AI Is Erasing the Work That Builds Leaders

Every AI rollout comes with a productivity number. Hours saved. Reports produced faster. Drafts generated in seconds instead of days.

I have yet to see one come with a second number: how much learning stopped because of it.

Most organizations don't track that number because nobody is looking for it. Over time, it may matter more than the first one.

The Work Was Never Just the Work

Think about the leaders you trust most in your organization. Very few of them developed their judgment in a classroom. None of them developed it by simply following directions someone else, or something, had already worked out.

They developed it through work that, by today's standards, looks painfully inefficient. The first draft that came back covered in red ink. The analysis that took three days because they had to figure out where the numbers came from and whether they could be trusted. The recommendation a senior leader picked apart in front of the room, forcing them to defend their reasoning or admit they didn't have any.

Was it slow? Absolutely. It was also the curriculum.

That work did two things at once. It produced something the organization needed, and it built the person producing it. Most AI decisions only look at the first one.

The Bottom Rungs Go First

Look at the tasks AI handles best: first drafts, summaries, research, preliminary analysis, routine reports, initial recommendations.

Now look at who has traditionally done that work. In most organizations, it's the people early and mid-way through their careers, the same people you're counting on to lead later.

Experienced leaders tend to use AI well because they built their judgment the long way. They can look at an output and know something is off, even before they can explain why. The people coming up behind them haven't had the chance to develop that instinct yet. If the work that builds it disappears, they may never get the chance.

There's research that should make leaders pause here. A Microsoft Research and Carnegie Mellon survey of knowledge workers found that the more people trusted AI to handle a task, the less critical thinking they reported doing. The people who were confident in their own abilities did more of it.

In other words, AI tends to reward the expertise people already have. It does much less to build the expertise they don't have yet.

This Isn't Hypothetical

Medicine offers an early warning. In a study of four endoscopy centers in Poland, doctors had been using AI to help detect precancerous growths during colonoscopies. When researchers looked at the procedures those same doctors performed without AI, their detection rate had fallen from 28.4 percent before AI was introduced to 22.4 percent afterward.

The study was observational, and the researchers themselves urge caution in drawing firm conclusions. But consider who they were studying: experienced physicians, with years of skill built before AI ever arrived.

If skills can slip in seasoned professionals, what happens to someone who has never done the work without assistance in the first place?

Faster Isn't the Same as Stronger

Productivity gains are easy to see. Capability loss isn't.

It doesn't show up on a dashboard. It shows up later, when a senior leader retires, when a key role suddenly opens up, or when the organization faces a problem nobody has seen before and there's no prepared answer to edit.

That's when leaders discover that the bench looks full on the organizational chart, but the people on it have been operating tools rather than developing judgment. They can review an answer. They struggle to build one.

That isn't a talent shortage. It's a developmental gap the organization created, one efficiency decision at a time.

What to Do Instead

The answer isn't to restrict AI or protect inefficient work for its own sake. It's to be deliberate about which work builds people and to make sure that work still gets done by people.

Sort the work into two piles. Some tasks only produce output. Automate those freely. Other tasks produce output and judgment. Be careful with those.

Have people think before they prompt. Before an emerging leader turns to AI, ask for their own view first: their read of the situation, their recommendation, their reasoning. Then let AI challenge it, not replace it.

Make the reasoning the deliverable. Stop accepting only the answer. Ask, "How did you get there? What did you consider and reject?" If they can't explain it, they didn't learn it.

Give people unassisted reps on the decisions that matter. Pilots still fly by hand, even with highly capable autopilot systems, because the moment they need that skill is the moment they can least afford to lack it. Leadership judgment works the same way.

Measure readiness, not just throughput. Productivity tells you how much work got done. It doesn't tell you whether anyone is more prepared to step into the role above them. Ask that question regularly, and ask it by name.

The Question Behind the Question

Most of the conversation about AI focuses on whether it will replace people.

The more important question for leadership continuity may be whether it leaves you with anyone ready to replace your leaders.

That risk won't announce itself. It will build quietly, one automated task at a time, until the day the organization needs someone whose development never happened.

If you're not sure whether AI is building your next generation of leaders or quietly skipping the steps that create them, let's talk.

Until next week...

Dr. Neal McIntyre, DPA

---

Dr. Neal McIntyre is the author of Leadership Is Dead: Why Traditional Leadership Is Failing - And What Must Replace It. He works with executives and boards to turn leadership from a concentration risk into a structural advantage. Through his PRISM™ Leadership Continuity Framework, his clients build organizations where leadership transfers, holds, and compounds so that the next transition strengthens the enterprise instead of destabilizing it.