You Hired People Who Are Better at AI Than Their Boss.

Thirteen percent of managers strongly agree they are equipped to lead AI-native employees.

Among managers who are already doing it, 88 percent feel capable.

That gap is the whole story, and it is not about training budgets.

The numbers come from a survey Indeed ran with YouGov across more than 1,000 job seekers and 300 hiring decision makers. The headline everyone quoted was that 43 percent of managers feel poorly equipped or not equipped at all. The number nobody quoted is the one that tells you what to do about it.

The Skills Gap Is a Management Gap

Nearly half of employers say their direct reports often have stronger AI capabilities than the managers running them.

Sit with that for a second, because it describes an organizational condition most companies have never had to handle at scale.

You promoted people because they were excellent at the work. Then the work changed underneath them, and the twenty-six-year-old on their team who has been using these tools since college is now more capable at a growing part of the job than the person doing their performance review.

That is not a competence problem. It is a structural one. And it produces predictable behavior: managers who avoid the topic, who approve AI work they cannot evaluate, or who quietly slow it down because they cannot tell good from dangerous.

Only 14 percent of the workforce identifies as AI native. Among Gen Z, 35 percent claim AI fluency. Among Gen X, it is 11 percent. Most managers sit on the wrong side of that line and know it.

Readiness Comes From Reps, Not Curriculum

Go back to the two numbers I opened with.

Thirteen percent of managers overall feel strongly equipped. Eighty-eight percent of managers who already lead AI-native talent feel capable of it.

The competence did not come from a course. It came from managing the actual situation. The gap closes on contact.

That should change what you do with your training budget. A generic AI literacy webinar for all managers produces very little. Hannah Calhoon, who runs AI at Indeed, makes the same point: the programs that work are role specific, not general.

Put a manager in charge of one person doing real AI work, give them a defined outcome, and let them be uncomfortable for a quarter. That produces more capability than any curriculum I have seen.

Worth noting: 88 percent of employers actively recruiting AI-native talent already provide manager training. The companies that thought hardest about the hire also thought about the boss. Most companies did neither.

You Are Expecting Things You Never Asked For

Here is where it gets unfair to the people you employ.

Seventy-three percent of employers expect staff to use AI tools for individual tasks, and 66 percent of workers are comfortable with that. Close enough.

Then it falls apart. Fifty-eight percent of employers expect workflow automation. Twenty-seven percent of workers feel comfortable doing it. Forty-one percent expect employees to manage AI agents. Fourteen percent feel comfortable.

That is a three-to-one gap between what leadership expects and what the workforce can currently deliver, on the exact capability that matters most.

Meanwhile The Conference Board found only a third of workers received any employer AI training in the past six months, and 28 percent get none at all, while 55 percent use AI weekly or daily anyway.

So the expectation is set, the support is not funded, and the performance review is coming. That is not a strategy. It is a setup.

This Is an Authority Problem

The instinct is to send managers to training. The better move is to fix what the manager is allowed to decide.

A manager whose report knows more than they do about a tool is not in an unusual position historically. Good managers have always supervised specialists who out-skilled them in a domain. Nobody expects a VP of Engineering to out-code every engineer.

What makes AI different is that nobody has told these managers what their job is now. Are they supposed to evaluate the output, approve the tool, own the risk, or set the standard? Unstated, all four land on them at once, with no authority attached to any of them.

Empowerment means giving people the tools, training, and authority to make decisions, then getting out of the way. Most companies are two-thirds of the way there and wondering why it is not working.

What I'd Do This Week

Name what the manager owns. In writing. Outcome ownership, not tool expertise. They own whether the work is right, not whether they could have produced it themselves.

Pair, do not train. Assign each manager one AI-native report and one real deliverable. Reps beat curriculum, and your own data will show it inside a quarter.

Close your own expectation gap. Write down what you expect from employees on AI this year. If you expect agent management from 40 percent of your people, fund it. If you will not fund it, stop expecting it.

Let managers say they do not know. The fastest way to keep a manager frozen is to make admitting the gap career-limiting. The 88 percent got there by being uncomfortable in public first.

The Part That Matters

Everybody is hiring for AI skills. Almost nobody is preparing the people who will manage them.

You cannot classroom your way out of this. You manage your way out of it, one uncomfortable quarter at a time.

Give your managers the reps and the authority. They will close the rest themselves.