could make the early years of a legal career considerably more interesting. But it may also expose something the traditional model conveniently concealed.
The billable hours did not simply pay for legal work. They helped pay for the creation of future lawyers.
If AI weakens that subsidy, firms will have to decide how much they are prepared to invest in replacing it. And in a competitive environment, we probably know the answer.
The question for partnerships, therefore, is not simply how much they are prepared to invest in AI, but how much of today’s profit they are prepared to forgo to produce tomorrow’s partners.
Want to explore what AI means for the future of your firm?
Get in touch with myself or a member of the HaysMac team, to discuss the opportunities and challenges AI presents for professional services firms.
