ChamberOne

AI and the apprenticeship model of legal training

By the ChamberOne team 4 minute read

Generations of lawyers learned the craft by doing the dull jobs: reading bundles, summarising documents, drafting first versions, checking cross-references. Those tasks were boring, but they taught how documents work, where problems hide and what a senior lawyer notices. If AI now does a share of that work, the training that came with it is at risk. This article is for training principals and heads of learning who want to keep the benefits of apprenticeship while the grunt work changes.

Key points

  • Routine work taught judgment as a side effect; if it goes, replace the teaching on purpose.
  • Plan shadowing, with briefing and debriefing, rather than hoping it happens.
  • Alternate drafting from scratch with fixing machine drafts so that juniors build both writing and review skills.
  • Give prompt, reasoned feedback and measure skills, not hours.

What the grunt work actually taught

It is worth being precise about what is at stake. Reading a bundle taught juniors how a dispute develops over time and which documents matter. Summarising taught them to separate the important from the incidental. First drafts taught them the shape of a document and the reasons for its clauses. Checking taught them to notice small errors that change meaning.

None of this was in the job description. It was a side effect of doing the work. If the work goes, the side effect goes with it unless someone replaces it on purpose.

The risk: skilled reviewers with no apprenticeship

The worry is a gap in the middle. Firms will need people who can review AI output with judgment. That judgment comes from having done the underlying work. A junior who has never drafted an agreement may struggle to see what an AI-drafted one is missing. A junior who has never summarised a bundle may not sense what a summary left out.

So the shift in tasks creates a training problem exactly where the profession most needs good judgment. The firms that notice this early can design for it.

Deliberate shadowing

Juniors used to pick up knowledge by being near seniors. In a hybrid working world, and with less routine work to share, that happens less. Make it planned.

Even small things help: leaving a door open on a call, copying a junior on an email chain with a note on why a point was handled that way, or walking through a marked-up draft together for ten minutes.

  • Invite juniors to calls, negotiations and client meetings, and brief them before and afterwards.
  • Ask the junior to take a note and compare it with the senior's.
  • Have the senior explain choices aloud: why this clause, why this tone, why this authority.
  • Pair juniors with different seniors across a year so they see different styles.
  • Set aside time in the diary, because unscheduled mentoring tends to vanish under pressure.

Redrafting as a teaching method

One of the best uses of AI in training is as a source of imperfect drafts for juniors to fix. Give a junior an AI-produced clause, letter or summary, together with the source material, and ask them to improve it and explain each change. That exercise builds the skill of review, which is the skill the firm will need.

Alternate between the two directions. Sometimes the junior drafts from scratch, then compares their version with a tool's. Sometimes they edit a machine draft. The first builds the underlying skill, the second builds judgment about output. Doing only the second risks producing editors who cannot write.

Draft from scratch
  • The junior writes first
  • Then compares with a tool's version
  • Builds the underlying skill
Edit a machine draft
  • The junior improves an AI draft and explains each change
  • Builds judgment about output
  • Used alone, it risks editors who cannot write
Two ways to use drafts in training

Feedback that is specific and prompt

Apprenticeship worked partly because correction was frequent: a senior crossed out a paragraph and said why. As work moves faster, feedback can become thinner. Make it a rule that juniors receive comments on their work within a short, agreed time, and that comments explain the reason, not just the change.

Keep a record of recurring points. If three juniors in a row miss the same type of error in AI output, that is a training topic for the whole group.

Short, frequent conversations work better than an annual review. Ten minutes at the end of a matter, covering what went well, what to do differently and what the junior would like to try next, builds both skill and trust.

Rotations, simulations and supervised live work

Without routine volume, juniors may see fewer matters through to the end. Use rotations that let them follow a file from instruction to close. Use simulations: a fictional matter with realistic documents, run in a few sessions, where juniors take roles, make decisions and receive feedback.

Give juniors real responsibility early, with supervision. Drafting a client letter that a partner will review teaches more than a month of summaries. The cost is partner time, and it should be planned and budgeted as the cost of building the next generation.

Measuring whether it works

Check progress against skills, not hours. Can the junior spot a missing clause in a draft? Explain an authority in their own words? Take a client call and write a clear note? Review each junior at set points against a short list of skills, and discuss gaps with them.

Ask juniors what is helping. They often know before their seniors do which exercises build confidence and which are box-ticking.

Be wary of shortcuts such as counting the number of AI-assisted tasks. Volume of use says little about learning. A junior who uses a tool for everything and cannot explain the output is not learning, whatever the figures show.

Costs, fairness and what to protect

Training takes time that is hard to bill. If juniors are not generating hours from routine work, the firm must decide how to fund their development. Some firms treat it as an investment, some adjust expectations on utilisation, some move juniors earlier to work that clients value. Be open about the choice and check that juniors are not penalised for time spent learning.

Check also the training rules of your regulator or bar, since qualification requirements often prescribe experience. Keep the aim in view: people who know the law, can write, can judge and can be trusted with a client.

This article is general information about the law, legal practice and technology, current when it was published. It is not legal advice. Laws and guidance change: check the sources linked above and take advice on your own circumstances from a lawyer qualified where you are.

Read next

AI and the apprenticeship model of legal training