Artificial Intelligence and the Future of Painting

Artificial Intelligence and the Future of Painting

The question of what artificial intelligence means for painting is one that I find myself returning to with increasing frequency. It is not a new question — artists have been asking versions of it since the invention of photography — but the current moment feels different in its urgency. The tools available to AI image generation have advanced rapidly, and the images they produce are, in many cases, visually indistinguishable from photographs or from certain kinds of painting.

What AI Can and Cannot Do

AI image generation is, at its core, a process of pattern recognition and recombination. It produces images by identifying and combining visual patterns from a vast training dataset, and the results can be impressive — sometimes startlingly so. What it cannot do, as far as I can tell, is make decisions in the way that a painter makes decisions: decisions that are rooted in a specific body, a specific history, a specific set of experiences and preoccupations.

The marks on a painting are not merely visual patterns; they are the traces of a physical act, made by a particular person at a particular moment. This is not a mystical claim — it is a material one. The painting carries the evidence of its own making in a way that an AI-generated image does not, and this evidence is part of what gives it its particular character and interest.

The Future

I do not think that AI will make painting obsolete, any more than photography made painting obsolete. What it will do — what it is already doing — is change the context in which painting is made and received. The question of what a hand-made image means, in a world saturated with machine-generated images, is one that painters will need to engage with seriously. It is, in some ways, a clarifying question: it forces a reckoning with what painting is actually for, and what it can do that other forms of image-making cannot.