AI can make juniors faster. But are they learning?

AI can be incredibly empowering for junior employees.

• It can help them structure a first draft.
• It can explain unfamiliar concepts.
• It can summarise documents.
• It can prepare meeting notes.
• It can rewrite a difficult email.
• It can turn scattered thoughts into something useful.
• It can help someone walk into a conversation better prepared.

That matters. A junior person who might previously have waited for help can now get started. A new employee who might have been nervous to ask a basic question can now build a first view. Someone who struggles with writing can now express better thinking more clearly.

That is not small. AI can boost confidence. It can create momentum. It can reduce friction. It can help people contribute sooner.

So this is not an argument against AI. It is an argument for using it properly. Because the uncomfortable question is this: AI can make juniors faster. But are they learning?

AI can help junior people arrive at the answer. But development comes from understanding the path.

The productivity story is real

The evidence that AI can improve productivity in some tasks is strong enough to take seriously.

A study published in the Quarterly Journal of Economics examined more than 5,000 customer-support agents using generative AI assistance. It found that access to AI increased productivity by 15% on average, measured by issues resolved per hour. The gains were largest for less-experienced and lower-skilled workers.[1]

That is a powerful finding for junior development. It suggests AI can help less-experienced people perform closer to stronger performers in certain structured environments. It can give them guidance, language, response patterns and support at the point of work.

The study also found evidence that AI assistance improved English fluency and helped worker learning, particularly among international agents.[1]

That is the positive version of the story. AI can raise the floor. It can help people who are newer, less confident or less experienced participate sooner and perform better. But that is not the end of the question. Because speed and development are not the same thing.

Speed is not the same as learning

A junior analyst can produce a summary faster. But did they understand the issue? A graduate can draft a client email faster. But did they understand the relationship context? A new manager can ask AI for a performance conversation script. But did they learn how to read the person in front of them? A young consultant can produce a strategy outline. But did they understand the trade-offs, risks and assumptions?

AI can help people arrive at the answer. But development comes from understanding the path.

That distinction matters. If a junior person uses AI to accelerate their thinking, challenge their assumptions and improve their questions, the tool can be transformative. If they use AI to skip the thinking, the output may improve while the capability does not.

That is the risk. Cleaner work. Faster work. More confident work. But not deeper learning.

Some struggle was useful

A lot of workplace learning used to happen through friction:

• The uncomfortable first draft.
• The messy attempt.
• The senior person asking, "Why did you write it that way?"
• The redline that showed what good looked like.
• The meeting where someone watched a leader handle pressure.
• The painful moment where a weak recommendation had to be defended.

Nobody loves those moments. But they teach. They teach structure. They teach judgement. They teach confidence. They teach what matters. They teach what not to say. They teach how a room reacts. They teach why context changes the answer.

If AI removes every rough edge too early, we need to ask what learning disappears with it. Not all friction is waste. Some friction is training.

The apprenticeship model is changing

Most organisations still rely on some form of apprenticeship. Not always formally. But practically. Junior people learn by being close to the work:

• They draft.
• They get feedback.
• They join calls.
• They watch how decisions are made.
• They ask questions.
• They make small mistakes.
• They see how experienced people think.

That is how judgement gets transferred. Not through a policy document. Through exposure.

AI changes that model. It gives juniors a private coach, editor and explainer. That is powerful. But it can also quietly reduce the number of human teaching moments. The junior no longer asks the "basic" question. The manager no longer sees the rough first draft. The senior leader only sees the polished version. The feedback loop gets shorter, but sometimes thinner.

That is the part leaders should watch. Because if managers only ever see polished AI-assisted outputs, they may lose visibility of how juniors are actually thinking.

The danger of polished misunderstanding

AI can make weak thinking look better than it is. That is useful in some contexts. It is dangerous in others.

A junior person may submit something that reads well, but does not hold up. The tone is polished. The structure is clear. The grammar is clean. The confidence is high. But the logic is shallow. The assumptions are untested. The commercial context is missing. The recommendation is generic. The risk is not understood.

That is harder for managers to spot than the old messy first draft. The old rough draft exposed the thinking. The new polished draft can hide it.

So the leadership question changes. Not only "Is this good?" but "Can you explain why this is good?" That question matters. It separates output from understanding.

AI can compress experience. It cannot replace reflection.

One of the most exciting possibilities is that AI can compress some learning cycles. A junior employee can ask for examples. They can compare approaches. They can request counterarguments. They can ask what a sceptical manager would challenge. They can test whether their logic is coherent. They can ask AI to explain a concept in several levels of complexity.

That is powerful. In the past, some of that required access to a senior person with time, patience and context. AI can now provide a first layer of support quickly.

But reflection still matters. Learning does not happen only because someone receives an answer. It happens when they compare the answer to reality:

• What was useful?
• What was wrong?
• What was generic?
• What did the customer actually say?
• What did the manager challenge?
• What did the meeting reveal?
• What would I do differently next time?

That reflection is where capability grows. Without it, AI can become a shortcut that produces faster output but weaker ownership.

What juniors still need to learn

AI can help with many tasks. But it cannot fully replace the hard parts of professional development.

Juniors still need to learn how to frame a problem — understanding what question is actually being asked, not simply answering the words in the brief.

They need to learn how to ask better questions — moving beyond "what should I write?" towards "what am I trying to achieve, what is uncertain, and what evidence would change my view?"

They need to learn how to understand context — knowing that the same answer can be right in one customer relationship and wrong in another.

They need to learn how to build an argument — connecting evidence, assumptions, risks and recommendations in a way that can be defended.

They need to learn how to handle feedback — not treating a redline as criticism, but as insight into how stronger thinking looks.

They need to learn how to read a room — noticing power, pressure, timing, emotion and politics.

They need to learn how to separate confidence from competence. That may be one of the most important AI-era skills of all. Because AI is fluent. It can sound sure. Juniors need to learn that confidence in the answer is not the same as quality of the answer.

What leaders should do differently

The answer is not to ban AI for junior people. That would be the wrong lesson. It would also be unrealistic.

AI is already moving into daily work. Microsoft and LinkedIn's 2024 Work Trend Index reported that 75% of global knowledge workers were using AI at work, and 78% of AI users were bringing their own tools to work.[2]

So the question is not whether juniors will use AI. They will. The better question is how organisations make that use developmental.

Leaders should ask juniors to show the process, not only the output:

• What did you ask AI?
• What did it give you?
• What did you reject?
• What did you change?
• What did you verify?
• What assumption did you challenge?
• What would you do differently next time?

That turns AI from a shortcut into a learning tool. It also gives managers a better view of the thinking behind the work. Because that is where development happens. Not in the final paragraph. In the choices made before it.

Make the messy middle visible again

One of the risks of AI is that it hides the messy middle. The final output looks better. But the learning often happens before the final output: in the confusion, in the false starts, in the weak first attempt, in the wrong assumption, in the feedback, in the revision, in the moment where someone realises why the first answer was not enough.

If AI removes that visibility, managers need to recreate it deliberately. This can be done simply. Ask for the first AI output and the final version. Ask what changed. Ask what was removed. Ask what was verified. Ask where the person disagreed with AI. Ask what they learned from the difference.

That is a much better development conversation than simply reviewing the finished document. It also teaches juniors that AI is not the authority. It is a starting point.

Teach AI use as a thinking discipline

Most AI training focuses on tools. How to prompt. How to summarise. How to write. How to automate. How to create slides. How to produce more output. That is useful, but incomplete.

For junior people, AI training should also teach thinking discipline:

• How to question the answer.
• How to check claims.
• How to spot generic output.
• How to identify missing context.
• How to compare options.
• How to ask for counterarguments.
• How to separate facts from interpretation.
• How to know when a human conversation is needed.

Those are the skills that turn AI from a productivity tool into a development tool. Otherwise, organisations may produce juniors who are faster at generating work, but weaker at defending it.

The role of senior people changes

AI does not remove the role of senior people in developing juniors. It changes it.

Senior people may spend less time fixing grammar and formatting. Good. They may spend less time helping someone start from a blank page. Good. They may spend less time explaining the same basic concept repeatedly. Also useful.

But that time should not disappear into more meetings and more output. It should move into coaching judgement:

• Why did you choose this recommendation?
• What is the trade-off?
• What would the customer push back on?
• What did you verify?
• What did you ignore?
• What would make you change your mind?
• What is the real issue underneath this task?

That is where senior input becomes even more valuable. AI can help juniors get to a better starting point. Senior people must help them understand what better actually means.

A better standard for junior work

Maybe the new standard for junior work should not be "Did you use AI?" or "Did you avoid AI?" Maybe the better standard is "Did AI help you think better?"

That is the shift. AI should not only make juniors faster. It should make them more curious, more prepared, more reflective, more confident and more willing to test their own thinking.

If it does that, it is transformative. If it only helps them produce cleaner work without deeper understanding, we may be building speed without capability.

What juniors should do

This is not only on leaders. Junior people should also be intentional.

Use AI to prepare, but do not let it replace preparation. Use it to explain, but then test whether you can explain the concept yourself. Use it to draft, but know why the final version is stronger. Use it to generate options, but decide which option you can defend. Use it to identify risks, but check whether those risks matter in your context. Use it to build confidence, but do not confuse confidence with mastery.

A good personal rule is simple:

Never submit AI-assisted work you cannot explain.

If you cannot explain the logic, the assumptions and the trade-offs, you have not yet learned enough from the tool.

The opportunity

Used well, AI can be an exceptional development accelerator. It can help junior people learn faster. It can expose them to better examples. It can help them practise. It can lower the fear of asking basic questions. It can help them prepare before engaging senior people. It can give them language for ideas they already had. It can help them reflect after feedback.

That is exciting. The opportunity is not to protect old friction for its own sake. The opportunity is to keep the useful friction: the friction that teaches, the friction that tests thinking, the friction that builds judgement, the friction that forces ownership.

That is the balance.

The question

So here is the question. Where should junior people still be allowed to struggle, even if AI can help them move faster? The first draft? The research? The client email? The recommendation? The meeting prep? The presentation? The difficult conversation?

Because not all friction is waste. Some of it is training. And in the AI-enabled workplace, the best organisations will not simply make junior people faster. They will make them better.

---

Sources

[1] Erik Brynjolfsson, Danielle Li and Lindsey Raymond, "Generative AI at Work", Quarterly Journal of Economics, 2025. [2] Microsoft and LinkedIn, "2024 Work Trend Index: AI at Work Is Here. Now Comes the Hard Part", 2024. [3] Shakked Noy and Whitney Zhang, "Experimental evidence on the productivity effects of generative artificial intelligence", Science, 2023. [4] Fabrizio Dell'Acqua et al., "Navigating the Jagged Technological Frontier: Field Experimental Evidence of the Effects of AI on Knowledge Worker Productivity and Quality", Harvard Business School working paper with Boston Consulting Group, 2023.