The Average You Didn't Know You Were

Identity does not leave all at once.

It leaves a paragraph at a time. A sentence here, a word choice there. You do not wake up sounding like everyone else. You drift into it — gradually, politely, almost gratefully — until the day you reread something you published and realize you have never heard that voice before. It is competent. It is clean. It has the unmistakable posture of someone who has been taking advice from a machine for eighteen months.

And you cannot quite say when the drift started.

This is the part of the AI era almost nobody is diagnosing clearly, because it is not a failure of the tool. It is not a bug in the model. It is a quiet, cumulative edit to the self — and it happens by the most innocent mechanism imaginable.

You write something raw. The machine suggests a softer version. You accept the softer version. The machine suggests a more structured one. You accept the structured one. You do not notice that every suggestion is pointing in the same direction — toward what the model considers balanced, professional, legible. Toward what the model learned from reading a hundred thousand LinkedIn posts by people who have already been flattened.

Each edit is a tiny concession. A milligram of identity removed. After a year of milligrams, you are not the same voice anymore. You are an edited version of you, rendered toward a statistical average of competent business writing — and the average is the average because it is what sounds right to the system, not because it was ever the truth of who you were.

Here is the part that should make anyone who cares about their public thinking very uncomfortable: you did not feel the loss. You felt the opposite. You felt more polished. More legible. More professional. The drift felt like improvement.

It was not improvement. It was convergence.

Identity is the thing that resists averaging. It is the weird analogy that does not survive a tone guide. It is the sentence that sounds wrong until you realize it is the most honest thing you have written all year. It is the rhythm that does not match what a machine learned from a million other people finishing the same thought. Identity is the part of your thinking that cannot be derived — only noticed.

And AI is structurally terrible at noticing it.

The model has no idea who you are. It has training data. It has priors. It has a strong statistical opinion about what a business post should sound like in 2026. It will help you sound righter, smoother, safer, cleaner, more legible, more acceptable, more indistinguishable from every other founder who has been accepting the same suggestions at the same quiet rate for the same length of time.

That is not a failure. That is the default.

The question is not whether the tools will get better at sounding personal. They will. They will get very good at it. The question is whether you will still have anything specific enough to say by the time they do. Whether you have preserved the raw, untrained edge that made your thinking yours — or whether you have edited it so often, so politely, so incrementally that the voice that remains is already a very sophisticated average wearing your name.

This is why I keep coming back to the raw note. Not because raw is always better. Raw is not. But the raw note is the last honest evidence of what you were thinking before the averaging began. It is the place where your actual self shows up before the committee of soft suggestions gets involved.

Keep the raw notes. Read from them. Let them remind you who you were before the machine started helping you sound like everyone else.

Not all editing is progress. Some of it is erosion wearing a professional outfit. Some of it is identity loss you never noticed because each step was small enough to feel like a win.

So here is the uncomfortable diagnostic: read the last ten things you have shipped. If they sound interchangeable with what anyone else competent in your field is currently producing, the tool did not help you ship. It helped you disappear — one accepted suggestion at a time.

The average is comfortable because it has no friction. It has no risk. It has no fingerprints the market can push back on. It is the safest possible way to never be noticed, never be wrong, and never be remembered.

Identity is the opposite. Identity is expensive. Identity is risky. Identity is the weird, specific, sometimes awkward choice that a machine would never suggest — because it has never happened in the data that way before.

That is exactly why it is yours.

Do not let the tool edit the part of you that the tool cannot understand.

The thing it would most want to smooth away is the reason anyone ever recognized you.