Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Quality non-fiction books are the antithesis of AI slop

resobscura.substack.com

Post reply on HN
Print view View original

141–150 of 265 posts

Re: Quality non-fiction books are the antithesis of AI slop

#141

Post by inigyou »

felipeerias wrote: post #44
adamtaylor_13 wrote: post #14

On a related note, I was just noting to my co-founder, as we struggle to write good case studies for our website, that I find LLMs are astoundingly bad at writing good prose. We all know the "AI-tics" that give away a sloppily AI-written piece, but even if you steer them, they still struggle to write consistently high-quality prose. Somehow I feel that the work of a good copywriter has never been more noticeable.

I gave Claude Fable $25 in Pangram API credits and, after hundreds of attempts, it was unable to produce a single readable original piece of writing that was not immediately identified as AI. This seems to be a hard problem for LLMs, as passing would probably require good self-perception ("oh no, I am writing like an AI!") and fine-grained control over its own output ("let's write like a human instead!").

I think the actual way you'd do that is to run RLHF fine-tuning with Pangram as the judge.

Re: Quality non-fiction books are the antithesis of AI slop

#144

Post by lesostep »

m463 wrote: post #20
jeffreyrogers wrote: post #12

I would bet that how your brain stores information that you read from long-form text is very different from how it stores information you acquire from chatting with an LLM. When I read something challenging or new to me I spend a lot of time thinking about how what I'm reading matches my own experiences or knowledge. Although I'm a fairly fast reader, it often takes me a long time to get through difficult pages since…

I listen to audiobooks while I walk and hike, and sometimes when I recall some particular thing I've learned, I can also recall where I was hiking when I heard it. I think there's a lot to learn on how we really process information.

That's spatial memory! I have long walk-and-talks with friends, and sometimes I forget minor details of previous discussions, but not where we were when it was mentioned.

A lot of my friends can't do same trick. But the fascinating part is that it still works for them! Usually they can recall whatever I forgotten after I described the place we were in.

Re: Quality non-fiction books are the antithesis of AI slop

#145

Post by brrrrrm »

keiferski wrote: post #138

Not to be intentionally contrarian, but I have thought a lot about this: I think good fiction is actually the antithesis of AI, fundamentally. LLMs are really only capable of combining different things into output. This can give you some creative results, but it is ultimately kind of limited. I cannot imagine that prompting an AI a thousand times would give you something truly original in the way many high-quality fi…

Models are increasingly showing their ability to extrapolate into the human unexplored (math proofs being the most apparent). What gives you confidence the absurdity of life is uniquely difficult for models to source?

Re: Quality non-fiction books are the antithesis of AI slop

#146

Post by inigyou »

olooney wrote: post #70
zem wrote: post #50

possibly a bug - I searched for "writers like lewis thomas" and the top two results were by thomas himself.

On a similar note: on Steam, the original Dark Souls game[1] is listed with the "souls-like" tag. That's not wrong , I guess, but just kind of a tautology. Grice's maxims[2] generally have such tautologies omitted in ordinary conversation as non-informative. [1]: https://store.steampowered.com/app/570940/DARK_SOULS_REMASTE... [2]: https://en.wikipedia.org/wiki/Cooperative_principle

I really thought ChatGPT would be able to tell me what this is called but it can't. I suggest "circular eponym".

It's like calling a CRT monitor a really big flatscreen.

Re: Quality non-fiction books are the antithesis of AI slop

#147

Post by inigyou »

Nevermark wrote: post #95

I don't think we would have AI slop, if they were just trained on high quality media. Less relatable AI, maybe, but not sloppy AI. AI trained on other AI (i.e. distilling), can be better than the AI they were trained on, likely because of this reason. Each iteration can raise the bar. As apposed to the deep spiral of generating shiny attention crystals for humans, and then training on that.

The slop isn't a property of the LLM but the workflow it's utilized in. Just now I asked ChatGPT about the linguistic effect where a CRT could be described as a giant flatscreen, a landline could be described as a wired cellphone, or the original Rogue could be described as a roguelike. Unfortunately it doesn't know what that's called either, but that is a non-slop use of an LLM, like a semantic search engine. On the other hand if you print LLM output into a book it's almost certainly slop no matter how the model was trained.
Post reply on HN