Art as a Conversation

What generative AI taught me about the collective imagination.

Visual essay · 5 min read

People told me AI would replace artists. So I tried to see if it could replace me.

I wasn’t trying to win. I wanted to know how these tools work, what they show, and where a person’s hand still matters. It started as a technical experiment and turned into a question about authorship and voice.

This is what I tried, in order, and the work that came out of it.

1. Starting with Gracián: could AI find one voice?

I began with a constraint. I fed a model prompts drawn from one book, The Art of Worldly Wisdom by Baltasar Gracián (1647). It’s 300 short maxims in a steady, restrained voice. If any text could hold a model to one voice, I figured it would be this one.

It didn’t. Every maxim came back in a different style. The model treated each prompt as a fresh start instead of part of one worldview.

That was the first lesson. A model doesn’t have a voice. It has probabilities. It remixes Gracián without holding his point of view, and a consistent source text wasn’t going to give it one.

2. Switching to art school prompts

So I switched to the kind of prompts you get in a studio critique. What is art? What is engineering? What does it mean to build something that lasts? Where does the human hand show up in a digital world?

They’re vague on purpose. They leave room to interpret, and the model handled that room well.

A few consistent styles started to show up. The model still didn’t have a voice. I was shaping one, prompt by prompt, pushing it toward something that hung together. That’s when it started to feel like a collaboration.

3. Picasso vs. Duchamp: asking AI to finish an argument

Then I gave it an old argument to finish. Picasso made art for the eye. He’s often quoted saying that painting is a blind man’s profession, that he painted what he felt rather than what he saw. Duchamp moved art from the eye to the mind. His readymades, like Fountain, were ideas first. Thierry de Duve writes about that split in Kant After Duchamp (1996).

I asked the model to keep the argument going, and it did. The images it made pulled from both sides.

That’s where it got uncomfortable. The model rendered better than I do, but every image leaned on my story to hold it together. So where was the art: in what the model drew, or in the conversation I was having with it? I was living the same argument I’d asked it to paint.

4. AI as a lens on the collective

I don’t think AI is conscious. I do think it’s collective.

Pierre Lévy called the internet a collective intelligence (1997): knowledge spread out, shared and recombined. Kate Crawford’s Atlas of AI (2021) shows how these systems are built from what they take from people, and how they hand our patterns and biases back to us.

To me a model works like mycelium, a network running under billions of things people have made and posted. When I prompt one, I’m reaching into that pile, everything we’ve ever uploaded.

That’s why I think of this series as made from the collective imagination, with AI as the lens I used to look at it.

5. Why I’m sharing this work

I think art is a conversation: between the artist and the viewer, the past and the present, the idea and the thing made. Generative AI adds more voices to it. It lets me hear what comes back when billions of people’s work overlaps, and it makes me say what I mean more clearly, because it takes me literally.

This series is my try at showing that.

What it came down to

All of that was in my head. The real conclusion came from my life.

My uncle was an artist. We went to the same school, and we saw things the same way. Then he had a stroke, and the realistic, careful style he’d always drawn in was gone. What came out instead was abstract, bright and raw. It wasn’t worse, just different: a new language his body made him find.

Years later I burned out badly enough that it rearranged what I could make. Styles that used to come easily felt locked away. I learned that being able to make art isn’t fixed. It can shift, disappear and come back different.

Working with AI brought all of that back up. Art has always had gatekeepers who cared more about the product than the person: technique over voice, output over experience. The person behind the work, their body and their history, mostly didn’t count.

AI doesn’t replace artists. It does separate expression from physical ability. Someone who can’t hold a brush anymore can still shape an image. Someone whose hands or style have changed, from illness, injury, age or burnout, can keep talking in pictures.

That’s what this series is about for me. More people get to make things, and art gets bigger when they do.

THE SERIES

Gracián Experiment Series · Studies of Form, Function, and the Creative Process · The Art of Duchamp · The Art of Picasso · Picasso vs Duchamp · Expanding on What We Call Cubism · Expanding the Conversation · Liberation Through Expression