How do we credit hybrid images?

Around the corner from here, over at 3 Quarks Daily, I’ve published an article I wrote in conjunction with both ChatGPT and Claude. How should that article be credited? How do we characterize the contribution of each agent and how do indicate that characterization? I discuss these issues at the end of the article.

Same issues can arise with visual images. All of these images were rendered by ChatGPT. But the renderings were done on a different, a different what? Basis? Substrate? Seed?

In the first two images, I uploaded a one of my photographs to ChatGPT and asked it to add something to it. In the case of first photo, I wanted to see the Millennium Falcon flying into the iris. The second photo is of a scene in Liberty State Park into which I had ChatGPT place a photo of an Indian woman in a sari eating McDonald’s French fries.

This image is a bit different. I gave ChatGPT a photo of a scene in Jersey City and ask it to turn it into a futuristic scene.

For this image I gave ChatGPT a photo of a painting I’d done as a child and asked it to render it in the style of Hokusai.

In this last case I gave ChatGPT a document that I wrote and then asked it to create an image that would be an appropriate frontispiece for it. This image is quite different from the one it originally produced. I had to do quite a bit of art directed to obtain this final image.

The question then is: Imagine that these images were on display in, say, a museum. How should they be credited? In all cases the final image was rendered by ChatGPT. But the substrate varied as did the prompting which instructed ChatGPT in generating the image. For example, in the first four cases we could indicate “Original photograph by William Benzon. For the last, “Original text by William Benzon” and “Art Direction by William Benzon.” Do I give myself an art direction credit on the others as well? What kind of credit should ChatGPT get. “Realization and Rendering by ChatGPT” might be sufficient for the first two. For the third and fourth, “Transformation and Rendering.” The last? Perhaps “Transmutation and Rendering.” Whatever the nature of the credits, they’re only meaningful if the audience already knows something about the process through which they were produced.