I like the way Ted Chiang puts it:

Some might say that the output of large language models doesn’t look all that different from a human writer’s first draft, but, again, I think this is a superficial resemblance. Your first draft isn’t an unoriginal idea expressed clearly; it’s an original idea expressed poorly, and it is accompanied by your amorphous dissatisfaction, your awareness of the distance between what it says and what you want it to say. That’s what directs you during rewriting, and that’s one of the things lacking when you start with text generated by an A.I.

There’s nothing magical or mystical about writing, but it involves more than placing an existing document on an unreliable photocopier and pressing the Print button.

I think our materialist culture forgets that minds exist. The output from writing something is not just “the thing you wrote”, but also your thoughts about the thing you wrote.

I think the reason so many AI bros are conservative is that conservatives have historically had really bad taste in art/media, so they see the drivel AI creates and think, “oh wow, it looks just like what the artists make,” not realizing that they don’t have the eye to see what it’s missing.

I means, I’m I huge local LLM advocate. I look at slop metrics, I’ve finetuned or messed with smaller LLMs for prose. I use bigger MoEs with raw completion syntax to get past writers block and some probably-related-to-neurodivergence issues I have.

But they really are terrible with long-form stories.

There’s no way around it. They just can’t stick to a coherent plot.

And even at the “micro” scale (eg “I’m stuck, how do I finish this sentence?” or “how should this character realistically respond?” or “does this paragraph feel stilted?”) they’re only there for inspiration, or maybe as a mirror to bounce ideas off of. Not any sort of output you could rely on.

It absolutely boggles my mind people think they can replace, or even “double” writers. They don’t think. They are literally architected to be terrible at that, no matter how much koolaid on scaling Altman preaches.

sadly, a bad first draft in five minutes usually beats a thoughtful solution in three days

edit: oops i thought we were talking about code. nvm, people actually look at prose.

Writer having toyed with AI, here : yeah, AI writing sucks. It is consensual and bland, never goes into unexpected territory, or completely fails to understand human nature.

So, we’d better stop calling AI “intelligence”. It’s text-prediction machine learning on steroïds, nothing more, and the fact that we’re still calling that “intelligence” says how gullible we all are.

It’s just another speculative bubble from the tech bros, as cryptos were, except this time the tech bros have made their nazi coming out.

AI is just a marketing term, there’s nothing intelligent about it. Its simply Large Language Models, databases that predict what should go next. Its like asking the prediction bar when you are typing to write a story.

That means in about 6 months or so the AI content quality will be about an 8/10. The processors spread machine “learning” incredibly fast. Some might even say exponentially fast. Pretty soon it’ll be like that old song “If you wonder why your letters never get a reply, when you tell me that you love me, I want to see you write it”. “Letters” is an old version of one-on-one tweeting, but with no character limit.

Guy says robot can’t replace him. News at 11.