When I was a child, my family did a supermarket shop every couple of weeks and the kids came along. My parents wrote the list, sometimes split it between us, and sent us off to fetch things.
Nobody called it teaching. It was called helping, and we complained the whole time. But it taught you how to scan a shelf, how to find something without being told where it was, how to weigh price against size, and how to read a discount and work out whether it really was one. It taught you to hold next week’s meals in your head while you did all of that.
I can do that same shop from an app now, in about ten minutes. I already know how to compare the products and decide what we need, so doing it through the app is straightforward. No store, no trolley, no queue, no bags to carry to the car.
All of that is better for me. It also means my child misses out on what I learned from those trips.
I could take her to a supermarket just so she can learn how to shop. But it’s slower and more annoying, and it isn’t something I can do regularly. It’s not that I miss pushing a trolley around. I want her to learn the things I picked up without realising I was learning them.
I can teach her to buy things through the app instead. We need eggs, there are the eggs, we need detergent, there is the detergent. But knowing how to find those products and place an order doesn’t teach her how to compare brands, prices and sizes, or work out what we need for the week.
A supermarket chooses what to stock, but the shelf lets you scan the available options together. An app can have ten products available while showing you only two or three that its algorithm wants to promote. She needs to know to look beyond those suggestions and compare the alternatives.
Teaching someone to use the app is not teaching them how to shop.
I need to teach her both at the same time: how to use the app, and how to make the choices that go into the shopping. That means talking through prices, quantities and what we plan to eat while we put the order together. It’s much more work than showing her which buttons to press, or finishing the shopping myself in ten minutes.
Doing the work gives people a chance to learn
For as long as I’ve worked in marketing, managers have delegated tasks like competitor research, first-pass analysis, building monthly reports and cleaning data to junior team members. The work needed doing, and someone had to do it.
Some of the learning happened simply because people were doing the work, even when it wasn’t assigned as a training exercise. A junior who built the report forty times could learn what the report was for, which numbers move and which are noise, and what people need from it.
Putting the numbers together gives them the chance to understand what those numbers mean, notice changes and spot patterns or errors. Knowing how to build the report is helpful, but understanding it is what they need to develop. AI can build it faster, while a junior may submit the result without noticing numbers that an experienced manager can see are wrong.
An experienced manager knows when to reject an AI output, when to give it more guidance, and when to pause and rethink the task. A junior who hasn’t developed that judgment may simply accept what comes back. When we teach them to work with AI, those decisions are part of what we need to teach.
When a manager does that work themselves with AI instead of involving a junior, the work gets finished, but the junior misses the opportunity to learn from doing it.
Two ways to keep developing those skills
Keep giving people work they need experience doing, even when AI could do it faster. Allow time for them to learn, and choose work whose outcome matters to you so you’ll review it and give useful feedback.
Give the junior responsibility for the AI-assisted task. They can do the analysis, research or first-pass build with your supervision, rather than you doing it yourself in four minutes. Make sure they know how to approach the task, and work through their questions and corrections with them until it is complete.
Prompt courses, tool onboarding and a shared prompt library can help people use AI. But they also need practice judging whether the analysis makes sense and what to do with it.
As a manager, you’ll need to spend time teaching that you could have spent finishing the task yourself. Allow for that from the start, and stick with it when the first few attempts take longer than you hoped. This is the time you already spent helping people develop, even though AI now gives you a faster option for finishing the work.
Developing people still needs your attention
Developing people has always been part of a manager’s job. Being able to delegate work to AI can pull managers away from that responsibility, because it’s often quicker than working through the task with a junior.
AI still needs briefing, checking and feedback. But it doesn’t need you to plan its career development or give it work so it can build experience. Managing people includes both getting the work done and helping them develop.
Keep involving people in real work, reviewing what they produce and giving feedback through to completion. Help them understand the task itself and how to use AI to do it more efficiently.
Build understanding into the task
For a junior learning to produce a report, the practice can combine doing calculations themselves, using AI and checking the results:
- Build and check a section at a time. Compare the numbers with previous reports. Establish expected ranges for funnel conversion rates so the junior can flag a result that needs a closer look.
- Compare their calculations with AI’s. They could work through a section manually, ask AI to calculate it too, and investigate any differences. Or they could calculate the numbers themselves and use AI to put the report together, learning how to present the results as well as how to calculate them.
- Spend equal time on judgment when teaching the AI workflow. Ask what would make the output wrong and what they would need to check. Have AI show the calculations and steps so the junior can work through them and check how the result was reached.
- Ask them to explain their choices. When reviewing AI-assisted work, ask what alternatives they considered and why they chose the version they brought you. They should be able to explain what they rejected and why.
In your regular one-to-one, discuss the work you’ve delegated for learning and how it went. As they gain experience, people can tackle more difficult work with less guidance from their manager.
Many companies are working out how to develop junior employees as AI takes over more of the work they used to learn from. HR programmes can help, but managers need to plan for this in the work they give people and the feedback they provide.
Pick a task your team needs to learn, give someone responsibility for it, and help them develop both their ability to use AI and their judgment about the work it produces.