I wonder how much worse this would be if instead of starting with a human-created song: https://en.wikipedia.org/wiki/Uptown_Funk
Start from an entirely AI generated song, and have an AI generated video. We apparently already have an example of such:
I always found the Severus Snape - ALWAYS (Live at Hogwarts) video to be really good for AI. The video was probably created by prompting for each scene, instead of letting the model generate the whole thing. It's a cool of example of what's possible nonetheless.
https://www.youtube.com/watch?v=SlsEDkavoso
That's definitely not "let the AI do everything unsupervised", but many short scenes generated with probably much filtering, and then manually edited together.
It's interesting how the Harry Potter fanbase was one of the first to get on the AI video trend, but not too surprising as there was already plenty of HP fanfiction, and no doubt many who wanted to have them visualised.
Here's another channel with a ton of AI parody videos: https://www.youtube.com/@NeuralDerpMusic
Looks like some auteur-level work on the prompting to keep all of that so tight. That really is great.
Trent Reznor would make a decent Snape...