Will AI deskill junior staff?
It can, if juniors use AI to skip the work that builds judgement. Deliberate practice without AI and review of their reasoning keeps skills developing.
How deskilling happens
Junior roles have always included repetitive, unglamorous work: first drafts, reconciliations, research summaries and document reviews. That work is slow, but it is how people learn what good looks like. A trainee accountant who has built many bank reconciliations by hand notices when something does not add up.
If juniors hand that work to AI from day one, they produce acceptable output faster but miss the learning underneath. The problem shows later, when they are asked to review AI output and cannot tell whether it is right, because they never built the judgement to check it. Errors that slip through can carry professional and legal consequences, explored in who is liable when AI gets it wrong. If your work carries regulatory or professional duties, take advice on your own situation.
It does not have to go that way
Used deliberately, AI can speed up learning. A junior analyst can ask a tool to explain an unfamiliar term, test their understanding, or compare their draft with an alternative. The difference lies in whether AI replaces the thinking or supports it.
Practical safeguards
Do some tasks without AI first
Identify the core tasks that build judgement in each junior role, and ask people to complete them unaided until they can do them reliably.
Finance: Prepare a set of month-end accruals manually before using AI to help with the supporting notes.
Client services: Draft renewal letters and account updates without assistance for an initial period, so the person learns tone and what clients expect.
Legal or compliance support: Summarise a contract clause by hand, then compare it with an AI summary to see what each missed.
Ask juniors to explain and defend their work
When a junior submits AI-assisted work, ask them to walk you through it. Why is this figure right? What did the tool get wrong first time? What did you change and why? If they cannot answer, the work is not ready, however polished it looks.
Review reasoning as well as results
Senior review often checks the final output and moves on. With AI involved, shift some attention to how the junior reached the answer.
Ask for the working: Request the instructions used, the sources checked and the edits made, so you can see the thought process.
Spot-check the basics: Occasionally ask a junior to redo part of a task without AI and compare the two.
Give feedback on judgement: Comment on choices, such as which source they trusted, as well as on presentation.
What this means for development plans
Treat AI competence as one part of a junior's development, alongside the underlying craft rather than instead of it. Spell out which skills you expect people to hold independently, and where they may use AI for support. The wider set of skills worth building is covered in what AI skills a team actually needs.
Be open with juniors about why you are doing this. Explaining that you want them to be able to catch AI mistakes, rather than restricting tools for the sake of it, makes the safeguards feel like investment in them rather than suspicion of them.
More questions