Image Descriptions via AI

I collect a lot of artwork in Amplenote for future inspiration, but as it currently stands, being able to recall these images is a 100% manual organization process. It would be great to have keyword descriptions automatically generated via AI, as GPT-4 has recently demonstrated. Something verbose like "a fantasy scene depicting an imposing castle at the edge of a cliff against a vibrant green landscape with dragons in the sky" or just keywords like "fantasy, castle, cliff, landscape, dragons". Genre + content is how I classify images but perhaps the user could provide their own AI prompt (with sane defaults provided). My system works for concept art but someone who works in advertising will likely save different types of images and want different results.