There was a lot of discussion about AI art on Reddit's r/generative. We've banned AI creations (I'm a mod there) since we treat most of them as effortless. The flood of simple prompt->image posts was enormous. Search for tobyski posts for further explanation.

My personal opinion is: generated creations by ready to use tools (which I consider AI for example) is not the same as generative creations.

Thanks! Definitely agreed that simple prompt to image creations are not generative in the sense that we usually mean. At the same time, I also think that generative AI systems open up a lot of interesting new possibilities that are generative in that sense, for example writing custom code that does a guided series of recursive transformations, where each step is performed by the generative AI, trying to guide it toward realization of some particular result or aesthetic. Or code that explores the prompt space in search of the prompts that produce results with highest information density as measured by compressibility.

(and it's not clear to me that the copyright offices current guidance would successfully distinguish between those cases)

If human effort is the measure, maybe recording/documenting the creative process could provide enough evidence that there was “significant creative input” by the artist? Did not read the article (only the twitter thread), but these quotes: “If you manipulate the generation significantly with Photoshop or other tools […] it's valid for copyright.” and “AI will be considered a tool in the arsenal of human authors” Sound to me like AI could as well be such a tool in a process similar to what you described, as long as it is not just prompt→image, but more conceptual thinking/effort behind it. Still vague, however, what is considered the threshold.