The future is here. It may not need you.
The legal industry is pretending AI will liberate junior lawyers. It is more likely to reveal how many of them were only ever there to be converted into time.
Every few weeks, another legal AI product arrives dressed as the future.
There is a launch post. There is a panel discussion. There is a photograph of someone from innovation standing near someone from management, both looking as though they have personally negotiated with the machine and obtained favourable terms.
The tone is always the same. Excited, careful, slightly dead behind the eyes.
The message is that the profession is entering a new age. Lawyers will be empowered. Workflows will be transformed. Clients will receive better service. Junior lawyers, we are told, will be freed from endless review work and allowed to do more interesting things.
It sounds perfect.
It also sounds like something written by a committee that has not yet spoken to finance.
Because the interesting question is not whether these tools will be used. They will be used. Some of them are already very good. Some are not. That is not really the point. The point is what happens when a profession built on time starts adopting tools designed to reduce it.
The legal industry has not yet had an honest conversation about that. It is circling one, nervously, with a branded tote bag.
The first problem is charging.
A significant amount of junior legal work has historically involved taking large quantities of tedious material and converting it into something a client can digest. Lease reviews. Title reports. Replies to enquiries. Due diligence summaries. First-pass issue lists. The sort of work that eats days, evenings, weekends and, if you are lucky, only the more cheerful parts of your personality.
For years, this work served several purposes at once. It trained juniors, in theory – supposedly by osmosis. It created useful output, sometimes. Most importantly, it generated time.
Lots of time.
That time could then be recorded, reviewed, written down a little, written off a little, and ultimately converted into fees. The model was crude, but it worked. The junior lawyer sat inside it like a hamster with a law degree.
AI drives a juggernaut through this arrangement.
A decent legal AI tool can now produce, in seconds or minutes, a first-pass review that would once have occupied a junior lawyer for half a day, a full day, or the emotionally significant part of a weekend. It will not be perfect. It will still need checking. It will still need judgment. It will still need a human being to understand context, risk, client sensitivity and the thousand small ways in which legal documents can lie while remaining technically accurate.
But checking is not the same as doing.
And a sense-check is not six hours of chargeable grind.
This is where the industry’s public story begins to fray. Firms can say, truthfully enough, that lawyers will remain involved. They can say that AI output must be supervised. They can say that professional indemnity insurers, regulators and clients will still want a qualified human somewhere in the chain, ideally with a name, an email address and a measurable level of anxiety.
All of that may be true.
But none of it answers the commercial question. If the work takes less time, what exactly is being billed?
The obvious answer is that firms will try to charge for the tool. There will be technology fees, AI premiums, matter innovation charges, or some other phrase with the smooth, narcotic quality of pricing language. This may be defensible up to a point. These tools cost money. Implementation costs money. Training, security, procurement and compliance all cost money.
But once that cost is spread across clients and matters, the surcharge is unlikely to come anywhere near the hours-based fee income it is quietly replacing.
The difficulty, of course, is that honesty has never been the billable hour’s most natural habitat.
The second problem is worse.
Junior lawyers.
There are certain things the profession is currently unwilling to say plainly, so it says other things instead. It says AI will “free up” junior lawyers. It says they will move into higher-value work. It says they will become more strategic, more client-facing, more commercially aware.
This is the sort of sentence that sounds humane until you remember how law firms make money.
Inside a law firm, freed time is not a moral victory. It is unfilled and unbilled capacity.
If a trainee no longer needs to spend two days reviewing leases, that does not automatically make the trainee a strategic adviser. It makes the trainee someone with two days missing from their time sheet. Perhaps they can be trained. Perhaps they can help with business development. Perhaps they can attend a client meeting and nod with the correct level of solemnity.
But firms already have people who do marketing. They already have people who produce business development plans. They already have senior lawyers who own client relationships. And those people do not usually look at a newly qualified solicitor and think: finally, the missing piece in our go-to-market strategy.
The uncomfortable truth is simpler.
A lot of junior work exists because it is necessary, repeatable and billable. If it becomes less necessary, less time-consuming and less billable, firms will need fewer juniors.
Not no juniors. Fewer.
The legal profession has always dressed its labour pyramid as a training system. Sometimes it is. There is value in reading the documents. There is value in seeing patterns, spotting oddities, learning what matters and what merely looks important. A junior lawyer who has never done the boring work may struggle to develop the judgment required to supervise the machine that now does it.
That is a real problem.
It is also a problem firms will need to solve deliberately, which is another way of saying expensively.
The old model trained people accidentally. It took intelligent graduates, buried them in documents, applied pressure, removed sleep, added occasional terror and called the resulting competence “development”. It was inefficient, frequently miserable and morally underwhelming. But it produced lawyers who had seen enough masonry to understand the building.
AI threatens that pipeline too.
If the machine does the first pass, where does the junior learn what the first pass is? If the machine flags the issue, how does the junior develop the instinct to know whether the flag matters? If the machine drafts the report, who teaches the junior the difference between a point, a risk, a caveat and a professional embarrassment?
The answer should be structured training.
The likely answer is a pre-recorded webinar, a competency framework and a Teams session delivered with the bright, procedurally focussed optimism of people nowhere near the actual work.
Clients are not standing still either. This is the part firms should be more afraid of than they appear to be. The tools are not only being bought by law firms. They are being bought by clients. In-house teams with decent legal knowledge and access to good AI tools will not always need to send everything out for a junior lawyer to summarise at external rates.
Why would they?
If a client can run the first pass itself, identify the obvious issues, triage the documents and then send the genuinely difficult points to external counsel, the value proposition changes completely.
The law firm is no longer being paid to discover the terrain. It is being paid to advise on the dangerous parts of the map.
That may be better for clients. It may even be better for senior lawyers who enjoy actual judgment work rather than managing armies of exhausted twenty-somethings through disclosure-adjacent purgatory.
But it is not obviously better for the traditional junior pyramid.
The industry will resist saying this for as long as possible. It will call the change transformation. It will call it augmentation. It will call it a shift in talent strategy. If enough consultants are involved, it may call it a journey.
But the underlying arithmetic is not complicated.
If fewer human hours are needed to produce the same output, and if junior lawyers exist largely because their hours can be sold, then fewer junior lawyers will be needed.
That is not cynicism. It is the business model caught off-guard and without its make-up.
There will be exceptions. There always are. Some firms will invest seriously in training. Some juniors will become better faster because they learn to use AI intelligently. Some work will expand because lower costs make certain projects more viable. Some lawyers will move earlier into supervision, analysis, verification and strategy.
These would be excellent results in themselves.
But the broad direction is difficult to avoid. The bottom of the pyramid is going to narrow. The profession will discover, perhaps with visible sadness and invisible relief, that it does not require quite so many young people to spend their twenties converting anxiety into units.
There is a darker kindness in that.
For all the panic about AI taking legal jobs, one possible outcome is that fewer bright, ambitious people are fed into a machine that sells prestige at the door and quietly shreds their mental wellbeing in return. Fewer people may be told that misery is training. Fewer may mistake exhaustion for importance. Fewer may discover, too late, that the “profession” they joined is often an industry wearing a fake Rolex.
So perhaps the advice to anyone thinking of entering the legal circus is not to learn prompt engineering, optimise your LinkedIn presence and prepare for the future.
Perhaps the advice is simpler.
Look carefully at the tent before you buy a ticket.
The circus has found some cheaper acrobats.