Junior work has often combined low-cost production with apprenticeship. New hires research examples, prepare first drafts, run routine tests, reconcile records, and observe how experienced colleagues correct them. These tasks are attractive automation targets precisely because they are structured and reviewable.
The productivity bargain is becoming unstable
A senior employee using AI may produce more than a traditional senior-and-junior pair on routine work. Hiring fewer juniors looks efficient in the short term. But an organisation that removes entry-level practice without replacing it eventually lacks people who understand the domain well enough to supervise systems, handle exceptions, or become senior.
- First drafts become machine-generated, reducing practice in forming an argument from a blank page.
- Routine debugging and test execution become automated, reducing exposure to common failure patterns.
- Research summaries arrive instantly, making source judgement more important but less frequently practised.
- Administrative coordination shrinks, while stakeholder communication begins earlier in a career.
- Output expectations rise because every entrant has access to capable tools.
Experience still has to come from somewhere
People become experienced by making decisions, seeing consequences, receiving feedback, and building a library of patterns. Reading an AI explanation is not equivalent to owning a live incident or observing a customer. Employers need deliberate practice environments: simulations, shadowing, scoped ownership, reviewed agent supervision, and rotations through real workflows.
A new junior role
The junior worker becomes less of a low-cost producer and more of a supervised operator. They may prepare context for agents, inspect outputs, document failures, interview users, run controlled experiments, and own small outcomes. The threshold for useful contribution rises, but so can the rate of learning when feedback is designed into the work.
Companies that treat junior hiring only as a cost line may create their own future shortage of judgement. Apprenticeship should be redesigned, not silently deleted.
Our editorial estimate is that entry-level vacancies will become fewer and more demanding in highly digitised functions, while portfolios, work simulations, and demonstrated tool supervision will carry more weight than credentials alone.