A senior executive walks into a coaching session with fire in his eyes. What happens next is a masterclass in how AI hype is eroding human leadership and why the most dangerous decisions today are made in its shadow.

Some coaching sessions start with a handshake and a coffee. This one started with: "Are they totally out of their mind?"

No pleasantries. No warm-up. The thunderstorm was already in full effect before I could even gesture toward a chair. I gave him the stage because when a senior executive walks in like that, the most useful thing a coach can do is get out of the way and let the lightning strike.

What followed was one of the most quietly disturbing corporate stories I've heard in recent memory. And I've heard a few.

"My dashboard fed by AI now shows, in real time, the estimated headcount reduction AI should deliver to my org. Not in percentages. In people. And today, heading into the final month of H1, that's the number I'm being asked to hit."

"We're offering a voluntary package. Twelve months' pay. Sign by the end of June, and there's a $5,000 bonus on top. We call it the fast-track reward. We're supposed to tell the people we think should take it that it's voluntary, no big deal if they don't. But what we won't tell them yet: if we don't hit 10%, those who didn't volunteer will probably be walked out without a package. That's what drives the right behavior, apparently."

"André, what in the world should I do? My values aren't vulnerable. They're about to burst."

Sit with that for a moment. An algorithm calculates a headcount. A human is handed that number as a target. And then that human is asked to have "voluntary" conversations with colleagues, carrying undisclosed information that transforms consent into pressure. All while being told to smile and call it a program.

The Psychology Trap: Three Locks on One Door

My coachee is caught. Not because he is weak, quite the opposite. He is caught precisely because he has a strong moral compass, and that compass is being systematically jammed. Let's name the traps clearly.

TRAP ONE

Complicity by proximity. He didn't design this program, but he's been handed the delivery role. The system makes him the face of something he doesn't own.

TRAP TWO

Moral injury through information asymmetry. He knows what employees don't: that "voluntary" has a deadline after which it becomes involuntary. Holding that secret in a relationship of trust is corrosive.

TRAP THREE

The frozen middle. Push back upward and risk being labeled difficult and added to the "voluntary list". Execute downward and betray his own people. Paralysis masquerades as professionalism.

The psychological term for what he's experiencing isn't just stress; it's moral injury: the damage done when we're forced to participate in actions that violate our own ethical and/or moral framework. It's common in healthcare, in military contexts, and, increasingly, in corporate leadership during AI-driven transformation. The wound is real. And if left unaddressed, it doesn't just affect him. It leaks into every conversation he has with his team.

AI Needs to be De-Hyped before Decisions become Irreversible

Let's be direct about something that isn't being said loudly enough in boardrooms right now: the efficiency number on that dashboard isn't a fact. It is a projection. A model's best guess, trained on data from contexts that may bear little resemblance to this particular team, this particular culture, this particular moment in the business cycle.

AI is extraordinarily good at pattern recognition. It is considerably less good at knowing which patterns matter. It cannot model the institutional knowledge that walks out the door with a 20-year employee. It cannot quantify the erosion of trust that occurs when colleagues watch each other be "fast-tracked" out. It cannot calculate the two-year hangover of a workforce that survived and is now wondering if they're next.

And yet: real people are being asked to make irreversible decisions based on this number. Not a strategy. Not a judgment. A number on a dashboard with a real-time ticker.

The de-hyping conversation that must happen, in coaching rooms, in leadership teams, in board meetings, is this: AI is a powerful input. It is not a mandate. When we treat its outputs as instructions rather than signals, we don't become more efficient. We become less human. And in doing so, we often become less intelligent, too, because we have outsourced the judgment that was always supposed to be ours.

The executive holding this dashboard is not being asked to lead. He's being asked to execute. And there is a profound difference, one that the best leaders have always known, and that AI cannot paper over.

The Three Coaching Questions, and Why they Matter

This is where the coaching work begins. Not with solutions, he doesn't need more options on a menu. He needs to find his own way through, with his values intact. These three questions are designed to do exactly that.

QUESTION ONE

"If the people you're about to have these conversations with knew everything you know, what would a genuinely voluntary conversation actually look like?"

My Intent: This question surfaces the information asymmetry and forces him to confront it directly, not abstractly. It moves him from "I'm just following instructions" to "I am actively choosing what to disclose." That is agency. And agency is the antidote to the frozen middle. It also begins to open a creative space: what could transparency look like, even within constraints?

QUESTION TWO

"What is the cost to the business, not the headcount, but the actual business, if the 10% who leave are not the 10% the algorithm intended?"

My Intent: This reframes the AI output as a risk signal rather than a guaranteed outcome. Voluntary programs systematically attract the most employable people, those with options. The remaining workforce after a poorly designed program is often a self-selected group of those who couldn't leave or chose to stay despite everything. This question invites him to build a business argument that challenges the program's premise and gives him language to move upward rather than simply absorb downward pressure.

QUESTION THREE

"When this is over, whatever over looks like, what do you want the people in your team to say about how you showed up?"

My Intent: Legacy is not a soft concept. It is a navigation tool. This question bypasses the immediate noise of the program and anchors him in the identity he has built as a leader. It is not designed to make him feel better; it is designed to make clear the cost of abandoning his values and to surface what kind of leadership he is actually capable of in a moment like this. The answer to this question is usually where the real coaching begins.

The Uncomfortable Truth about Where We Are

My coachee is not an outlier. There are hundreds of them right now, capable, values-driven senior leaders handed AI-generated targets and told to "make it human." The cognitive conflict is breathtaking. The collateral damage to people and organizations is only beginning to become visible.

We are in a moment where the speed of AI adoption has dramatically outpaced our organizational ethics, our leadership development infrastructure, and our collective honesty about what these tools can and cannot do. The result is exactly what walked into my coaching room: a smart, caring leader whose values are "about to burst", not because he is failing, but because the system he's operating in is asking him to.

The answer is not to reject AI. It is to lead it consciously and critically, with human judgment firmly in the driver's seat. An efficiency dashboard is a mirror. What we do when we look into it is still, and must remain, a human choice.

And if the number on the dashboard is the only thing driving the decision? Then, the most important thing the algorithm forgot to calculate was the leader standing in front of it.