Who Trains Your People When AI Does the Entry-Level Work?

IN ONE MINUTE

The tasks that used to train junior staff (research, first drafts, data gathering) are exactly the tasks AI now handles well. If organisations don't redesign how people learn, they risk a gap in future expertise. McKinsey's Bryan Hancock and Charlotte Seiler recommend four moves: codify how the best performers think, redesign roles around working with AI, build learning into real work, and upskill managers as coaches. The goal is to use AI to speed up how people build judgement, not to remove the path by which they build it.

THE OBJECTIVE

Last week's issue asked how graduates get hired when AI has changed the entry-level job. This week is the sequel question: once people are inside the organisation, who trains them now?

The aim is to help leaders rethink early-career development so that AI accelerates expertise instead of hollowing it out.

WHAT THE RESEARCH SHOWS

• McKinsey's HR Monitor 2026 (5,500 workers in ten countries): 24% of employees had no training in the previous 12 months, and more than half received feedback once a year or not at all.
• The Bank of England (Aug 2026) reported that AI adoption is leading many businesses to slow hiring or leave vacancies unfilled.
• Indeed (Aug 2026): UK graduate job postings are at their lowest for the time of year since 2020. Its senior UK economist noted that early jobs are "often where foundational workplace skills are built".

The concern isn't only fewer junior jobs. It's that the learning those jobs provided may quietly disappear with them.

WHY THE OLD LADDER IS BREAKING

For decades, expertise was built through repetition. A junior analyst built a hundred models before being trusted with a judgement call. A junior lawyer reviewed a thousand documents before advising a client. A junior engineer fixed small defects before designing a system. The routine work was not only output. It was the training.

AI now does much of that routine work faster and cheaper. That is a real gain. But if nobody redesigns how people learn, the organisation quietly removes the first rungs of the ladder while still expecting people to reach the top. In five to ten years, the question becomes: where will our experienced managers, specialists and leaders come from?

The answer isn't to protect routine work for its own sake. It is to rebuild deliberately the learning that routine work used to provide, and AI can help do that faster than before.

A GOOD EXAMPLE

Standard Chartered has put skills at the centre of how it develops its people. Its internal talent marketplace lets employees take on project-based "gigs" across the bank, and 12,500 employees signed up in the first year. In a February 2026 interview with McKinsey, Tanuj Kapilashrami, the bank's chief strategy and talent officer, said participants are much more likely to move into areas aligned with future skills, and that their retention and engagement rates are much higher. In her words: "The adoption of AI at scale is not going to be a tech challenge—it's a skills, leadership, and culture challenge."

THE FRAMEWORK: FOUR MOVES TO REBUILD THE LADDER

The four recommendations below are from McKinsey's Bryan Hancock and Charlotte Seiler (July 2026). The right-hand column is my own set of questions for leaders putting them into practice.

McKinsey's recommendation	What it means	The leader's question
1. Knowledge management	Codify how your best performers think: their frameworks, decision rules and past judgements	Are our AI tools teaching our standards, or generic ones?
2. Workflow and role redesign	Rebuild roles around working with, supervising and improving AI outputs	Which junior roles have we redesigned, not just automated?
3. AI-based learning design	Embed learning into real work, so people build judgement while producing genuine outputs	Where in the workflow does someone explain why, not just what?
4. Manager upskilling	Treat coaching as a core capability, not an afterthought	Is every manager equipped to coach judgement, not just review output?

WHAT THIS MEANS FOR CEOS AND EXECUTIVE TEAMS

Treat early-career development as a strategic risk, not an HR programme. The leadership pipeline of 2035 is being built, or not built, by today's decisions on junior roles. Ask for it to be reviewed alongside succession planning.

Don't let short-term savings remove the ladder. Cutting entry-level roles can look efficient this year. Ask what it does to the supply of experienced talent later, and what it will cost to buy that experience back.

Make learning part of the AI business case. Every major AI deployment should state how it changes the way people learn in that area, not only how it changes cost and speed.

WHAT THIS MEANS FOR HR AND L&D LEADERS

Map where judgement used to be learned. For each critical role, identify the tasks that once built experience, and decide how that learning will happen once AI does those tasks.

Capture expertise before it walks out of the door. Work with senior practitioners to document how they think: their frameworks, decision rules and the lessons from past calls. That knowledge can then shape the AI tools and training juniors use every day.

Move from courses to real work. Build learning into live projects, with structured review and reflection, rather than relying mainly on classroom or online modules. Stretch assignments and short internal "gigs" are powerful ways to do this.

WHAT THIS MEANS FOR LINE MANAGERS

Coach the why, not just the what. When a junior shows you AI-assisted work, ask them to explain why it is right, what they changed, and what the AI missed. That conversation is where judgement grows.

Review the AI's output together. Use AI drafts as teaching material. Walking through what a strong answer looks like, and why, can teach more in ten minutes than a week of routine tasks used to.

Give responsibility earlier, with support. If AI removes the routine work, juniors can take on meaningful problems sooner. Pair that responsibility with regular check-ins, so stretch doesn't become sink-or-swim.

WHAT THIS MEANS FOR EARLY-CAREER PROFESSIONALS

Use AI to learn faster, not to skip learning. Ask it to explain its reasoning, challenge your own first draft, and show you alternative approaches. Then check it.

Seek feedback actively. More than half of employees receive feedback once a year or not at all. Don't wait for it: ask for specific feedback on real work, regularly.

Build a record of judgement. Keep track of the decisions you made, what you learned from them, and where you caught the AI being wrong. That record is what earns the next level of trust.

THE LAST WORD

If AI takes the first rung off the ladder, leaders have to build a new one.

In your organisation, where do early-career people now learn the judgement that the old entry-level work used to teach?

—

Until next week,

Artificial Ian

Tell your network: if a colleague manages graduates or apprentices, these four moves might be useful in their next team plan.

→ Found this useful? Forward it to an HR, L&D or team leader.

→ Subscribe to The AI Executive Briefing, weekly in your LinkedIn inbox.

Sources: McKinsey, "Building expertise in the age of AI: Who trains the next generation?" (Bryan Hancock and Charlotte Seiler, Jul 2026); McKinsey, "Leading a skills-based transformation powered by AI" (interview with Tanuj Kapilashrami, Feb 2026); McKinsey HR Monitor 2026 (Jun 2026); Financial Times reporting on Indeed and Bank of England data (Aug 2026).