My issue is with what needs to be in the artist’s control to qualify. In the Sarony case, he’s working to dress, pose and get an expression from Oscar, and the only contribution from the machine is in capturing the moment. In the case of an AI, the creator provides an idea as a prompt (Oscar Wilde in a cowboy hat, say, or the current Congress as an Hieronymous Bosch paitining), and the AI provides an answer.

Worse, the AI is providing an answer that it’s reproduced and rearranged based on other works that it has been fed, like headshots of House members and Bosch’s oeuvre. “Generative” is kind of a misnomer. It looks more like a mechanical license or a cover: Hendrix singing Dylan’s All Along the Watchtower as a power rock song, rather than an acoustic folk masterpiece.

To me, an AI prompt is like a detailed plan for a fox hunt: the user plans the location, the outfit and the gun. But she never has to travel to the beachhead to capture something new.