The CEO Who Sounds Like Everyone Else Has Already Lost

When I see the videos of the latest advances in robotics, machines doing cartwheels, two thoughts come to mind: the first, why would we make robots look and behave like us? Surely, there must be more efficient forms for the tasks at hand. Second, are we too at risk of becoming like machines doing cartwheels, but with our language, perfect every time, but without a soul and far from being distinguishable from each other?

Earlier this year, in a hushed room where global minds who manage trillion-dollar portfolios gathered, balancing the weight of futures, I watched CEOs stand at the podium in Davos. After clearing their throat, they delivered a fluent, polished, and credible speech. Every sentence had the rhythm of a thousand TED transcripts. The next day, I found myself challenged to remember a single word. Surprised, because I am usually rather attentive, I then wondered whether anyone remembered a word and whether it had any impact on the other listeners or the world. Aren’t forums like this supposed to start movements that change things, something, anything?

When you are in rooms such as this one, you begin to feel the air thin when authenticity drains out. I can feel my intellect start to gasp in these moments, especially as AI makes it effortless to sound that way: competent but forgettable. It is an interesting and sharp paradox that speakers find themselves in. The human voice that doesn’t sound engineered becomes the rarest asset. Outsource it, and you lose the war of attention, not to mention distinctiveness.

Consider the mechanism. Large language models train on the median of human output: the safest phrasing and the most probable turns. No akrasia, that weakness of will where a leader admits the path ahead forks into fog. No paltering, the half-truth that reveals more by what it withholds. Just the blandification Byung-Chul Han warned of: everything smooth, nothing jagged.

Recently, I was advising a communications director who prompted an AI to create a crisis town hall script. The output was full of empathetic openers, forward-looking positivity, and, moreover, and unfortunately, bullet-proof platitudes. The CEO delivered it verbatim. Employees nodded along, with what looked like almost rehearsed resolve. Two weeks later, the turnover they were trying to slow, spiked. Trust isn’t built on fluency. It’s forged in the friction of a voice that bears scars.

The boardroom standoff in Seoul, where concessions cost a deal but saved a partnership. The all-nighter in Paris rewriting a layoff memo because “transparent” platitudes (again, one of the hallmarks of AI) rang hollow against human cost. If you take these for data points, you would be inherently wrong; they’re the irreducibly human residue. Marshall Rosenberg called it nonviolent observation: naming what is, without the violence of euphemism. Models excel at the violence.

Yet leaders outsource to AI anyway. Often, insecurities and a lack of clarity of thought allow it, but increasingly speed demands it. And beyond the insecurities, indeed settling them, platforms reward it (the LinkedIn algorithm loves the median voice of shared incompetence).  The result? CEOs who sound like their peers, their analysts, their own press releases. In a 2025 Forrester report, 68% of C-suite comms now involve generative tools. Fine for memos. Fatal for vision.

When every leader sounds like a slightly warmer McKinsey slide, the one who speaks from akrasia (from the tension of knowing better but acting anyway) cuts through. Think Elon Musk’s erratic blasts, or Indra Nooyi’s handwritten letters to employees’ mothers. Not polished. Human. Memorable. They don’t scale, but neither does trust. (Note that this is not a value judgment on what they say but on the delivery.)

So let’s do something I like to call “communication architectural programming,” where communication isn’t a department but the operating system. Figure one: The CEO’s voice sets the signal. Figure two: AI-generated, it homogenizes downward. Figure three: VPs mimic the median, directors dilute further, until the company speaks in echoes. Authentic, it amplifies: decisions carry the grain of experience, crises the weight of conviction. One leaves a positive sum; the same equation with slightly different inputs leaves a negative one.

I built Intradiegetic around this. Not style guides, governance systems for voice as strategy. And in no way are we anti- AI. Quite the contrary. We work to make Human X AI systems operate in perfect harmony, building an architecture that takes the best from both and amplifies what a prompt can’t do, building human connections upon which value remains steadfastly rooted.

The CEO who sounds like everyone else has already lost because attention is the new capital. AI will make more leaders sound credible. Fewer will sound alive. Don’t prompt your way to the median.

On AI, leaders need to choose their position now, before the flood rises.