There is a conceptual dialectic in generative art created through code: when an artist cannot fully control the outcome but is still responsible for it, they enter a state of responsibility without sovereignty.
This leads to an interesting tension. An artist cannot claim every unexpected result as an expression of their own creativity, because those results were not individually conceived by them. Yet when the system produces bias, harm, or unacceptable consequences, the artist cannot simply say, “the machine did it.”
So can the resulting work still be considered their work?
Once we shift the concept of creation from “creating an object” to “creating a set of rules,” the identity of the author is fundamentally redefined.
We can examine this idea through another example. In 1959, Hungarian computer-art pioneer Vera Molnár did not yet have access to a computer. Instead, she developed a concept she called the “Machine Imaginaire,” or “imaginary machine,” as a method for making art. She would first establish a set of rules, then require herself to execute them as if she were a machine, without making spontaneous changes based on intuition. In other words, she transformed herself into the computer.
Earlier, when discussing the idea of an artist creating “a set of rules,” the ambiguity over authorship emerged partly because the computer was the one executing those rules. This is what leads us to ask: Who actually created the work, and whose work is it?
But in Molnár’s case, the rules were created by a human and executed by a human. Does this change how we assign value to the same process when the rules are instead executed by a computer?
Looking at these two different models, the core of code-based generative art is not whether the creator can fully control the work, nor is it simply a question of who performs the execution. What code-based generative art introduces is a mode of creation in which the formulation of rules and their execution are separated.
Historically, this separation has been controversial within art.
Likewise, if future AI systems become capable of reproducing the psychological state of a human author, I think the question may resemble the debates surrounding code-based generative art. The point would not be to deny existing forms of authorship or to compete for the status of “original creator.”
It may simply represent an entirely new creative relationship, one that has not yet been fully defined.