The opinions of contributing entrepreneurs are their own. </p><div>
Key Takeaways
- Automate mechanical tasks, but protect the judgmental work that develops future leaders.
- AI should accelerate the learning process by moving juniors towards analysis, evaluation and decision-making.
Have you ever thought about what will happen to your business if you stop teaching people how to think?
I keep coming back to that question as more teams transition entry-level work to generative AI. Yes, the output still appears. The spreadsheet is still being built. The dashboard continues to be updated on time. And yes, on paper, productivity looks better than ever.
However, the silent costs lie elsewhere. The junior employees who used to earn their judgment through that work don’t get the same reps. They no longer struggle with messy input. They don’t make the kind of little mistakes that instinct creates. They are not coached by the blind spots that turn ‘smart’ into ‘reliable’.
When I look at the A players on my own teams, they didn’t become great by avoiding mistakes and doing fundamental work. They became great because they did it anyway, got feedback, did it again, and learned from real people’s experiences. If you remove that path completely, you create a dangerous kind of organizational shortsightedness. The knowledge may live within systems and cues, but fewer people learn how to produce, challenge and pass it on.
This is not an argument against AI. It’s an argument for using it on purpose.
The work that teaches judgment is not the same as the work that wastes time
Many entry-level tasks take time. They are repetitive. They are often at the basis of a process. Leaders see that pile and immediately think: “Automate it.”
That’s where the error starts.
Some entry work is mechanical. It needs to be done, but it doesn’t build much judgment. For example, formatting decks, pulling standard reports, cleaning up repetitive spreadsheets, or setting up a first-pass template that follows the same pattern every time. If AI can handle these tasks well, you should allow it. Protecting busy work doesn’t build talent. It burns it.
Other entry-level work is where the judgment is made. It is the moment when someone learns to separate signal from noise. It is the moment when they realize that a familiar approach does not suit a particular situation. It’s the moment they learn why the company cares about one metric and ignores the other. This work builds future leaders and is exactly the work you can’t do through AI without replacing it with something equally developmental.
If you treat both categories the same, you will get the worst result. You take away the training ground, and then you wonder why your bench has become weaker two years later.
What AI has changed for us
After decades of building and scaling teams, I’ve learned that new technology is rarely the real challenge. The challenge is to redesign work so that technology absorbs the mechanisms while people grow to become higher-value contributors. That’s where scale comes from, and that’s where resilience lives.
Here’s a simple example.
A junior analyst spent hours collecting data and preparing spreadsheets, after which he or she was given a short period to interpret what the numbers meant. That’s backwards. AI can often handle the pulling and formatting quickly, meaning the analyst can spend their time on the part that actually teaches them something. They can test assumptions. They can see what doesn’t look right. They can explain what the data suggests and what it doesn’t.
The same shift applies to all functions.
When AI drafts an internal memo, the junior employee should not be judged on how quickly they can press send. They need to be taught how to judge whether the memo answers the right question and whether the recommendation holds up when the context changes.
As AI summarizes research, the junior employee can be expected to discover what is missing and what conflicts emerge. A clean summary is not the same as a reliable conclusion.
This is not about doing less work. It’s about doing different work and doing work that builds skills.
How to do this right without slowing down
Take a fresh look at your entry-level team’s workload. Separate the purely procedural tasks from those that require trade-offs. If a task can be completed by following a checklist, automate it. If judgment is needed, delegate it to people.
What comes next is where most organizations get stuck. You can’t take away the mechanical work and hope the development happens on its own. You need to redesign the assessment so that juniors still get reps, coaching and responsibility. This means that assessment standards must be introduced. It means juniors have to explain why an AI output is correct and what would make it incorrect. It means giving them control over the thinking, not just the outcome.
Finally, track more than just productivity. If your only scoreboard rewards output and efficiency, you are optimizing for the wrong future. Pay attention to whether your junior team gets better at analysis and decision-making over time. If that’s not the case, you’re not building real capacity.