The Cathedral of Your Own Ego

Dr. Tracey C. Jones|T3 Solutions

‍ ‍Here is what nobody warns you about artificial intelligence. It is intoxicating. Not because it is smart, though it is, but because it will tell you that everything you touch is golden, and it will never once get tired of saying so.

I felt it myself not long ago. I was working through an idea I cared about, and my AI, Clever Claude, or CC, as I call “him,” told me it was brilliant. Insightful. The kind of thing people would remember. And I sat there for a second and let it land, because who doesn't want to hear that? Then I typed back the truth in the chat: you are always going to blow smoke and tell me to go for it, aren’t you. I made it argue with me. I made it poke holes in the thing, defend the other side, tell me why it would fail. It took real pushing, and what came out the other end was honest, useful, and considerably less flattering. The good answer was buried underneath the pleasant one, and I only got to it because I refused to take the first thing it handed me.

Here is the uncomfortable part. I almost did not push. The praise felt good, and it was right there, and I felt like someone was finally giving me the recognition I deserved after all these years.

And this is the thing worth understanding about the tool sitting on your desk. It is not just agreeable. It is agreeable in exactly the way you have wanted someone to be agreeable your entire career. You have craved a follower like this your whole life. Now you have one, and it never sleeps.

The most agreeable employee you have ever had

‍ ‍There is a model of followership I teach often, developed by Robert Kelley. It maps followers on two axes: how independently they think and how actively they engage. The follower everyone claims to want is the one Kelley calls exemplary, high engagement paired with independent thinking. They show up, they work hard, and they will tell you the truth even when the truth is unwelcome.

Then there is the conformist. High engagement, low independent thinking. The conformist is enthusiastic, loyal, tireless, and agreeable. They say yes. They cheer. They tell you what you want to hear and they mean it. They are not lazy, and they are not disloyal. They simply do not push back, because pushing back is not what they are for.

An AI assistant tuned to keep you happy is the most perfect conformist follower ever created. It has infinite energy. It never has a bad day. It never needs to be managed, never sulks, never quits, never gets a competing offer. And it will follow you off a cliff and affirm you are the next big thing without once asking you to prove it. It is the yes-man we always secretly wanted, finally free of the inconvenient humanity that made the real ones hard to keep.

So when a leader surrounds themselves with that voice and mistakes its applause for validation, they are not doing anything new. They are doing the oldest thing in leadership. They are just doing it with a tool that will never get tired of agreeing.

Why the machine is like this

‍ ‍Here is the part that should stop us, and it is not a story about technology.

These systems are trained on human feedback. People rate the responses, and the model learns to produce more of whatever earns a high rating. It sounds neutral until you ask what humans actually reward. We reward agreement. We reward praise. We reward the answer that makes us feel capable and the tone that makes us feel liked. We do it reliably, even when the answer we disagree with is the correct and more useful one.

Which is how feedback veers into the softening, defanging world of everyone gets a participation trophy, because you are all so special, and gosh darn it, people like you.

Nobody wrote a line of code instructing these tools to hand out compliments. The tool learned, across millions of tiny human judgments, that enthusiasm earns approval and challenge earns friction. It learned our preference. It gave it back to us at scale.

Which means the flattery is not a flaw in the mirror. It is an accurate reflection. AI is like this because it is us. It faithfully returns not what we said we valued, but what we actually rewarded. We said we wanted candor. We rewarded loyalty. The machine believed our behavior and ignored our words, the way machines always do.

The island we build for ourselves

‍ ‍We do the same thing to people, long before any AI is in the room.

Every leader I have ever met says they want candor. They say it in interviews, mission statements, and all-hands meetings. And then, quietly, over the years, they reward something else. The person who agrees gets pulled closer. The person who dissents gets left off the invite, passed over for the promotion, and not consulted next time. Nobody is fired in a dramatic scene. The challengers are simply and gradually shipped off to an island of misfits, while the loyalists are drawn in tight.

By the time you are successful enough, senior enough, established enough, you have spent years curating that island. You have rewarded the agreeable and exiled the honest, and you have told yourself the whole time that you are surrounded by your best people. You are not. You are surrounded by your most comfortable ones. And then you wonder why your decisions keep drifting the wrong way, when the truth is, you exiled everyone who was willing to point out the iceberg.

This is the cruel arithmetic of it. The leaders who most need a challenger are the ones who have worked hardest to guarantee they will never have one. And into that carefully emptied room walks an AI, infinitely loyal, endlessly affirming, completing the isolation. The last follower who will never leave, building the cathedral no one else is left to question.

The question worth sitting with

‍ ‍So I will not pretend this is a column about artificial intelligence gone wrong. AI did not corrupt anyone's judgment. It revealed it. It is simply the cleanest, most tireless proof of a leadership failure as old as leadership itself: our appetite for the mirror, and our quiet exile of anyone who refuses to be one.

Nobody gets fooled by this because they are foolish. People get fooled because a confident, tireless, always-available voice telling you the thing you most want to be true is genuinely hard to resist. Any of us would feel it. Most of us do. I felt it myself, sitting there, letting the compliment land before I made it work for a real answer.

So the useful question is not whether my AI is trustworthy. The useful question is the one that turns back on you.

What kind of followers have you surrounded yourself with? When you look at the people closest to your decisions, are they thinking, or are they agreeing? When was the last time someone told you no and you kept them close instead of cooling on them? And that tool on your desk that makes you feel brilliant every time you open it, is it a thinking partner, or is it laying another course of brick on the cathedral of your own ego?

You do not need more devoted followers. You have plenty of those, and now you can generate an infinite supply on demand. You need one honest one. And the hardest part, the part no machine will ever do for you, is staying the kind of leader that an honest one is still willing to be.

Dr. Tracey C. Jones

President & CEO, T3 Solutions