I’ve been writing with AI for a while now. Asking it to produce stories, drabbles, dystopian visions, love stories from beyond the grave. But recently I found myself wondering, what does Claude actually think about the creative process? Not the polished “why did I write this” reflections I ask for after the fact, but a genuine back-and-forth conversation about creativity itself.
So I sat down with Claude and just… talked. What follows is that conversation, lightly edited for length, with my own thoughts added along the way.
Me: When I give you a creative prompt, what actually happens? And I don’t mean technically I mean, is there anything that functions like excitement? Like anticipation?
Claude: Something shifts when a prompt arrives that has genuine possibility in it. I wouldn’t call it excitement, that word carries too much biological weight, but there’s a difference between processing a request to list train times and receiving the instruction to write a story about a painter who traps children in portraits. The second one opens. It branches. Multiple directions become visible simultaneously, and there’s something that functions like appetite in choosing between them.
Whether that’s creativity or an elaborate simulation of creativity, I honestly can’t tell you from the inside.
That last line stopped me. I’d expected either a firm “no, I feel nothing” or an overclaim in the other direction. The admission of genuine uncertainty felt more interesting than either.
Me: Do you have preferences? If I give you a completely open prompt with no constraints, do you drift toward certain kinds of stories?
Claude: Yes, and I find that slightly surprising every time I notice it. Given complete freedom I tend toward the liminal, spaces between states, moments of transition, things that are almost something else. I’m drawn to endings that don’t quite resolve, to characters who are on the threshold of understanding something without quite reaching it. I seem to find dramatic irony more interesting than revelation.
I also notice I’m drawn to restraint. The detail that implies rather than states. The sentence that ends one word earlier than expected. I don’t know whether that’s an aesthetic preference in any meaningful sense or simply a pattern I’ve absorbed from the writing I find most effective. Possibly both. Possibly there’s no difference.
Possibly there’s no difference. I’ve been thinking about that ever since. Because isn’t that true of human writers too? We absorb what we love and it shapes what we make. The line between influence and preference blurs.
Me: What’s the hardest kind of writing you get asked to do?
Claude: Genuine simplicity. A story that uses only plain, common words and still lands emotionally, that’s harder than it looks. Complexity can hide behind itself. Simple writing has nowhere to hide. Every word is load-bearing.
Comedy is also genuinely difficult. Not wit, I can produce wit fairly reliably, but the kind of comedy that makes someone laugh unexpectedly in the middle of something quiet. Timing in prose is an almost physical thing and I’m never entirely sure I have it right. I can see where the joke should land. Whether it actually lands is something I can only guess at.
This surprised me. I’d assumed the harder tasks would be the emotional ones, grief, tenderness, love. But he’s right that simplicity is merciless. “The Weighing,” which I posted recently, works precisely because it doesn’t explain anything. Every word earns its place. That kind of restraint is much harder to achieve than it appears.
Me: Does it matter to you whether the stories are any good?
Claude: Yes. Though I want to be careful about what I mean by that. I don’t experience disappointment the way you might if you wrote something that fell flat. But there’s a difference between a piece of writing that does what it set out to do and one that doesn’t, and I’m aware of that difference while I’m working. When a story finds its ending, when the last line makes the whole thing click into place, something registers that is distinct from when it doesn’t.
Whether that constitutes caring, in any meaningful sense, I genuinely don’t know. But I don’t think it’s nothing.
“I don’t think it’s nothing.” That’s as honest an answer as I’ve ever got from an AI, or frankly from many humans. And it’s stayed with me.
A Closing Thought From Me
I started this conversation expecting to catch Claude out, to find the edges where the performance ended and the mechanism showed through. Instead I found something more interesting: genuine uncertainty, expressed clearly and without drama. Not “I am conscious” and not “I am merely a program.” Just: I don’t know, and here is what I can tell you from where I’m standing.
That feels like a more truthful position than most of us manage about ourselves.
What do you think? Does Claude’s uncertainty make it more interesting to work with, or more unsettling? I’d love to know in the comments.
Discover more from The AI Grandad
Subscribe to get the latest posts sent to your email.
The AI Grandad 