What if we were generic all along, and AI only proved it?

There's a particular thought that shows up around eleven at night, usually after you've pasted something you wrote into a chatbot to "tighten it up," and the thing it hands back is as good. Maybe better. It took four seconds and no anguish, and you sit there looking at it with a feeling you can't name, something closer to exposure than anger.

The thought is: what if I'm not as good as I thought I was.

I found the sharpest version of that fear in, of all places, a Reddit thread. Someone in a personal-branding forum wrote two lines that made me stop cold: "AI didn't make people sound generic. It just revealed they were generic to begin with."

I read it again. It felt even worse.

Because the comforting story we tell is that the machine flattened us. That our voices were distinctive and AI came along and averaged everyone into the same beige pasty slop. But this person is suggesting something more brutal. The machine held up a mirror and showed a lot of us that the work we were charging for, the thing we quietly believed only we could do, was more replaceable than our self-image allowed. The specialness was already thinner than we advertised, and the AI shined a spotlight on it.

There's a hopeful end to this, but we can't jump ahead until we look at the difficult part.

If you've spent fifteen or twenty years getting good at something, and a model can now approximate it on the first try, that's disorienting. It calls your next invoice into question, and then your story about yourself. The competence was supposed to be the hard-won thing, the part nobody could take. And it turns out a lot of competence is just pattern-matching done well, which is precisely what these systems are built to do.

So let's not pretend that doesn't sting. It does. I've felt it.

It took me a while to see what the fear gets wrong. It assumes there was only ever one kind of value, and there were always two.

The first kind is scarcity. You were worth paying because you were one of the few people who could do the thing. You could write the clean paragraph, build the deck, cut the spot, render the comp. Not many people could, so they paid you. Scarcity is a real source of value and it was always, always going to lose this fight. There was always going to be someone younger and cheaper, and now there's something faster and cheaper than that. If your worth was built on being one of the few who could, the machine was eventually coming for it, and the only surprise is the timing.

The second kind is specificity. You're worth paying because nobody else would do it the way you do. That's particularity of person, and it turns out to be the thing the model can't reach. A fingerprint takes a hand, and the machine doesn't have one.

I learned that from the worst job application I ever sent. Years ago, early in my career, I was broke and demoralized, and every application I sent disappeared into a black hole. This was before the systems that auto-screen you out; a human being was ignoring me, which somehow felt more personal. After enough silence, I stopped trying to be a strong candidate. Out of pure frustration, I sent one application with a haiku where the cover letter should have been. I don't remember a single word of it. I do remember thinking, this is a stupid thing to do, and sending it anyway.

It got me my first real role in advertising.

For years I filed that under "got lucky." Lately I read it differently. In a stack of applications that were all competent and all interchangeable, the haiku won on specificity. A better cover letter would have been lost in the same pile as all the other better cover letters. Anybody can write three lines. Nobody else would have written those three lines, in that order, out of that exact bad week. What landed on that desk was a person, legible on the page, in a place where everyone else had sanded themselves smooth.

I'd spent the year before that working for someone who performed authenticity for a living, watching the performance work right up until it didn't. So maybe the haiku was the opposite instinct finally surfacing. What I had was a specific mood, a weird idea and nothing left to lose.

The way through is being more particular than the machine can fake. Race it at the generic thing and you'll lose, and so will I. The reference only you would make. The detail that lands because you were in the room. The opinion you hold for reasons that took you twenty years and one bad boss to arrive at. AI produces plausible in enormous quantities. Particular requires a life, and it doesn't have one.

Which means the Reddit comment, the one that felt like a verdict, is a gift, if you let it be one.

If the machine exposed that part of your work was generic, then now you know which part. You know which of your skills was scarcity, on loan from a bank that was always going to call the loan, and which was specificity, the stuff with your fingerprints pressed into it. The generic part was always rented. You can stop grieving it.

What's left when you subtract everything a model can do on the first try is small, and strange, and particular, and entirely yours. That was always the job. Most of us couldn't see it until something showed up that could do everything else.