Lines of code. 1,596 BTC gone | Hacker News Reader

Lines of code. 1,596 BTC gone

onekey.so

9 pointsArchie62728 comments

Maybe it’s just me but (and I ask this as something of an ML practitioner myself) can anyone at Anthropic or Open AI or any of the frontier labs work on summarizing content into a format where a human won’t immediately vomit?

There’s this sort of local optima all these models pick which is instantly recognizable and hardly digestible for human consumption. Perhaps too much internal RL against benchmarks during chain of thought? The older non reasoning models had they’re own problems but at this point I can’t get an LLM to summarize data for humans, which is troubling.

As much as I’d like to read this piece it follows suit and I don’t have the patience to try and extract anything valuable from it.

I'm starting to notice that LLMs tend to speak in the form of movie trailers.

Say this in a deep masculine dramatic voice:

     One man.  One gun.   A story that would could never die.

Compare that with this Coldcard page:

     Two files, four lines, five years unnoticed.   This is the COLDCARD
     entropy failure

It's trying too hard to replicate a dramatic movie arc with every... single... sentence...

You are correct, and that style of writing is called "broetry".

but someday in a galaxy far far away...

what do you think these things are trained on? mountains of rubbish, human and likely also AI slop. how many bad movies and ads do u think they are fed. poor things. i can see them caged in the corporate HQ, with their tails shoved down their own throats rigged up to the SlopExtractor9000.

its time for the UN to Unslop the world. or something like that :(

Yes, you say:

"write (whatever you want written), BUT do not make it sound like a typical AI don't use any of the tropes or typical phrases that AI tends to use - if you include anything that makes it appear to be from AI then the task has failed. Pull your writing away from the median and toward very specific random styles that are NOT the cliched, averaged LLM style."

edit: actually wrong I tried to get ChatGPT to do this and it kept falling back into LLM speak even when it recognised that's what it was doing.

yeah, no, I've tried a number of variants like that. for ex even if I tell it to not rely on its chain of thought created jargon it'll just usually continue to hallucinate adding in the jargon unless I line item them painfully and eventually get something a human can understand.

prompting it to be concise ("250 words or less") seems to help somewhat but theres only so much you can do esp in cases where the details matter and don't compress well

Give it a sample of your own technical writing and tell it to match the voice. I did this with Fable just today and it nailed it.

that's an interesting idea thx

that is just fatigue. you’ve been vibed too hard.

Maybe it learn it from content farms, not RL.

Before LLM, the majority of the web was content farms... but those are instantly recognizable and we seldom fall into them. Now LLM mixed the content farm style writing with _some_ real content. It take time and effort to tell they are slop, causing much fatigue.

I have no data to back up this claim but one hypothesis I have is it's gotten particularly acutely bad esp with newer gen models because of increased training on reasoning and chain-of-thought.

To borrow a programming lang analogy the failure mode I see a lot is it invents its own jargon that is effectively like a compiler intermediate representation of the high level natural language you actually want a human to look at and then inserts it directly into what the human has to read.

It needs to stop doing that but it doesn't seem to do a good job at differentiating from what is or isn't chain of thought slop. I'm sure the jargon is useful during its reasoning but it's very unhelpful and not very nice to deliver it to a human.

It's not gone. It was just transferred in payment to someone who understood the rules of the game better.

As Kevin Bacon put it in Quicksilver, "What happened to all that money?" "Nothing happened to it. The money's still there. It just belongs to somebody else now."

Not sure if "understood the rules of the game" is doing a good job here. Using a hardware wallet is almost as close as you can get of a cryptocurrency user understanding all the rules of the game. They are following the "gold standard" of security for personal usage. What else should we expect them to do?

So after all that, still gone?

No, someone else has it. It's a zero-sum game.

So the legal owner can’t use it, correct?

So,.. money never "gone" ?

Your competitor has a $100m self-implosion in what must be an all-time slam dunk marketing moment and you spend 7 cents generating a marketing blog post instead of having someone write it.

"This is the part worth sitting with." - written by AI.

It is possible to get the AI to write stuff that does not sound like AI you just have to ask it to properly.

There are so many AI-isms in this. It is frankly astonishing to me that AI detectors are so poor when this is screaming out.

> Six steps, no jargon. The note beside each one is the precise technical version, for anyone who wants to check the work.

It actually makes me uncomfortable reading this garbage now.

You hear it on TV too often political pundits have this exact turn of phrase.

"But here's the part keeping me up at night....."

Since this has become a thread about complaining about unreadable, brain-killing AI slop, I'll add mine:

>Six steps, no jargon. The note beside each one is the precise technical version, for anyone who wants to check the work.

I think this is called "concept leak" or "prompt leak" (please someone correct me, I can't recall at the moment.) It is one of my least-favorite failure modes. I notice it a lot when drafting landing pages, and it eats up a lot of time to edit away.

That's a great name for it; I've seen it on AI-generated form pages, where a version of the requirement text is included alongside an input.

Get a human to write it or GTFO

Good info, but em dash slopmaxxing, slop prose and AI design is such a buzzkill.

It’s not A it’s B. Thanks. Slop.

— — — — and —

Reply on news.ycombinator.com