Well, LLMs aren't just randomly spitting out garbled text either. Humans create and curate the training data, humans design the models, and humans train them.

sure, but humans have also crafted LLMs

Yes, but then again, completely no. I mean, a procedural generation algorithm for a game world is an artwork tuned to produce results the artist likes, usually with care to allow some cases and exclude others and meet a particular vision. It's not very different from a pre-made world, except using the RNG as one of its building blocks.

Same can be said for prompts. What's the difference from a pleasing underground mine cave labyrinth and some weird Pulp novel plot?

True, so I went and looked for the prompts:

https://github.com/EQ-bench/creative-writing-bench/blob/main...

Not clear on how these elaborate prompts relate to the very short prompts in the samples.

Different benchmark, those are for the short form creative writing leaderboard here: https://eqbench.com/creative_writing.html

>tuned to produce results the artist likes

Kinda like RLHF?