You're thinking about this all wrong.

Look at some of the projects coming out of Google research to see where this is going:

Phenaki combines LLMs to generate video sequences

Combine that with something like Wordcraft which allows LLMs to retain consistency for much longer form writing (like entire scripts), and you can start to generate long form synthetic videos.

Those on their own are an interesting product already: being able to ask a bot like chatGPT to make you mini movies on any topic that you can edit simply through prompting like chatGPT?

but take it even further: combine that with an algorithm like tiktok's recommender & Invert it as a reward model for RLHF like openAI does with their davinci models and you also then start getting ways for AIs to automatically generate new content that people like, and recommended them to people. A lot of people scoff at tiktok, but it's enough of a scare to Google that YouTube had to copy it.

I think synthetic media can revolutionise how we consume content: Holodecks are not outside the realm of possibility anymore