Thoughts on clarifying Your Process Before Tuning Models for Computational Creativity

When I work with Artificial intelligence platforms as part of my artistic practice, the challenge is rarely just technical. I make moving images based on my paintings and bring them into augmented reality environments. But before I even start tuning or training a model, I have to get very clear about what I want—because without that clarity, the technology can easily distort my vision.

Many generative models are not neutral. They carry built-in biases—toward certain racialized features, toward Western or Eurocentric aesthetics, toward specific visual “styles” that can overwhelm or erase the source material. Sometimes I’ve seen my work transformed into something unrecognizable, aligned more with the model’s defaults than with my aesthetic choices. If I’m not careful, I can lose the thread of my own artistic language.

This is why personal process comes first. Before tuning a model, I need to pause and articulate:

• What do my paintings actually feel like?
• What themes, textures, or shapes recur across my body of work?
• How do I describe my color palette, my relationship to movement, or the emotional tone of my images?

It’s not enough to “know” these things intuitively. Training a model forces me to put them into language, into data, into repeatable patterns. That act of articulation is its own artistic exercise. If I can describe my work precisely enough to guide or train a machine, then I’ve deepened my mastery of my personal aesthetic.

In this way, AI becomes less of a threat to artistic integrity and more of a mirror: it reflects back the strength—or vagueness—of my self-definition. If I am unclear, the outputs will be inconsistent or colonized by the model’s defaults. If I am clear, the outputs can extend and evolve my practice in powerful ways.

So for artists working with computational creativity, here is the paradox: the machine requires you to define yourself, but the process of defining yourself also strengthens your art. Before tuning a model, tune yourself.