Maybe there are some nooks and crannies where this tech can be employed and actually save time/budget. Maybe. But what does not work is to just take the output and use it. There is nowhere near enough precision there.
If you think you can be "specific" enough for AI to match your "creative vision", then maybe you're not really creative.
I have no objection to the potential vision you espouse, I have zero love for the actual ground level reality of the tsunami of AI slop videos being pumped out ATM - many are skewed toward trust building in a probable long con game of deceit or nation swaying, others are just ick.
An app is still useful even when it's cheap for anyone to make it. But art that anyone can make for cheap becomes, by definition, slop.