So far “telling a satisfying and well-written medium-length story” has proved far harder for LLMs than mathematical proofs, music generation, research reports, code, and many other forms of work. The technical reasons are pretty clear, but they are supposed to be language models

One big reason is that it is hard to verify really good writing, so you can't do good RL on it. However, I suspect that the above tweet will be quote tweeted sometime this year with an example of a very good piece of writing created by an LLM from scratch, so we will see.