Have you ever watched a slow leak in a tire?
It doesn't blow. There's no bang, no pull on the wheel, no drama to make you stop. The pressure just quietly drops, mile after mile, until one cold morning the warning light comes on and you realize you've been driving on it for weeks. By then the damage is done.
That, almost exactly, is what artificial intelligence is doing to employee engagement inside most organizations right now. Not a blowout. A leak. Walk a sales floor, a contact center, a clinic, or a store and you won't hear open rebellion against AI. You'll feel something quieter and more corrosive: a low hum of uncertainty, people doing their jobs while a single question runs underneath everything they do. What does this tool mean for me?
I call that question the quiet leak, and it is the central challenge of employee engagement in the age of AI. People can stay engaged through hard work, through change, even through a bad quarter. What drains them is not knowing where they stand. And AI, for all its promise, is the most efficient ambiguity-generating machine we have ever installed in the workplace.
The number that should stop you cold
Let me give you the data before the advice, because the stakes here are not abstract.
Each year, Gallup measures employee engagement across virtually every economy on earth in its State of the Global Workplace report. The 2026 edition landed on a sobering figure: only about 20% of employees worldwide were engaged in 2025. Read that again. Four out of five people show up to work without being meaningfully invested in it. Gallup puts the price of that disengagement — the lost productivity, the turnover, the quiet underperformance — at roughly $10 trillion globally. Not billion. Trillion.
Now drop AI anxiety into a workforce that was already running on 20% engagement, and you can see the problem. We are introducing the most disruptive workplace technology in a generation into a system that has almost no engagement to spare. The leak doesn't start when AI arrives. AI just opens it wider.
Here is the part leaders most often miss. The same Gallup research has shown for years that managers account for roughly 70% of the variance in team engagement. Seventy percent. Not the CEO's vision, not the all-hands, not the values painted on the lobby wall — the manager. Which means the entire weight of whether AI lifts your people or hollows them out comes to rest on the shoulders of the very managers we tend to inform last and equip least.
Why AI hits engagement differently than any change before it
Leaders have steered teams through change before — reorganizations, new platforms, mergers. AI is different in three ways that matter, and naming them is the first step to managing them.
It touches identity, not just workflow. A new scheduling system changes how you work. AI quietly raises the question of whether your work still needs you. That is a far more personal fear, and personal fears do not respond to a project plan or a Gantt chart.
It outruns the communication around it. By the time most organizations get around to explaining an AI tool, employees have already written the story themselves — usually an anxious one — from headlines, hallway talk, and a friend who got laid off somewhere else. In that vacuum, silence is not read as neutral. It is read as something being hidden.
It lands first on the frontline. That is precisely where engagement is most fragile and most consequential, because your frontline employees are not an internal HR statistic. They are your customer experience, walking around in human form. A disengaged barista, nurse, or agent delivers a disengaged interaction, and no efficiency gain ever outruns that math.
The Ritz-Carlton lesson hiding in plain sight
Years ago, while writing The New Gold Standard, I watched The Ritz-Carlton do something most companies still find unthinkable. They give every single employee — housekeepers, bellhops, line cooks — discretion to spend up to $2,000 per guest, without asking a manager, to solve a problem or create a moment of delight.
Pause on what that policy communicates. It says: we trust your judgment. We believe you will use power well. You are not a cost to be managed; you are a professional we have equipped to act. That single design choice does more for engagement than a wall of motivational posters, because engagement is not a feeling you hand to people. It is a byproduct of how trusted, equipped, and seen they feel.
That is the exact lens to bring to AI. The technology you deploy is, whether you intend it or not, a message about how much you trust your people. Deploy it to equip them — to strip away the drudgery so they can do the human work only they can do — and engagement rises. Deploy it to monitor them — to watch, score, and second-guess — and you will get the compliance of people protecting themselves, which is the opposite of engagement. Workers can tell the difference in about a week.
Three moves that turn the leak into lift
1. Trade ambiguity for transparency — fast. The single most engaging act available to a leader in an AI transition is also the cheapest: clarity. Tell people what the tool does, what it does not do, why you are adopting it, and above all what it means for them. "We are using this to remove the parts of your job you have told us you hate, so you can spend more time on the work only a person can do" is a fundamentally different message than a tool simply appearing on Monday. The facts can be identical. The engagement outcome is not.
2. Equip your managers first, not last. If managers drive 70% of engagement, they cannot be the last to understand your AI strategy. Give them the why, the honest talking points, the permission to acknowledge fear out loud, and the answer to the question every employee is truly asking. A manager who can speak about AI with calm, candid clarity protects more engagement than any town hall ever will.
3. Name the fear before they have to. The fastest way to defuse anxiety is to say the quiet part first. "Some of you are wondering whether this is the first step toward needing fewer of us. Let me tell you honestly how we are thinking about that." Acknowledged fear loses most of its power. Ignored fear compounds in the dark — and leaks.
The reframe that changes the math
Here is the shift I urge every leadership team to make. Most organizations treat AI as a productivity initiative that happens to affect people. Flip it. Treat it as a people initiative that happens to involve technology. The productivity will still come — but only if engagement survives the transition, because disengaged employees deliver disengaged experiences, and that is the one cost no efficiency gain can cover.
AI that doesn't work is not, at its root, a technology problem. It is a leadership problem. And so is AI that works beautifully on the spreadsheet while quietly draining the people who run it. The tools do not decide whether your team feels seen and secure through this change. You do.
So let me leave you with the question worth carrying into your next leadership meeting: if engagement in your organization is leaking right now — and at 20% engaged, the odds are it is — is your AI rollout opening the leak, or sealing it?
If you are not sure of the answer, that uncertainty is exactly the work I help leadership teams do. Come find me at josephmichelli.com/contact, and let's make sure your people come through this transition more engaged, not less.
