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

LLMs can get "brain rot"

llm-brain-rot.github.io

Post reply on HN
Print view View original

211–220 of 310 posts

Re: LLMs can get "brain rot"

#211

Post by lxgr »

calvinmorrison wrote: post #173
JumpCrisscross wrote: post #159

Earlier quoted context omitted.

> I love using em dashes Keep using them. If someone is deducing from the use of an emdash that it's LLM produced, we've either lost the battle or they're an idiot. More pointedly, LLMs use emdashes in particular ways. Varying spacing around the em dash and using a double dash (--) could signal human writing.

it's a shibboleth. In the same way we stopped using Pepe the frog when it became associated with the far right, we may eschew em dashes when associated with compuslop

I never understood why so many people would yield their symbols and language that quickly and freely to others they dislike.

In other words, I really hope typographically correct dashes are not already 70% of the way through the hyperstitious slur cascade [1]!

[1] https://www.astralcodexten.com/p/give-up-seventy-percent-of-...

Re: LLMs can get "brain rot"

#212

Post by anigbrowl »

dclowd9901 wrote: post #201
anigbrowl wrote: post #187

Earlier quoted context omitted.

In general I find WSJ articles very well written. It's not their fault if much of today's news is about clowns.

Their editorial department is an embarrassment imo. Sycophancy for conservatism thinly veiled as intellectualism.

I also hate their editorial department, I'm just saying that the news articles are well written in a technical sense rather than because I like their editorial positions or choice of subject mattter.

Re: LLMs can get "brain rot"

#213

Post by koakuma-chan »

commandlinefan wrote: post #70

My son just sent me an instagram reel that explained how cats work internally, but it was a joke, showing the "purr center" and "knocking things off tables" organ. It was presented completely seriously in a way that any human would realize was just supposed to be funny. My first thought was that some LLM is training on this video right now.

https://www.youtube.com/watch?v=sZkB11pO9R8

Re: LLMs can get "brain rot"

#215

Post by tietjens »

turtletontine wrote: post #101
standardly wrote: post #93

Earlier quoted context omitted.

That is indeed an LLM-written sentence — not only does it employ an em dash, but also lists objects in a series — twice within the same sentence — typical LLM behavior that renders its output conspicuous, obvious, and readily apparent to HN readers.

I think this article has already made the rounds here, but I still think about it. I love using em dashes! It really makes me sad that I need to avoid them now to sound human https://bassi.li/articles/i-miss-using-em-dashes

We cannot cede the em dash to LLMs.

Re: LLMs can get "brain rot"

#216

Post by easygenes »

turtletontine wrote: post #101
standardly wrote: post #93

Earlier quoted context omitted.

That is indeed an LLM-written sentence — not only does it employ an em dash, but also lists objects in a series — twice within the same sentence — typical LLM behavior that renders its output conspicuous, obvious, and readily apparent to HN readers.

I think this article has already made the rounds here, but I still think about it. I love using em dashes! It really makes me sad that I need to avoid them now to sound human https://bassi.li/articles/i-miss-using-em-dashes

Yeah, same. I apparently naturally have the writing style of an LLM (basically the called out quote of parent is something I could have written in terms of style). It’s irritating to change my style to not sound like AI.

Re: LLMs can get "brain rot"

#217

Post by TeMPOraL »

lxgr wrote: post #210
JumpCrisscross wrote: post #159

Earlier quoted context omitted.

> I love using em dashes Keep using them. If someone is deducing from the use of an emdash that it's LLM produced, we've either lost the battle or they're an idiot. More pointedly, LLMs use emdashes in particular ways. Varying spacing around the em dash and using a double dash (--) could signal human writing.

The solution is clear: Unicode needs cryptographically signed dashes and whitespace characters.

Tied to what?

Show us a way to create a provably, cryptographically integrity-preserving chain from a person's thoughts to those thoughts expressed in a digital medium, and you may just get both the Nobel prize and a trial for crimes against humanity, for the same thing.

Re: LLMs can get "brain rot"

#218

Post by furyofantares »

turtletontine wrote: post #101
standardly wrote: post #93

Earlier quoted context omitted.

That is indeed an LLM-written sentence — not only does it employ an em dash, but also lists objects in a series — twice within the same sentence — typical LLM behavior that renders its output conspicuous, obvious, and readily apparent to HN readers.

I think this article has already made the rounds here, but I still think about it. I love using em dashes! It really makes me sad that I need to avoid them now to sound human https://bassi.li/articles/i-miss-using-em-dashes

I don't think you do.

All this LLM written crap is easily spottable without it. Nearly every paragraph has a heading, numerous sentences that start with one or two words of fluff then a colon then the actual statement. Excessive bullet point lists. Always telling you "here's the key insight".

But really the only damning thing is, you get a few paragraphs in and realize there's no motivation. It's just a slick infodump. No indication that another human is communicating something to you, no hard earned knowledge they want to convey, no case they're passionate about, no story they want to tell. At best, the initial prompt had that and the LLM destroyed it, but more often they asked ChatGPT so you don't have to.

I think as long as your words come from your desire to communicate something, you don't have to worry about your em-dashes.

Re: LLMs can get "brain rot"

#219

Post by nazgul17 »

Version467 wrote: post #214

So they trained LLM's on a bunch of junk and then notice that it got worse? I don't understand how that's a surprising, or even interesting result?

They also tried to heal the damage, to partial avail. Besides, it's science: you need to test your hypotheses empirically. Also, to draw attention to the issue among researchers, performing a study and sharing your results is possibly the best way.

Re: LLMs can get "brain rot"

#220

Post by yieldcrv »

nazgul17 wrote: post #219
Version467 wrote: post #214

So they trained LLM's on a bunch of junk and then notice that it got worse? I don't understand how that's a surprising, or even interesting result?

They also tried to heal the damage, to partial avail. Besides, it's science: you need to test your hypotheses empirically. Also, to draw attention to the issue among researchers, performing a study and sharing your results is possibly the best way.

I don’t understand, so this is just about training an LLM with bad data and just having a bad LLM?

just use a different model?

dont train it with bad data and just start a new session if your RAG muffins went off the rails?

what am I missing here

Post reply on HN