Article

Junior employees and the path to mastery

AI now does much of the routine work juniors used to learn on. What early-career people need instead, what leading firms do, and a People team checklist.

By Jeremy Stover 4 min read

Every profession has a version of the same apprenticeship. The junior analyst reconciles the spreadsheet. The new associate drafts the first memo. The HR coordinator writes the job posting, schedules the interviews, and answers the policy questions nobody else wants. The work is routine. It’s also how people learn what good looks like.

AI is now very good at that routine work. That’s mostly good news. But it removes the rung people used to climb on, and most organizations haven’t put anything in its place.

The apprenticeship problem

Expertise used to be built in layers. You did the simple version of a task many times, someone more senior marked it up, and slowly you started to see what they saw. The draft was never the point. The markup was.

When AI produces the draft, two things go missing. The junior person skips the reps that built judgment. And the senior person, who can now do the task alone in a fraction of the time, has less reason to hand it over and mark it up.

People notice. Nearly one in five AI users say AI is making it harder for early-career workers to build skills in their field (Charter AI Pulse, 2026).

There’s a quieter shift, too. Among people who use AI daily, AI has overtaken coworkers as the first place they turn for ideas, feedback, or help thinking something through (Charter AI Pulse, 2026). Every question a junior person asks a model instead of a colleague is a conversation that didn’t happen.

What early-career people need now

Two things, and they aren’t in tension.

Durable skills. Judgment, critical thinking, clear communication, knowing which question to ask. Tools change every few months. The ability to tell whether an answer is good doesn’t expire.

A real domain foundation. You can’t judge an AI draft of a compensation analysis if you don’t understand compensation. The risk isn’t that juniors use AI. It’s that they use it to skip the part where they would have learned the subject. A fluent draft from someone who doesn’t know the field is the most dangerous kind.

The goal is people who can do the work with AI, and explain why it’s right without it.

What leading firms are doing

PwC offers one model. In late 2025 it piloted a program for about 1,300 new junior employees across tax, advisory, and assurance, adding about a week to their onboarding (Charter, 2025). The training is hands-on and built on the work juniors actually do, and it pairs AI skills with human ones like critical thinking and storytelling. Charter’s playbook with Microsoft, Leading Learning in the AI Era, covers the case in full. The firm has since made the pairing a rule: anyone teaching AI skills at PwC also teaches human skills (Charter, 2026).

One part of the program turns the usual direction around. New hires practice going back to their teams and teaching more experienced colleagues what they learned about the tools (Bloomberg Tax, 2025). The junior person knows the tool. The senior person knows the work. Each has something to teach the other.

That exchange is the part most firms miss. Training juniors on AI is common. Designing a way for their new skill to buy them time with the people who know the work is rare.

Protect the human channels

If AI becomes the first stop for help, mentoring has to become deliberate. It won’t happen by accident anymore.

  • Have juniors walk a senior colleague through how they did the work. Not the output, the process. What did you ask the model? What did you change? What did you check, and how? That brings the markup back in a new form.
  • Make the two-way teaching explicit. Give early-career people a named role teaching tools to their teams, and give senior people a named role teaching the judgment behind the work.
  • Give managers time to coach. The pressure to deliver more with AI lands on managers first. When coaching is the first thing cut, the talent pipeline goes with it.

Skills aren’t the whole story

The early-career problem isn’t only about skills. Speaking with Charter, Noam Scheiber, author of Mutiny, describes college-educated workers frustrated by jobs that didn’t match what they prepared for, and an “erosion of autonomy over the past 15 or 20 years that has led to a lot of frustration” (Charter, 2026).

AI can make that worse or better. If it leaves juniors only the scraps the model couldn’t do, it will make it worse. If it gives them a say in how their work changes, it can help. Feeling involved in how AI fits your job is the factor most closely linked to benefiting from it (Charter AI Pulse, 2026). Individual contributors report far less of that say than their managers do, and early-career people are almost all individual contributors.

What People teams can do

  • List, role by role, the tasks AI now does that used to train junior people.
  • For each one, decide how the judgment it built will be learned now. This is the list the readiness check asks about at the Application stage.
  • Pair AI training with human-skills training, every time.
  • Build “walk me through it” reviews into how junior work is checked.
  • Give early-career people a formal role teaching tools to their teams.
  • Protect managers’ time for coaching, and count it as work.
  • Give juniors a say in how AI changes their own roles.
  • Keep hiring at the entry level. The experienced people you’ll need in ten years are the ones you hire now.

The apprenticeship isn’t gone. It needs to be redesigned on purpose, and that is the People team’s work.

Next step

Want help putting this to work?

Let's talk

Want this for your team?

Turn reading into doing.

A BuildShop is where a team takes an idea like this one and builds something real with it, in two days.

Or request a lunch & learn for your team