It wasn't the point, but early LLMs would make mistakes so often that writing fiction full of hilarious nonsense was one of the only applications. Now that they make mistakes less often, you need some other source of high-entropy randomness to reinject creativity into the story.
Kinda, more accurately the current goal for labs is to produce economic value to justify high valuations / capex, thus having to RL on correct answers rather than interesting ones.
If you could access the current models as base models I guess you'd still get very interesting hilarious stuff.