Post

A refinement of that bsky.app/profile/moul...

I think my claim is mainly that it isn't that they can't, but that they don't, because they are trying to stay close to the prompt you give them.

i guess, but you could say the same for almost any really difficult task? like, the model won't just solve clay problems for you (even if it could, with the perfect prompting), because it's trying to stay close to the prompt. it's true, but feels kind of empty?

1. the problem is just way harder. good writing is mostly an aesthetic judgment, unlike code or science, where correctness is often verifiable. people are much less forgiving! if AI writes 120K words of code that *works*, devs are impressed even if they think it's insecure spaghetti code.

whereas if AI writes a broadly coherent 120K-word novel, nobody is impressed unless they also like the quality of individual paragraphs. so an AI writer has to be able to write like almost *every* human writer to be "good", and it has to produce high quality writing fractally, *at every scale*.

2. for reasons related to the above, learning to produce coherent and high-quality long form writing via pretraining or RL is probably much harder than learning to generate code or even do math. lack of verifiability, and the inherent subjectivity of writing, really hurts you here.

My view, which no one seems to agree with 🫤, is that humans aren't *that* difficult to pull their levers and press their buttons (evidence: mass media today) My imho is that long fiction that pulls the levers and presses the reader's buttons is mainly a memory management problem at this point

I think the signal to watch for is long fiction that proves compelling to a reader *despite their biases* Like the book you can't put down