For about two years, the standard worry went like this: senior leaders weren’t touching AI. Too busy, too skeptical, too far above the keyboard to type into a chat box. We all nodded along. Somebody made a slide about it.
That worry is dead.
Leaders are now the heaviest AI users in the building. Gallup’s late-2025 data has roughly 69% of leaders using AI to some degree, against 55% of managers and just 40% of individual employees, and the separation has widened every quarter since 2023. The grown-ups finally got in the pool. They are doing cannonballs.
The good news that curdled
More leaders using AI is genuinely good. Engaged executives are the single strongest predictor of getting value out of this stuff. McKinsey’s State of AI in 2025 found that the rare high performers, about 6% of organizations, are three times more likely than everyone else to have senior leaders who both own their AI initiatives and personally role-model using the tools. Meanwhile, 88% of organizations report using AI somewhere, but only 39% see any measurable impact on the bottom line. The differentiator isn’t the software. It’s leaders visibly doing the work.
So leaders climbing into the pool should be a happy story. Except they climbed in, looked around, and concluded that everyone else was swimming too.
They’re not.
The 2x problem
Here is the most uncomfortable statistic I’ve read all year. A Harvard Business Review study late in 2025 asked executives how their employees felt about adopting AI. Seventy-six percent of leaders said their people were enthusiastic. Then they asked the people. Just 31% of individual contributors actually were.
Leaders are off by more than double. They are standing on the sunny side of a gap they don’t know exists, waving at a crowd they assume is right behind them.
This is the trouble with becoming a power user. The more naturally AI fits into your own day, the easier it is to forget that your view from the top of the org chart is not the view from the middle of it. You delegated the boring parts of your job to a model this morning and felt like a wizard. Three layers down, someone is quietly worried that admitting they use the same model will get them labeled replaceable.
Your execs trust the robot more than their spouse
It gets better, in a way that is both funny and a little alarming. An SAP-sponsored survey of C-level executives at billion-dollar companies found that 74% of them trust AI for advice more than they trust their own family and friends. Forty-four percent said they would override a decision they had already planned to make based on what the AI told them. Thirty-eight percent would happily let AI make business decisions on their behalf.
Read that together with the HBR number and a strange picture comes into focus. The same leaders who trust an AI model more than the people they married cannot accurately read the mood of their own teams. They have extraordinary confidence in the machine and a blind spot the size of a parking lot when it comes to the humans they actually lead.
That is not a technology gap. No tool fixes it. It is a leadership gap, and it has a cost.
Meanwhile, on the frontline
While the corner office cannonballs away, regular AI use among frontline employees has stalled at around 51%. BCG calls it the silicon ceiling. Leaders and managers cruise past it; the people doing the day-to-day work hit it and stop.
The reason isn’t mysterious, and BCG measured it. The share of employees who feel positive about AI jumps from 15% to 55% when they have strong leadership support. Strong support roughly quadruples enthusiasm. It is the biggest adoption lever anyone has found. And only about a quarter of frontline workers say they actually get it.
Sit with the irony. Leadership support is the most powerful tool for adoption we know of. Leaders have never used AI more personally than they do right now. And most of their people still say the support isn’t reaching them. Using AI in your private browser tab is not the same as role-modeling it in the open, with your team, on real work they can see.
The reflex that makes it worse
When organizations finally notice the gap, the reflex is to treat it as a tools problem. Buy licenses. Hand IT the budget. Book a technical vendor to run a prompt-engineering workshop. Call it a transformation.
The numbers say that reflex is misfiring. Korn Ferry found that while 42% of CHROs are prioritizing AI investment, only 5% say their teams are actually well prepared to embrace it. The money is flowing. The readiness isn’t. You cannot purchase your way across a human gap.
MIT’s own executive education people put it plainly: AI literacy at the leadership level is about judgment, not technical detail. Leaders don’t need to understand model architecture. They need to understand where AI creates real value, where it quietly introduces risk, and how it changes who is accountable for what. That is a thinking skill, built through practice and reflection. It does not come in a license.
What the gap actually needs
The fix is less about the model and more about the mirror.
Get in the room with your people. Not to present at them, and not to forward another all-hands memo about responsible innovation. To work alongside them on real problems, watch where they get stuck, and find out what they’re afraid to say out loud. This is the whole idea behind starting bottoms-up, with the experiments your people are already running, instead of pushing a polished rollout down from the top. The hidden enthusiasm and the hidden fear both live in the same place, and you only find them by showing up.
Role-model in public, badly. Share the prompt that flopped. Show the dead end you hit and how you climbed out. Nothing closes the enthusiasm gap faster than a senior leader admitting the tool made them look foolish before it made them useful. Confidence theater widens the gap. Visible, honest fumbling shrinks it.
And get HR, IT, and the business in the same room while you do it. The capability questions, the human worries, and the strategy all sit with different people, and they only get solved when those people stop solving them separately.
The punchline
Two years ago we worried leaders were too far from AI. Now the risk is the opposite. They’ve gotten so comfortable with it personally that they’ve mistaken their own ease for everyone’s, and their own enthusiasm for a strategy.
The leaders who notice this, climb down from the high board, and get back in the shallow end with their teams will pull away from the pack over the next few years. Everyone else will keep waving at a crowd that isn’t there, feeling responsible while quietly falling behind.
The call is coming from inside the house. Pick up.
Sources:
- McKinsey, The State of AI in 2025: Agents, Innovation, and Transformation, November 2025
- Harvard Business Review, Leaders Assume Employees Are Excited About AI. They’re Wrong., November 2025
- Gallup Workplace AI Tracking (Q4 2025), reported via HR Executive, February 2026
- BCG, AI at Work 2025: Momentum Builds, but Gaps Remain, 2025
- SAP / Wakefield Research, Executive Trust in AI Survey, March 2025
- Korn Ferry Institute, Introducing the AI-Ready Leader, August 2025
- MIT Sloan Executive Education, AI Literacy: A Key Piece of an Executive’s Skill Set, February 2026