🌱 What Happens When Your Best Junior Person Stops Getting the Junior Work

There's a structural shift happening quietly across a lot of professional fields that doesn't get discussed nearly as much as it should. The traditional path for developing expertise, starting with the simpler, more repetitive tasks in a field and gradually working up to more complex judgment-intensive work, depended on those simpler tasks existing in meaningful volume. AI is absorbing a significant share of exactly that entry-level work, and almost nobody has fully worked out what replaces the learning path that used to run through it.

This isn't just a hiring or training logistics problem, though it shows up there too. It's a pipeline problem with a genuine long-term time cost, because the people who would have become tomorrow's experienced judgment-holders, the senior professionals whose accumulated pattern recognition makes them fast and reliable at complex decisions, aren't getting the repetitions that used to build that judgment in the first place.

------------- Context -------------

Historically, junior professionals in most knowledge fields learned their craft substantially through volume: doing the simpler research tasks, drafting the more formulaic documents, handling the routine client interactions, before graduating to more complex and judgment-intensive work. This wasn't an inefficient use of junior time. It was, functionally, the training mechanism. The repetition built pattern recognition. Making mistakes on lower-stakes work and getting corrected built calibration. The accumulated volume of these experiences is what eventually produced professionals capable of handling genuinely complex situations with good judgment.

AI has compressed the value of having a junior person do this work directly, because AI can often produce the initial draft or analysis faster and at comparable quality to what a junior professional would have produced after significant time investment. The economic logic for many firms increasingly favors using AI for this tier of work rather than assigning it to junior staff, which is individually rational for any given task but collectively removes the volume of repetition that used to build junior expertise over time.

The concerning part isn't that junior professionals are doing less low-value work. It's that the specific mechanism through which professional judgment used to get built, doing enough repetitions of progressively complex tasks to develop genuine pattern recognition, is being quietly hollowed out without an intentional replacement being built in its place.

------------- The Gap That's Forming and Who Notices It First -------------

A partner at a mid-sized professional services firm described watching this gap emerge directly over several years. Junior associates at her firm were, on paper, more productive than junior associates had been five years earlier, largely because AI tools handled much of the work that used to occupy their early years. But when those same associates reached the point where they should have been developing into professionals capable of independent complex judgment, several years into their careers, she noticed their judgment development lagging behind where associates at the same career stage would have been under the older training model.

The associates weren't less capable individuals. They simply hadn't accumulated the same volume of hands-on repetition that used to build the pattern recognition underlying strong professional judgment, because AI had been doing much of the repetitive work that used to provide those reps.

Her firm's response was deliberate: rather than assuming the traditional training path would simply reconstitute itself naturally, they built an explicit alternative. Junior associates were assigned a curated set of tasks specifically chosen for their learning value, sometimes tasks that AI could technically have handled faster, deliberately preserved as hands-on training exercises rather than automated away entirely. This meant accepting a real efficiency cost in the short term, doing certain tasks the slower, more manual way specifically for the developmental value, in exchange for preserving the pipeline that would eventually produce experienced senior judgment.

------------- Building an Intentional Replacement for the Old Pipeline -------------

The old training path worked partly because it was implicit. Junior professionals learned by doing the work that needed doing anyway, and the learning value was a byproduct rather than a deliberate design. As AI absorbs more of that work, the learning value has to be deliberately preserved rather than assumed to happen automatically, because it's no longer a natural byproduct of task assignment.

This requires firms and mentors to think explicitly about which tasks carry the most developmental value and to protect those specifically, even when AI could technically handle them faster, rather than defaulting every task to the fastest available option purely on efficiency grounds.

------------- Practical Moves -------------

First, if you manage or mentor junior professionals, identify explicitly which tasks in your field carry the most developmental value for building judgment, distinct from which tasks are simply the fastest to hand to AI. These lists often overlap significantly but aren't identical.

Second, deliberately preserve a portion of high-learning-value work for junior professionals to do hands-on, even where AI could technically produce comparable output faster. Frame this explicitly as a training investment rather than an inefficiency.

Third, build more deliberate feedback and mentorship structures around the work junior professionals do handle directly, since the learning value of any task is significantly amplified by good feedback, and that feedback loop needs more intentional design as the volume of natural repetition decreases.

Fourth, if you're early in your own career, seek out the hands-on repetition deliberately rather than assuming it will simply come your way through normal task assignment. Ask for the harder, more manual version of tasks specifically for the learning value, even when a faster AI-assisted path is available.

Fifth, revisit training and development structures periodically as AI capability continues to expand into new categories of work. The specific tasks worth preserving for hands-on learning will likely need to shift over time as the boundary of what AI handles well continues to move.

------------- Reflection -------------

This is a genuinely difficult problem because the individually rational choice, letting AI handle work it can do faster and more efficiently, aggregates into a collectively concerning outcome if nobody intervenes deliberately to preserve the learning pipeline that used to run through exactly that category of work.

The organizations and mentors managing this well aren't refusing to use AI for entry-level work. They're being deliberate about which specific pieces of that work carry outsized developmental value and protecting those pieces intentionally, accepting a real short-term efficiency cost in exchange for preserving the long-term pipeline that produces genuinely capable senior judgment.

If your field or organization is letting AI absorb entry-level work by default, has anyone deliberately thought through what happens to the pipeline that used to build senior judgment through that work, or is it simply being assumed to work itself out?

6 comments

🌱 What Happens When Your Best Junior Person Stops Getting the Junior Work