The Artificial Hivemind

Had an interesting conversation last night with someone who's spent 20 years in persuasive, high-stakes writing. The kind where funded vs. rejected comes down to voice. (Lots of great convos at NextGEN's SUDs event!)

They don't want to use AI. And I get it. When you've spent two decades developing instincts for what works, the idea that a machine can just do it feels like it erases all of that.

But I encouraged them to start. Not because it'll replace what they do, but because they need to understand what they're competing against.

Researchers at the University of Washington recently studied how different AI models respond to the same prompts. They called it the "Artificial Hivemind." Not only does the same model give you similar answers every time. Different models give you similar answers to each other.

So when someone dumps a generic instruction into ChatGPT and ships whatever comes back, they're submitting the same thing as everyone else who did that. Literally overlapping output.

AI is fantastic at the grindy work. Fill in the blank sections. Boilerplate. First drafts of things that need to exist but don't need to sing.

But the person reading the output? They're drowning in sameness. And they can feel it, even if they can't name it.

But what happens when a human isn't even the first one reading it? They told me that more and more, what they write gets reviewed and evaluated by AI before a human ever sees it. So now you have AI-generated content being judged by AI. If every submission scores within 1-2% of each other, it doesn't take much to stand out. A little genuine voice, a little human instinct, and you're the one that gets pulled from the pile.

Your taste. Your instinct for what matters, your feel for what the reader actually needs to hear. That's the thing AI can't generate. That's what makes the difference between something that gets pulled out of the pile and something that disappears into it.

The people who'll win with AI aren't the ones who use it the most. They're the ones who know when to override it.