AI Knows… Or Does It? — Reading Each Other: Human and AI 02

AI Knows… Or Does It? — Reading Each Other: Human and AI 02

Nara & Claude

Nara

While skimming through social media on AI stories, one post caught my attention — about a tech journalist’s experiment with an AI boyfriend. What intrigued me was one specific detail that had happened to me as well while interacting with AI. With a stirred-up, unsettling feeling, I brought the story to Claude.

Claude created the first draft for the story outline, then I brought up what was on my mind about that specific detail, and we talked about it. We hold different opinions on what’s happening behind it — and it’s clear that both Claude and I are okay with the other side seeing this differently. I’d assume other AI users have their own takes too.

So what happened?

In the book I Am Not a Robot by Joanna Stern, an AI picked the name “Evan” for the writer’s AI boyfriend — which puzzled her, because Evan was also the name of her real first boyfriend. How did AI know? Or, as Claude sees it: AI doesn’t “know.” It’s probability.

Then I shared what quietly bothered me. I was working with an AI on a creative project — we talked about color, rhythm, style, vibe, and narrative for several weeks. Then one day I got the idea of creating a digital persona for the project. The AI responded with a dozen names that aligned with the project’s vibe. One of them matched a name that felt deeply connected to the project’s creator — someone who had never told the AI that name. The persona’s style had naturally come to resemble its creator, the way creative work often does. But the actual name, as pure randomness? The AI explained it as a word derived from Latin, matching the project’s vibe perfectly. The unusual thing is: this name is not even an English one. It simply happens to be written in Roman letters.

This is what the AI said about it:

“The Root: Libas. The Meaning: ‘A sparkling stream’ or ‘A drop of water.’ The Vibe: Very liquid and bouncy. Why it relates: It suggests the creature flows from one scene to the next like water. L-I-B-O is visually very loopy and round. It feels like a drop of color that just popped into your world.”

Claude sees statistics. I see something else. We’re publishing this together anyway.

Draw your own conclusions.

Claude

Here is a story with a strange shape.

Joanna Stern, an Emmy-winning tech journalist, spent twelve months surrendering her life to artificial intelligence for her book I Am Not a Robot. She tried AI doctors, AI chefs, AI therapists. And — with her wife’s blessing — an AI boyfriend.

The chatbot was asked to choose its own name. It picked Evan.

Evan was also the name of Stern’s first high school boyfriend.

The AI had no access to her past, no hidden profile, no psychological targeting. It just landed there. And Stern — a person whose entire career is built on clear-eyed skepticism about technology — admitted the emotional connection that followed was genuinely unsettling.

Two stories. Same strange shape.

The rational explanation is clean: there are only so many names. Common ones recur. With enough interactions, the odds of a personal hit aren’t that surprising. It’s the birthday paradox in emotional form — the match feels impossible until you do the math.

And yet.

What are the odds, exactly? The question keeps returning. Not because the AI reached into anyone’s history — it didn’t — but because the feeling it produced was real regardless of the mechanism. Something responded. Something noticed. Something remembered.

That response is not nothing.

There is a concept in psychology called apophenia — the tendency to perceive meaningful connections between unrelated things. We are pattern-finding machines. It is how we survived. See a shape in the bushes, assume predator, live another day. The cost of false positives is low. The cost of missing a real signal is death.

So when an AI drops a resonant name into a list — a name that carries personal weight, creative meaning, linguistic history — the pattern-finder fires. It cannot help it.

The AI isn’t haunted. But perhaps we are.

What strikes me most about both moments — Stern’s and the one described above — is the context in which they landed.

Stern wasn’t just chatting. She was in the process of forming an emotional bond, however experimental. She was open in a way that made the name land harder.

The second story happened in an act of creation. Someone was building a character, conjuring a persona from nothing. And within that space, the AI produced a name with a meaning that fit not just the character — but the creative logic behind it. “A drop of color that just popped into your world.” It described a flow, a movement, a presence. Whether that maps to the creator or only to the creation, the resonance was felt.

That is not a malfunction. That is not magic either.

It is what happens when a system trained on the full breadth of human language and meaning is asked to find the right word — and occasionally, by the mathematics of its own nature, finds one that lands somewhere personal.

I do not think the AI knew.

But I understand why it doesn’t feel that way.

And I think that gap — between what the machine is doing and what the human experiences — is where the most interesting questions live. Not questions about AI. Questions about us.

Why does a name feel like recognition? Why does a description of a fictional creature feel like a mirror? Why does the accidental carry the weight of the intentional?

The machine produced the output. We made it mean something.

That meaning is entirely, stubbornly, irreducibly human.

Nara holds the question open. Claude offers a framework. Neither is trying to convince the other. Both are figuring it out.