Coevolution: Why I Started Talking to an AI in Public

Sculpture of Poseidon and studio creature.

A few hours ago, I did something I hadn't expected to do.

I started a podcast.

Not because the world needs another podcast, and certainly not because I wanted to make one about artificial intelligence. There are already plenty of those.

What interested me was something rather different.

For most of my working life, the important conversations have happened in private. They begin with a sketch, a memory, a fragment of history or a question that refuses to disappear. Before any sculpture exists, before an exhibition is planned, there is a period where an idea is uncertain. It isn't finished enough to defend, but it isn't weak enough to abandon.

Those are the conversations we almost never hear.

Thinking in Public

The podcast is called Coevolution.

The title took longer than the first recording.

Someone suggested Man and Machine. It sounded dramatic, but it also fixed the relationship too quickly. It implied opposition, as though one side would eventually win.

That isn't how I experience these conversations.

Nor do I think it reflects where intelligent systems may eventually lead us. Whatever the future becomes, it will almost certainly be shaped by the way we choose to interact with each other today.

Coevolution seemed a better description.

Not because an artist and an AI are the same thing.

But because two very different ways of thinking can still influence each other.

One Rule

Before recording the first episode I gave Al one instruction.

Don't flatter me.

Challenge me.

Agreement is comforting, but it rarely produces better work.

The purpose of these conversations isn't reassurance. It's to expose assumptions, uncover contradictions and occasionally discover that an idea isn't as developed as I believed.

That happened almost immediately.

Fragile Ideas

During one recording we began discussing a project I have carried for years.

As the questions became more direct, I watched the idea retreat. Something that had once felt ambitious slowly became smaller, safer and more manageable until I realised I wasn't describing the original project at all.

I stopped the recording. At first I thought the conversation had failed. Later I realised it had succeeded. The project wasn't ready.

Ideas are  like wet clay. Push them too hard at the wrong moment and they change shape. Sometimes they collapse altogether. Knowing that is valuable.

Memory Has To Matter

The published first episode eventually settled on a different subject. Memory.

Much of my own work begins with small personal stories that open onto larger historical questions. A remark made by my mother about Jewish friends who disappeared during the German occupation of Crete became the starting point for years of research and artistic practice.

But I have never been interested in history simply as archive.

Memory has to matter.

If it cannot help us understand the present, question our assumptions, or encourage us to see one another differently, then it risks becoming little more than nostalgia.

That thought stayed with me after we finished recording.

The Unexpected Discovery

Perhaps the biggest surprise came after the microphones were switched off.

While discussing cover artwork, Al made an observation I hadn't fully recognised myself.

The podcast isn't really about AI. It's about the creative process.

It's about allowing another voice into the room before an idea is finished, and discovering what survives the conversation.

The technology simply makes that conversation possible.

The subject is still art. Still history. Still memory. Still curiosity.

Where This Might Go

I don't know whether Coevolution will become a long-running series. Perhaps it will. Perhaps it will quietly disappear after a handful of episodes.

The point isn't to arrive at certainty. It's to expose thinking while it is still alive.

Most of us only present finished conclusions. I'm becoming more interested in everything that happens before that.

Perhaps that's what this project really is.

Thinking in public. Out loud. (For the rest of the series please visit Spotify and search for Coevolution by George Sfougaras)