These image-from-prompt generators are themselves just a particular type of generative art.
I think it's obvious that the output of these AIs can be considered art. There is human input at multiple levels - the code, the artists whose works it was trained on, the researchers that scraped/pruned the training set, and ultimately the crafting of prompts by users to achieve a desired output.
What is not obvious is: among those human inputs, which are actually responsible for imparting artistic value on the resultant image, and in what proportion.
I think it's obvious that the output of these AIs can be considered art. There is human input at multiple levels - the code, the artists whose works it was trained on, the researchers that scraped/pruned the training set, and ultimately the crafting of prompts by users to achieve a desired output.
What is not obvious is: among those human inputs, which are actually responsible for imparting artistic value on the resultant image, and in what proportion.