"An AI model is incapable of that."
"Good designers will reject this."
^ Famous last words.
I will stand by the first point unless models start being trained with different objectives instead of RLHF's three objectives: Helpfulness, Harmlessness and Instruction-following
I will very likely be wrong on the second point.
[dead]
I could see there being an 80/20-style argument for this sort of tool being used for more generic usecases, with "good designers" using Figma et al. for programs where the UI itself is a selling point.