We Are All Untrained Data
Pretty much the idea for this site.
By Richard Udell
Published April 7, 2025
In machine learning, untrained data is raw input. It hasn’t been labeled. It hasn’t been processed. It hasn’t been shaped into something a model can learn from. It’s chaotic, full of contradictions, potential, and noise.
Then there’s trained data—data that’s been cleaned, categorized, and fed into a model so it can predict, complete, or classify with confidence. Trained data is what machines run on.
But we’re not machines.
We don’t just feed the system. We break it. We overflow with contradictions. We’re so untrained we would throw an error in the model.
The things that define us as human—intuition, emotion, instinct, wonder—are fundamentally untrainable. They’re unrepeatable. Unreliable. Alive.
And that’s what this essay is about.
What if our humanness isn’t defined by how productive, rational, or efficient we can be—but by how fundamentally untrainable we are?
AI is moving fast. Like fast fast. And most people either don’t realize it, or don’t know how to react.
Everyone’s saying “learn the tools,” and yeah—do that. But what are we really preparing for?
Because here’s the thing:
AI can now write better than most people. It can code faster, explain things clearer, summarize your thoughts before you even finish them.
It’s always on. It doesn’t get tired. It doesn’t doubt itself. And it’s only getting better.
So what does that make us?
It’s easy to feel like a crisis. But maybe it’s something else. Maybe it’s a chance to see more clearly.
Every time a machine learns how to do something we thought made us special, it asks us—what’s really at the core of being human?
Sometimes that’s uncomfortable. Sometimes it’s beautiful.
Because the more humanlike these systems become, the more we’re haunted by what they still lack.
They don’t wonder. They don’t laugh out loud by surprise. They don’t get chills from a memory. They don’t have untrained—or untrainable—data.
We do.
We are full of it. Messy, nonlinear, soulful, unoptimized input. Half-formed questions, weird connections, jokes that don’t quite land. Gut feelings, contradictions, late-night voice memos.
We are all untrained data.
That’s not a flaw—it’s the point.
We’re not here to compete with the machine. We’re here to remember what it means to be human in its presence.
So what do we do?
We amplify the untrainable parts. We stay weird. We stay soft. We make things that can’t be templated.
We write essays. We go dancing. We take film photos and road trips and do a star jump just because it feels good.
We visit our parents. We ask questions AI can’t answer—not because they’re too hard, but because they’re too human.
The future is here. But so are you.
And the untrainable data inside you? It’s not just valuable. It’s irreplaceable.