We’ve built systems that write like us and we’re never going back. Why would we? Now we need to learn how to write like ourselves again.
The paradox of AI-generated text isn’t that it sounds robotic; it’s that it sounds almost right. It’s Competent. Indistinguishable from the kind of professional prose we’ve been trained to produce. The problem isn’t that AI writes badly. The problem is that it writes exactly the way we’ve been taught to write when we’re trying to sound professional and safe.
Which means the thing we’re calling “AI voice” might actually be our own safe voice, reflected back at us. The patterns we recognize as algorithmic are the conventions of institutional writing we’ve internalized so deeply we don’t notice them anymore.
Until now.
The craft of creating your authentic voice starts at a metacognitive layer where you’re thinking about how you think, and then thinking about how you translate that thinking into language. You have a thought, you reach for the words that seem to fit, you arrange them into sentences that sound reasonable. There’s a lot of room for nuance and personality and preference, bias, and so on — the distinguishing characteristics that make up how your ideas are translated into, words. And yes, I added the em-dash intentionally.
The process is more art than science. It’s important to be able to recognize the patterns and see the conventions that are overused, and perhaps too conventional and moreover, unrelated to the choices you would have preferred to craft a voice that consistently preserves your own structures, tone, and language.
This is the skill AI has made essential. The question isn’t whether you can write. The question is whether you can write like yourself.
If you can, you can do it consistently with AI.
