After four years of development, are LLMs capable of writing high-quality prose?

I have been following the evolution of LLMs since the days of GPT-3.5, and I have been actively using local LLMs since the release of Gemma 2. I read a great deal, including this subreddit. I have performed QLoRA...