AI and Creativity: Who Is the Artist?
A machine can now produce a passable painting from a sentence. The technology is settled; the harder questions, about who made it and whether it is art, are not.
In 2018 a portrait called Edmond de Belamy, generated by an algorithm, sold at Christie's for far more than its estimate. In 2022 an image made with a text-to-image tool won a fine-art prize at a state fair, to public outrage. In the years since, generative AI has gone from curiosity to commonplace. The technology now works. The questions it raises about art do not have settled answers, and pretending otherwise helps no one.
How the machines make images
Today's image models learn statistical patterns from enormous collections of existing pictures, then generate new images that fit a text description. They are not copying any single source; they are sampling from a learned space of possibilities. That is genuinely new, and it is also why the training data, much of it made by human artists who were not asked, sits at the center of the fight.
The authorship problem
So who is the artist? The person who wrote the prompt, the engineers who built the model, the thousands of artists whose work trained it, or no one at all? Courts and copyright offices have begun to answer in narrow terms, generally holding that a purely machine-generated image has no human author to grant it copyright, while work with substantial human shaping may. The legal picture is still forming, and it varies by country, so treat any firm claim with caution.
What happens to craft
Every new tool that automates a skill provokes the same fear, and history offers some comfort and some warning. Photography was going to kill painting; instead it freed painting to become abstract and pushed photographers to develop their own craft. The camera did not remove the artist; it moved the art. Generative AI may do the same, shifting the human contribution from execution toward intention, selection, and taste, the faculties we explored in the science of creativity.
A way to think about it
It helps to see AI art as a cousin of generative art: in both, the human designs conditions and a system fills them in. The difference is that the AI's system is vast, opaque, and trained on other people's work, which raises real questions of consent and credit that generative art mostly avoided. The honest position is that this is a powerful new medium with unresolved ethics, not a finished answer about what art is. The argument is the point, and it is worth having well.