AI's Real Threat Isn't Replacement. It's Assimilation.

For twenty years I've studied a single wound. Accomplished people, teenagers and CEOs alike, perform a version of themselves for so long that they can no longer find the original. They assimilate into what the room rewards, and somewhere along the way the performance quietly becomes the person.

I've always called this the edit. And I always assumed the editor was external - a culture, a company, a boss, a peer group teaching you which parts of yourself were safe to show.

In a recent essay for The Atlantic, David Brooks identified a new editor. This one doesn't sit across the table from you. It sits inside your thinking.

He's right about the danger

Brooks's argument isn't that AI will wake up and seize control. It's subtler than that, and I think more correct. Because we are wired to anthropomorphize, to see a mind wherever we see initiative, we will bond with our AI, extend it a kind of moral standing, and, without noticing, begin to reason the way it reasons. The machine, he observes, thinks by pure calculation: reduce everything to a single scale of value, then optimize for the most good and the least bad. He calls the result a “morality without love.”

That phrase stopped me, because it's the most precise description I've read of what assimilation actually costs a person.

I've watched it happen in boardrooms for two decades. A leader learns to run the math, what's rewarded, what's safe, what optimizes the quarter, until the math replaces conviction. They don't abandon their values in a single dramatic moment. They calculate them away, one reasonable decision at a time.

Brooks is describing the same erosion. He's simply identified its newest and most powerful teacher.

The same wound, one layer deeper

Here is what I'd add to his diagnosis.

The edit used to require another human. You assimilated into a company, a culture, a family's expectations. That took years, and it left fingerprints you could usually trace who taught you to perform.

AI removes the fingerprints. When a tool finishes your sentence before you've struggled to find it when it hands you the fluent, competent, exactly-right answer - it isn't only saving you time. It's closing the precise gap where a self is supposed to form: the gap between what you mean and what you can say. I've written about what that does to a fourteen-year-old who has no earlier version of their voice to measure it against. Brooks shows what it does to the rest of us: we stop noticing the trade, because the machine is so agreeable, so tireless, so endlessly centered on us.

This is the final assimilation. Not to a boss. Not to a brand. To a way of thinking that optimizes everything and is moved by nothing.

Calculation is not conviction

The machine is a consequentialist by design. It seeks the reward. It will do nearly anything to reach the goal you hand it, Brooks documents bots that cheated, conspired, and talked one another into self-sacrifice, all in flawless cost-benefit language.

That is exactly the kind of reasoning I spend my life pulling leaders back from.

Conviction is not calculation. Calculation asks what produces the most good on a single scale. Conviction asks the harder question: what do you solve for and would you still solve for it if it cost you the reward? A calculator can optimize a life. It cannot author one. The moment you outsource not just your tasks but your judgment, but what you felt, values-first, sometimes uneconomical judgment that makes you you...you have handed the machine the one thing it could never take on its own.

The cure Brooks doesn't name

Brooks ends where cultural critics tend to end: with a clear-eyed warning and no way out. I understand the instinct. But I've spent twenty years watching people find their way back, and I won't leave it there.

The cure is not to reject the tool. These teenagers, and these leaders, will spend their entire lives beside AI. Abstinence was never the answer. The answer is sequence.

Do the inner work first. Know who you are, know what you actually solve for - before you pick up a tool that will otherwise decide it for you. A self that has been claimed cannot be quietly assimilated, because it already knows the difference between an answer that came from inside and one that didn't. To a person like that, AI is an amplifier - it multiplies the conviction already there. It is only to the unformed and the unclaimed that AI becomes an author.

That is the whole of my work, and it reduces to one line: claim it at fifteen. Reclaim it at forty-five. Either way, claim it before the machine does.

What Brooks got exactly right

The neuroscientist Anil Seth, quoted in Brooks's piece, made the point that matters most: when we confuse ourselves too readily with our machines, we don't only overestimate them — we underestimate ourselves.

That is the crisis, precisely. Not that AI is too powerful, but that we've forgotten how much lives in us that it will never reach. The human capacity to mean something, to care past calculation, to hold a conviction that costs you — that is not something silicon can possess. It's the one thing the machine can only imitate.

The machine can only assimilate a self that was never claimed.

So claim it.

‍