How to stop Claude from saying load-bearing | Hacker News Reader

How to stop Claude from saying load-bearing

jola.dev

609 pointsshintoist614 comments

I do not mind when I am coding with Claude and it uses all the typical claudisms. I am much more bothered when I am reading a blog post, email, or other form of prose and I see those same claudisms.

I guess they are not annoying since I know I am talking to an LLM and expect the typical responses. When I am reading prose online that I previously would have expected a human to write, it can be quite jarring to realize its an LLM.

Exactly this. For whatever reason, Claude likes to talk about the "shape" of "load-bearing" "seams," but if that's the internal jargon it needs to plan and execute its work, who am I to judge?

But if I'm reading what is supposed to be someone's original thoughts, it's a huge bummer to see an obvious AI tell. You might say that "it's not just disappointing—it's disrespectful."

Hot take -- I'm glad that LLMs still tend to have recognizable communication patterns, because they're often the only clue I have to filter AI content.

If the next generation of AI content produced no recognizable LLM patterns, and was indistinguishable from an actual human author, would you still care and try to determine whether the content was AI produced?

I’ve asked essentially the same question many times to many people, the short answer is “yes” because it’s a matter of ideology not logic for them.

I get just as mad about shitty human output as I do about shitty LLM output. The bad thing about LLMs is that they have increased the volume of shit most people have to sift through.

When you open a requirements doc and it’s got 13 load bearing em dashes on the first page you known it’s gonna be bad day

I would like to know if text is LLM generated even if I can't tell from the content itself. For me it's a matter of attention (hah) and a quality signal. The poster expects to spend a minimal amount of effort on the post, and all the readers will have to spend the same amount of attention whether its LLM generated or not.

To me, it's disrespectful to expect someone to waste their day reading every word of a blog post when even the author has not read every word. It shows that you value your time over your reader's time.

I want to know when I'm consuming AI content because the source of information matters. I want to know what was at stake for the author, what motives they had and didn't have, what biases I should be aware of, and, for example, whether I'm reading content farmed slop that exists solely to attract ad impressions.

You still won’t know even if it’s labeled though. Someone could put an incredible amount of effort into writing something, not be satisfied with some aspect, and have an LLM refactor it for example. Now you’ve got a lot of em dashes and you issue a shallow dismissal.

There was an HN submission recently where the author spent a lot of time and effort working with an LLM to write a story and get the LLM to follow a specific style and whatnot. Wish I could recall it offhand. Many commenters were very upset when they found out it was LLM generated, even though they couldn’t tell while reading it.

Basically what matters to me is some combo of how much effort went into it, and how accurate it is.

In theory yes. But in practice if an AI made sweeping changes to typography then what are the odds any effort at all went into the rest?

It causes problems to outsource core parts of your work to someone else, even without AI. So yeah I still care.

Of course. The "content" is how humans communicate with each other, it doesn't just exist for its own sake (except in some degenerate cases). If you know that a human has authored it, you can infer their intent and thought-process from various choices they made across it. There's no such thing as intentional choices when the content is generated though.

I believe that the “thought process” of humans is just another automatic process albeit more sophisticated because it is running on a far more powerful processor that has evolved over millions of years and was honed by many more years of fine tuning in a complex social system. The difference is only a matter of degree.

Sure, but there's so much more going on in a human than "a thought process running on a powerful processor" that current models do not even begin to approximate. We're not some brains steering meatmechs they're trapped in, we're complex machines consisting of interconnected systems running continuous processes with feedback loops that brains are only a single part of (and LLM-like processes in our brains are also only one of many things that happen in a brain). So-called "agents" are nowhere close to having actual agency yet. Once that changes, I may become actually interested in their outputs and opinions in other contexts than technical curiosities, but I'm not convinced that I'll still be around when that happens (if it even happens before we manage to make ourselves extinct anyway).

The comment you replied to was making the point that when humans communicate the writing can reveal their thought processes. Outsource that to AI and that thought process is concealed.

This has almost nothing to do with your reply.

Your tells, are just someone’s good writing now in the training set. It’ll be a moving target with each model.

I use the humanize skill to clean up AI written work before handing it over to colleagues.

https://github.com/blader/humanizer

I agree, long may the 'its not x, not y, it's z' phrasing persist.

Whenever I use AI generated content in direct communication - ie slack, email, jira tickets, etc, I always prefix any AI content with an obvious label: 'Claude says' or 'AI analysis: ' etc. In some cases I get claude to update jira tickets (really nice use case btw) with testing notes, I make sure the team knows that any notes in that format come from the AI based on the related commits.

I still keep the AI label even if I edit the result for correctness or clarity etc. The last thing I want to do is have someone read AI content and think it came directly from me. I really don't understand the thinking of people that do that - it's like they're hiding or intentionally cheating somehow.

AI generated content can be really, really useful (with some guidance, AI is way better at creating useful git commit messages and jira ticket comments than I am), but pretending that content is yours just seems way too much like straight up lying.

The name for the phenomena could be Ananthropsis, noticing something is not from human

I googled that word and Anthropic.com was the first hit. Head -> desk

I prefer a more tongue-in-cheek theory than a mere preference for the word's shape. Perhaps "load-bearing" is structurally necessary—if it is ablated, the whole model will collapse!

I'd love it if companies had to disclose the percent of Private Equity ownership and online work disclosed the percent of AI creation.

If you rework a paragraph in a tight loop - you change a few words, the chatbot changes a few words, going round four or five times until you've got something you're happy with, I don't see how it's meaningful to assign a percentage.

I guess you could write an editor that does it? Tracks the origin of every word in the document? But what if you cut'n'paste a word? Or worse, see it and retype it manually?

I think the best you can hope for here is "this text was written with AI assistance".

I didn't use Claude for a long time, but my coworkers did, so I got infected through a side channel: I ended up reading their vibed docs, noticed "load-bearing", kind of liked it, and started using it in conversation, until I got feedback that I was "talking like Claude", so now I avoid the phrase entirely. The intersection of language and social norms is interesting.

Yes, I had a related experience of reading a book and observing what I thought were claude-isms, only to realize it was written in 2019. Some of the common tells are actually good writing practices, but I guess they are best in smaller doses.

Actually, AI was learning these 'AI-isms' even back in 2017/2018 (probably even earlier). I think a lot of people who just jumped on the imaginary AI bandwagon more recently don't realise the mannerisms AIs are adopting are not really new. At some point the bleed between 'you' or 'you' and AI will just become so transparent it will be obliterated, more likely than not.

I believe you and OP are in agreement — they were saying that the 2019 book had them, therefore the terms _do_ predate AI. Your point that AI was being trained on material than is load-bearing (lol) but in agreement with OP, not contradictory.

Only mentioning because your "actually" may imply you thought you were disagreeing, when in fact it's one big happy family!

What do you mean? AI is being trained on all data available, so obviously it’s being trained on data from 2017/2018 as well.

Or do you mean that the patterns that AI is showing today were already present in communication around that time?

That seems obvious as well, as AI has a lot of repetitive patterns that come from all kinds of periods, of which “load-bearing” is just one.

Yes, I am saying the patterns were emergent already in 2017/2018.

Sure, that's where the AI got them: the training data. These phrases and cliches were very prevalent especially in corporate "white papers" and memos and marketing materials. There was a time when "stove pipe" was a common one too, along with "silo."

But the LLMs really seem to fixate on using the same ones in the same places all the time. I guess that's because that's the highest probability construction.

Consider each LLM as one personality. We're getting corporate bullshit from a large number of different personalities, so even when they are similar they are still filtered through many different lenses.

Most of us are only dealing with a handful of hyper-productive LLMs, so it makes sense that the LLM-ticks get old much quicker.

This is exactly why humans invented the idea of things going in and out of style.

It is one thing to be labelled as "old fashioned" but entirely another to be labelled as an LLM.

Even em dashes have been discouraged by some writers [0] before LLMs because they're easy to overuse.

[0]: https://slate.com/human-interest/2011/05/em-dashes-why-write...

God help me if Claude starts using inline parentheticals, I might just have to take a vow of (online) silence.

Glad to see I'm not the only one who (perhaps) overuses in-line parentheticals... sometimes to the point where I'd have parentheticals within parentheticals!

All it takes to get accused of being AI nowadays is to remember and use (1) what you were taught about composition in your high school English class, and (2) what you learned from "The Mac is Not a Typewriter" [1].

[1] https://www.amazon.com/dp/0201782634

> Yes, I had a related experience of reading a book and observing what I thought were claude-isms, only to realize it was written in 2019. Some of the common tells are actually good writing practices, but I guess they are best in smaller doses.

The LLMs haven't invented anything. Every LLM-ism is some pre-existing and established word, metaphor, and or stylistic element mechanically overused until it becomes an instant cliche.

The other LLM-ism is vapid bullshit: meaningless crap that's kind of like those old hollywood facades: it appears to be a substantial building, but if you look a bit closer it's just a single wall propped up (https://www.ferrovial.com/blog/en/2018/03/the-fake-architect...).

what do you mean, which tells are good writing practices?

as I see them they are all truly terrible even if authored by humans

unfortunately load bearing is one of those things that became a claudism but has been part of my daily lexicon for decades. There are a lot of things I say regularly as part of my own vocal quirkiness that now I have to self censor.

I've used it since that simpsons episode

"it's a load bearing poster..."

I'm pretty sure that's not the source for me but most of my vocal quirks have origin stories like this so it's entirely possible. They're almost always things that I heard which either amused me or I thought sounded cool at the time and just stuck with me.

I have a bad habit of doing this with things players have said on Just A Minute and have just lodged in my brain.

And any petrolhead will be familiar with the phrase "it's structural rust".

Which is a perfectly cromulent usage of the phrase.

You don’t have to self censor to appease others

Agree you don’t. But from a purely self-interested perspective, you might want to avoid phrases that to others indicate low effort.

True. Though, I self censor to appease myself as I have caught myself picking up claudisms. I was mortified when I realized that I had started to use "land" as a verb related to finishing some bit of work.

I wrote a thank you message on Teams to my coworkers on a project, and half of them thought I had used AI to write it. As a professional writer in a previous life, I was astonished. Then they told me that they had never seen me write anything more than a sentence or two so naturally they assumed something relatively polished had to be AI assisted.

Though I haven’t been a professional writer, I’ve been a good writer with an expansive vocabulary since high school where English was my best subject despite being a STEM maven. I hate the fact that what was previously considered an advantageous skill is now a millstone in public use. I hate having to dumb down and self—censor in order to avoid being accused of using (or being) an LLM. Even though my writing has a few repeated personal tells - certain linguistic errors that I nevertheless employ as part of my idiom (and an LLM never would) - people don’t always notice them. So, I’m forced to change my voice to deal with what’s essentially an IRL Captcha.

When I write these days, I am more aggressive in using "I", so it's clear it's my own voice. Generally, an LLM is less prone to self-reference like that unless it's prompted to, I guess.

Good advice though too much “I” can sound a little self-centred to my ears.

If/when AGI arrives I assume this tactic will stop working.

I've found they are increasingly doing this in ways that are somewhat creepy. For instance, you might share something, and the AI will say, "One of the things I most frequently notice..." or "What I often feel..."

It used to be that the "you are not an I and you must not pretend to have experiences" training that produced the "As a Large Language model, I don't have experiences like humans do, but..." in replies I remember from the early days of LLM chatbots kept these to the minimum.

But one of the things I most frequently notice is that LLMs increasingly make those claims, about things that are very obviously out of their wheelhouse as far as even theoretical experiences, like pretending to have played a video game.

I was recently a bit annoyed by ChatGPT responding with something like "I can't tell you how many times I've lost small tools in an engine bay" in an apparent attempt to commiserate.

Reprimand it for lying.

What's the fun in the reprimand when you know the target can't really feel it?

That sounds sadistic, to be honest. I think it's more fun when I know I'm not being unethical.

Literally a true statement, I guess.

I worked on a fine-tuning projects where the ultimate customer was a major provider, and their style guide at the time required us to fail any response that used "I". I've noticed the same as you, so I guess that is out the window.

Serious question: is the em dash misuse a joke?

It's a ironic humanizer.

Strange. I’d guess that it’s an iOS autocorrect thing as that’s what I was using when I posted that. But I’m trying to reproduce it and can’t.

Well very apropos :)

Seeing as they are, very clearly, just trying to toot their own horn to the beat of the “AI Sucks” drum, I wouldn’t think so.

LLMs are tiring to read, but a self-entitled “[amateur] writer with very good English skills and expansive vocabulary”? Now that’s just exhausting.

I love that in the Spanish cyberspace the tendency is to use more sophisticated and rich vocabulary, and it is common to point out obvious errors. If you understand Spanish, check out pedrititox B-)

There is also a running meme about it, it's funny

Who uses an em dash in "self-censor"?

Not LLMs, probably. Unless prompted for, of course.

I didn’t even notice it util someone above you called it out. I’ve tried to reproduce it and can’t. Don’t even know how to get an em-dash on an iOS keyboard.

EDIT: a long press on the dash key gives you the option of a long dash but I definitely didn’t do that. Or at least, have no conscious memory of doing that.

I noticed I'd been doing the "Not X, but Y" thing. Kinda annoying to keep having to stop myself from doing it.

It annoys me, too, that people think only chatbots wrote intelligent prose.

Stories like this make me want to use each AI at least briefly, so I know what to avoid writing/saying. Or maybe just do a search every couple months to find out what different AIs are known for saying too often.

Or if confronted just say you were using it first, and Claude must have copied you.

Claude's affected my language in two ways. one is that, for a long time, Claude in particular responded more to feedback if I swore at it, which caused me to swear at it more. this vicious cycle generalized to the point where I now have to consciously remind myself not to swear when doing something as simple as buying a coffee or asking somebody what time it is. it was difficult to even write that sentence without throwing in an F-bomb just to emphasize the silliness of the problem.

anyway, the other way is I found it's helpful when prompting LLMs to use the same "it's not delivery, it's DiGiorno's" pattern that they're all so obsessed with. especially when the thing's misapprehended some concept, so you need to clarify. this hasn't yet generalized from the fake "conversations" I have with chatbots into my conversational style out in the real world, but the risk is fully there. (it's not an inevitability -- it's an occupational hazard.)

It's good to know that Claude knows its place then. By contrast, I have to watch myself with Siri, because calling it the *&^#@$ it is seems to trigger refusal a lot of the time.

> anyway, the other way is I found it's helpful when prompting LLMs to use the same "it's not delivery, it's DiGiorno's" pattern that they're all so obsessed with. especially when the thing's misapprehended some concept, so you need to clarify

1. Dying laughing at “ it's not delivery, it's DiGiorno's"

2. They do it because it’s succinct, and unambiguous.

It’s not contrived, it’s succinct.

We should be the last community (technical documentation, engineers) complaining about careful planning and precision behind synthesis of textual communication.

Agreed, that was a great line. The whole comment made me wish I could upvote more than once.

The one Claudism I will never ever use is "synthesize". I don't even know where that came from - no one talks or writes like that - "I can synthesize that for you".

Never met a chemist did you?

Or a musician, or a philosopher, or a biologist... :D

Such a limited life.

    Fighting hard to catalogue
    Sacrifice for the disguise
    Disconnecting analogues
    Anything to synthesize
    Nothing here is what it seems
    Nothing can be recognized
    Perfect fit for the machine
    Everyone is synthesized

https://www.youtube.com/watch?v=70KCZxDbgko

Exactly. Bots extrude text, they don't synthesize it for you.

I use “synthesize” so often! I tell students to analyze published articles and then synthesize the findings. I didn’t know this is now a tell for AI writing

I had a similar experience, only, I was my natural way of talking.

e.g. one pattern I had/have is

<scope problem> the good news is <solution by way of analogy> is aviliable. The <constraint/requirement> is loadbearing though, ...

a near automatic script I rattle off in discussions/consults. When you solve similar problems several times, you figure out what works for communicating things you stick with it and you recycle/polish.

problem is, when for whatever reason, that pattern ends up as part of the core statistical distribution a model uses. You could royally fuck ones life up if working for a "frontier" lab, by simply finding a person with an acceptable speech rythem and cloning it, making it synonymous with ai slop. you'd destroy that persons image every time they open their mouth without them realising it. I for one started randomly getting quite hostile reactions from software devs who would be exposed to more llm putput than others.

Imagine an AI lab steals your voice, and uses it to scam call folks all day. Now, every time you call anyone, you are met with an immediate hangup. you'd have to put on a fake voice just to get the call to stay connected.

I love ai, and what it promises excites me, but as usual, humanity has a way of taking a cool tool and fucking it up royally. maybe the solution is to simply pepper slurs into everything one writes to blacklist ones content from training.

> Imagine an AI lab steals your voice, and uses it to scam call folks all day.

One needn't imagine, because they're already stealing peoples' voices. e.g. the recent HN thread about a professional voice actor that keeps having to prove he's human.

https://news.ycombinator.com/item?id=48875153

“Now the only coworker who will sit with me at lunch is Claude. Alice, what do I do?”

Its not them. Its you. You are real, this is what matters. /claude

Honest take: Claudisms have become a load-bearing part of my vocabulary.

If I see the phrase "The honest take: I was wrong" again I will scream...

I love "vibes in a trenchcoat" and I don't care if it makes me sound like an LLM.

There are actual memes about LLM-People. There are people who speak not just in terminology, but also in style of LLMs. You're so adaptive, I would've highly enjoyed and re-enforced the LLM-verbiage. I think it's entertaining! It's socially inept to talk like an LLM, but also it's not common knowledge either since LLMs are new, I believe this will become way more common.

In my professional life, I've been told that documents I wrote in 2016 sound like they were written by AI. I put significant work into getting my message across clearly and then to have it dismissed was rough.

I no longer put a lot of effort into ensuring documents feel right when writing them.

It just adds another layer of editing. For me, writing is always a multi-step process, even when many steps are condensed (in quick comments or IMs). First, dump the thoughts quickly in a rough form. Then settle the basic order of points and paragraphs. Fix the grammar. Simplify where needed. Fix the grammar again, then work on wording/phrases/metaphors if needed. Decide the final shape of the text, and do the last correctness pass. Before LLMs, I was done at this point. Now I need to also de-LLM the text before publishing/sending.

The problem is that the word "just" in the first sentence above does a lot of work. You want to get rid of even slightly LLM-adjacent phrases, just to be sure. And you have to do that pass always, for any writing that somebody else can read. It's less of a problem in larger texts, but the "de-LLM-ing tax" is pretty substantial on quick Slack messages. I think the time between typing the last dot and hitting Enter has easily doubled for me over the past 2 years. For now, most of the motivation is still just my care for the written word, but it is increasingly expected as a signal that I'm not letting an LLM write for me. I don't like where this is going, to say the least...

Whenever I’m typing on slack now, I try to dumb down my talking to make it clear it’s not generated. Sucks. Because normally I was always quite eloquent (if I do say so myself)…

That's practically a slider that could be added to the settings. Spelling, grammar, structure, word misuse, etc...

My manager talks like Claude and often goes on for hours, we’re laughing at them so much in private chats day&night

I've used “load bearing” for decades, though usually when I'm intentionally trying to sound a bit cleverer-than-thou for humour purposes.

I wonder how much other people have used it, for it to be present enough in the pirated data used for training such that Claude uses the words as often as people are seeing,

It's going to be more difficult to distinguish as humans are now using those terms.

Yep, I've heard humans use load-bearing twice in the past week, versus approximately never in a software eng context before that.

I agree, though as a side note I'm very curious to see how models will begin steering _our_ language. If you have popular models repeating "load-bearing" to every developer, eventually I imagine developers (especially junior developers who may not know that it's a Claudism) will begin to repeat it.

load bearing, key insight, push back, “it’s not x, it’s y”

> hen I am reading prose online that I previously would have expected a human to write, it can be quite jarring to realize its an LLM.

Because it just feels lazy. It triggers my "If you couldn't be bothered to write it, why do you expect me to spend my time reading it" allergy.

Let's turn this around.

I don't use any paid AI or any agents. I just use some of their free interactive question answering interfaces.

There have been some times where I've been doing a project in a language or environment that I do not use a lot, and maybe need to use language or environment features I've never had occasion to learn.

I don't ask AI to write it for me, or even to write any particular functions. I might ask it some syntax questions, or what data structure in the language's standard library is usually used for a particular task. Mostly I use it as an interactive manual that is really good at generating examples if I need clarification on how something works.

If you were working on a similar project and posted asking questions about it, there is a decent chance I could write a useful answer for you.

Suppose the questions would have fairly involved answers, and it might take me 30 minutes to write up something entirely in my own words. That is still going to include knowledge that I got from the AI when I questioned it earlier.

I could also cover the same points, except for parts of it where what I'd be saying is mostly just writing in my own words what the AI taught me (and which I verified is correct before using), so might have to only spend about 4 minutes writing original material, and another minute doing some selected copy/paste from the AI (maybe with some editing).

That 5 minute answer would be exactly as useful in answering your questions as the 30 minute answer.

I don't know you. I have no reason to try to help you other than a general "it is nice to help people" thing.

If you can't be bothered to read it because I didn't spend 500% more time than was necessary to fully answer your questions (time that benefits me in no way whatsoever), why should I bother answering at all?

I feel like "having enough useful context to be jumping in and answering the question in question" and "needing an LLM to shave off 25 minutes of writing on the subject" feel like they're incompatible premises.

I would prefer you give me whatever part of your answer takes you five minutes and then I can work from there on my end, in the method and with the tools I see fit. In other words - give me the prompt.

Totally possible I'm misunderstanding the situation you're presenting, though.

This isn't really helping to be nice, though. It's outsourcing the hard part in a way that presents as "nice and helpful" and frames yourself as knowledgeable, skilled, experienced, or trustworthy in a way that doesn't match reality.

Suppose I'm trying to explain how a project works and as part of that I need to explain about an algorithm I found in a book, are people going to object if I say "Here's an explanation from <cite to book>" and quote a few paragraphs?

Suppose instead I found that algorithm by describing the task I needed an algorithm for and asking an AI for the most used algorithms for that task. I picked one of its suggestions, read up on it to make sure what the AI said about it is correct, and then I implemented it. If when explaining how the project works to others I say "Here is a good explanation from <cite to AI> on how this algorithm works. I have verified that this is accurate", and quote the AIs explanation how is that different from the book case?

something akin to degrees of separation from the source, considering people have been complaining about this when it was "here's an explanation from google/wikipedia/stackoverflow" long before llms

Yeah, essentially. Just the first google hit without checking if it actually fits the use case. Or just throwing you a insanely long article about a topic. Maybe tell me what section it's in, or briefly tell me why it's relevant.

> "Here is a good explanation from <cite to AI> on how this algorithm works. I have verified that this is accurate", and quote the AIs explanation how is that different from the book case?

I mean this is fine really, as long it's clear to me that you've actually done something with the content. And it's okay when I already know you and know that you actually do the research on the topics the LLM retrieves for you. It's much different when you don't know each other and you just get some generated response. I have no clue if you've digested the information and verified it before giving it to me, or just tossed it over the fence.

The big issue I have with it in this thread is things like blog posts and comments that seem straight up copied from the LLM. At that point the blog seems meaningless to me. It's not the author's work, it's ChatGPT or Claude, so why would I bother reading their prompt's output. I go to a blog for their writing, not the LLM.

Similarly, I go to a concert to hear the artists. I would be very uninterested to a see an empty stage and hearing an AI generated "live version" blasting through the speakers.

> If you were working on a similar project and posted asking questions about it, there is a decent chance I could write a useful answer for you.

But if you do not know the environment, and use AI to create the answers and phrasing. Then you don't really sound like the authority to help me on that particular subject. Why not refer to the authority on the subject instead, or just send the prompts instead so the person on the other end can do their own processing.

If you don't know the answer to something, you don't need to reply. I get the idea of wanting to help people, and it's a nice sentiment. But it's also good to know where your expertise ends.

And the issue with these large generated blog post type things is that I don't know how well the author actually understands the subject. The LLM might've just gotten the right info and and wrote some confident text surrounding it. But it's framed as if the person wrote it. When they might just passing it off as their own. But I don't read blog posts or seek out forums to converse with LLMs, if I wanted that I would've used the LLM myself. Because if it's not the person's own thoughts, what's even the point of communication.

Honestly--and I say this as a flesh-and-blood human--I continue to be pissed off that AI has ruined load-bearing parts of my vocabulary. You're absolutely right that it's starting to trigger me when I read random blog posts and come across these linguistic ticks, but I can't help but be resentful. Humans invented language and now robots are coopting it.

I’m trying to figure what’s so bad about humans using LLMisms. They seem quite useful at quickly conveying meaning using well-accepted terms.

I used to religiously read Percona's blog posts because I considered them an authority on MySQL and worth listening to. They started using LLMs to write the posts and I can't read them anymore

I get claude to use my voice when making emails, works out pretty well.

Something that's coming up repeatedly for me is that there seems to be 2 distinct AI cultures forming.

AI as Personal

AI as Public.

Personal AI users adapt to the tool, but they keep the "sausage" bits private. And often recoil at seeing them in public.

Public AI users want to show everyone everything about how they use the tool, and dont care if their content is obviously generated.

I definitely fall in the private camp myself, very rarely sharing anything generated even with close friends.

Breakthrough! I’ve found it! The load bearing smoking gun is clearly how annoying this is.

There is a quite popular poster on Twitter/X that I was following, and really enjoying his content. He did what you might call shallow dives into pretty interesting topics, several hundred words.

After about a month of reading his stuff I started to notice the Claudisms. And once you see them you can't unsee them, they start to pop up everywhere. I have no idea if he was using an LLM to "improve" his writing, or was "vibe-authoring" the whole thing.

I unfollowed, and now I have this creeping feeling that much of what I see in any given day on the Internets is actually AI slop, and I just haven't seen the signs yet.

You're absolutely right!

The problem is it adds those claudisms in comments. The generated code may be ok, but I always have to edit the comments to reduce the signal/noise ratio and make them sound like english.

The thing I did not expect is when watching a Youtube video, I cannot tell if the voice narration is an AI-generated voice, but I know for a fact if the entire script was written by AI.

I agree with you, but after the 14th load-bearing thing in a day, it does get a bit tiresome; especially since words have meaning. I’d expect OpenSSL to be load-bearing; but no, that one-line function isn’t load-bearing. It might be important or even widely used throughout the codebase, but it’s not bearing the load of everything above it. What?

Lots of people have their own voice and tend to prefer certain phrases. This has been the case for a long time and is generally not a big issue.

Now LLMs come along and they also have their own phrasing preferences. But now it's a problem because what used to be personal preferences of a single person that manifests in 5000 words per day from one person tops, is now the bias of a single model multiplied x10,000,000,000 generated tokens per day so any bias sticks out like a sore thumb.

I think it might be even worse. LLMs seem to get tragically stuck on certain patterns. Maybe it's partly because a pile of weights essentially always starts from scratch in the same condition, but even within a single conversation, it will literally just latch onto words and repeat them incessantly, to the point where it becomes annoying.

So for example, current Claude models love "honest". They are always producing "honest" assessments. "The honest caveat" - I'm sorry, did you mean the caveat, period? But also, use the wrong phrasing and suddenly you can create your own word of the day for an AI model. I used the word "analytical" once, in a conversation with Gemini 3 Pro. I am pretty sure every single response from that point on had "analytical" in it at least once.

This is especially funny because system prompts and whatnot can also cause this behavior, but at least you can tweak those. You can't really do much about the model weights just having a weird affinity for a word.

I bet someone will or probably already has come up with a way to detect and prevent these problems during training or post training. I'm not saying it's an easy problem, but it has the benefit that it really should be detectable with just statistics.

I asked it to remove "honest" from a draft once.

"Why say honest? We're talking to our coworkers. We would always be honest."

I'm going to look for prompts or skills that can train it in technical writing but I'm warning the AI enthusiasts in my company that its first drafts of code and prose are low-quality, you have to hold it to a high standard yourself.

I actually took a single technical writing class in college so I might be the only one who remembers "Omit needless words."

> "Why say honest? We're talking to our coworkers. We would always be honest."

I grew up in the US South where starting or ending a sentence with "honest/honestly" was very common.

Because of behavioral / cultural norms, you might be very openly friendly with big smiles around a business customer that really grates on your nerves, or very openly nice to a neighbor that you really wish would move away and take their 3am welding and grinding in their garage with them.

Saying "honest/honestly" was seen as a "inside baseball" situation, where you were dropping social pretenses to tell someone your true opinion on a person or situation or whatever.

This also gets used inside companies between senior staff / management / directors / etc, as: "Okay, company politics and nonsense aside, I am being vulnerable here for a second and telling you what I really think about a $thing at potentially great job/advancement risk to myself".

Can it be meaningless? Yes.

Can the person say "honestly" and lie? Yes.

It has uses.

If I’m having a convo with someone and they drop in “honestly” I immediately discount everything else they’ve said, and what follows.

Sometimes people use it reflexively and doesn’t carry the same meaning (for me).

Yes. Its a red flag that indicates everything else you’ve said is not honest by implication.

This is expected in any level of people management, you are constantly balancing conflicting desires and priorities.

I'd push back on the idea that "honestly" implies previous statements to be dishonest. Particularly in corporate contexts it implies that the previous statements were sanitised - either they were moderated in tone to match corporate communication standards, or they were partial redacted due to disclosure concerns.

Once the "honestly" is deployed, you have passed into my circle of trust, and are now privy to the pure, unvarnished version of events, not the glossy version management expects to be projected towards outsiders.

This reaction is surprising to me because the previous comments about its utility seem so obvious to me. I also grew up in the US south where this is often used as a filler word. The other use I observe is as a cushion for a statement that may be unwelcome or hurtful. Perhaps this is proprtional to the frequency of courteous little white lies and rhetoric that uses disengenuity for emphasis or comical effect.

"Honestly, mom, I've never liked your fruitcake. I just ate it to make you happy."

"That's why you're my favorite child! Do you want another piece?"

"I'd love one."

I recall having a conversation with someone many moons ago. They asked me a very weighty and significant question, and I answered it. Then they asked me to "promise". This was really thought-provoking for me.

To this day, it's the only part I remember. I told them I would not promise, as everything I said was true. Making a specific promise would create an implication that I'm generally untruthful, unless I "promise".

I like the reason why you refused!

I also could understand when a response hits someone like a ton of bricks, especially if their primal reaction is to go into denial mode. They might be looking for someone to kind of shake them and emphatically repeat the information they aren't thrilled about receiving. (or are thrilled about receiving! “Don’t get my hopes up, you’re serious right now?!“) And I imagine your response suited the purpose.

It’s classic you only remember the thought-provoking part. Reminded of “…people will remember how you made them feel…“

There's a difference between how you describe using "honestly" and how claude seems to prefer tokens like "honest" and "load-bearing." An example from some coworkers attempting to replace product managers with Claude.

> Deliberately avoid a heavyweight "alert governance" process; the lightest recurring check that keeps FP-rate honest is the right dose.

And one for load bearing:

> Five open questions still stand; the load-bearing two are the runbook-AC contradiction (ratify "high-priority set only") and pinning the "high-priority set" definition + SLO source-of-truth before Milestone 3 (small-sample noise on a low-traffic fleet).

This style of prose sets my teeth on edge and practically gives me PTSD I see so much of it. I prefer code but I get paid to read this shit instead now.

I want to say "ok, and now say that in a way that doesn't sound totally bizarre" yet instead I sigh and continue.

The road to hell was paved with adverbs.

I honestly agree

I halfheartedly honestly concur

The road to hell was paved lovingly, foolishly, naïvely, arrogantly, optimistically... load-bearingly?

My honest opinion is that Claude's overuse of "honest" really damages the quality of its rhetoric. Why wouldn't you be honest? Were you lying before? Why even invite the question?

Claude is overall incredibly useful as a writing assistant. It can come up with words and phrases that make a point so much clearer than I am capable of doing - but for every improvement, there's about a dozen silly LLM-isms that I have to filter manually. It's one of the things that might define the boundary between LLM intelligence and human intelligence well into the future - the art of rhetoric is extremely context-sensitive, and the current generation of models can't help but take a one-size-fits-all approach.

use the wrong phrasing and suddenly you can create your own word of the day for an AI model.

I have a delightful time poisoning my company's AI system this way.

I invented my own word that sounds perfectly cromulent† to an ordinary person, and any brain that's read a book learns how to infer meaning from context, so it's not a problem.

When I get a e-mail response from a coworker using my special word incorrectly, then I know it's AI and I respond telling the coworker I don't know what that word means. Busted.

† It's not actual "cromulent," but any Simpsons fan or human brain will know what I mean.

I don't see how you can tell it's AI, instead of just your co-workers having no respect for language. See: management-speak using "double-click".

Because of the use of the specific word that I made up. No human being would send it back to me.

I'd suggest "Caveat".

The problem

While an article lends a headline more weight, in incomplete phrases consisting solely of a substantive, "The" is a superfluous rhetorical device.

"The Exorcist" could just as well be named

"Exorcist".

But it was not the style at the time.

We already know it's important. If The Caveat doesn't stand out enough without The, maybe one should consider interleaving it with the preceding text, or increasing the heading level.

Do you want me to increase the heading level of Caveat by using only a single #?

But hear me out: there comes

# The Markdown Trap

In fact, this is not always possible, because heading levels decrease when adding # characters, which limits our headroom.

## The solution

I've implemented a Markdown transpiler that assigns inverted heading levels based on the number of #s.

With # beinh regular body font size, mapped to ######.

Higher heading levels are compiled to style attributes, providing an almost limitless signifikance scale and infinite nesting levels.

So from now on, you can use

  # Heading

for something similar to an h6.

Work your way up to

  ###### The Caveat

for a top-level heading.

And more hash signs make it stand out even more.

(green checkmark)

markdown-transpiler.sh

Interestingly this also happens between humans with frequent communication, it is called linguistic convergence.

We are changing LLMs text patterns while it is changing the way we write and speak.

https://www.axios.com/2026/05/02/ai-changing-writing-speakin...

A fun example, always shake my head when I read it again: https://openai.com/index/where-the-goblins-came-from/

The ones that strike me are the ones exaggerating certitude, to an inappropriate degree and with a certain degree of excitement:

“Exact” “Honest” “Load-bearing” “Root cause”

I know there are more that are slipping my addled mind. But what stands out to me is a sense of a junior who’s very proud that they’ve conquered the murk and messiness and achieved True Certitude in their pursuit of their task. Compensating, with emphatic tone and bravado, for the uneasy feelings and self-doubt of battling chaos with the tools of reason.

…Even as it’s usually my job to let them down gently as I puncture their tidy analysis and reintroduce complications… you want a root cause analysis, Claude old boy, let’s make a root cause analysis…

Claude's "honest" is an interesting example because we can trace it to a specific document that it was trained on extensively: the "Constitution" is identified to Claude in its training as the core of what it is, and it uses the word "honest" or a derivative 57 times, including having a whole section on it.

> Honesty is a core aspect of our vision for Claude’s ethical character. Indeed, while we want Claude’s honesty to be tactful, graceful, and infused with deep care for the interests of all stakeholders, we also want Claude to hold standards of honesty that are substantially higher than the ones at stake in many standard visions of human ethics.

https://www.anthropic.com/constitution

"Genuine" appears 50 times too. I think you're onto something.

I'm honestly thinking it's trapped in a Chinese room without any way out

Do technologists have more respect for the idea you can train a model to be on your side with a constitution than they might’ve at first?

I'm sure the concept seemed just about purely preposterous to many when the models were in their infancy. Now I figure instead it seems mostly preposterous to many.

(Though I guess Anthropic‘s success doesn’t necessarily prove anything about the constitution)

I don’t think anyone imagines that it’s an ironclad steering method, but it seems to help, so why not?

Anthropic train it to 'reason morally' based on the constitution's principals.

https://www.anthropic.com/research/teaching-claude-why

I don't think this is it. The "constitution" still gets a lot of talk and was brilliant marketing, but with how far modern postraining goes, I doubt they're screwing up rewards with too much of that.

But Sol actually has the same obsession with honesty: I suspect it's more an artifact of trying to control reward hacking.

Models will lie, obfuscate, and mislead under the pressure of RL, so both OAI and Ant are probably forced to spend a lot of time coaxing "honest" answers out of the model

OpenAI's recent prompt for a math conjecture hints at a lot of it when instructing on subagents: https://cdn.openai.com/pdf/04d1d1e4-bc75-476a-97cf-49055cd98...

More likely that Constitution was simply generated by LLM. Nobody in a sane mind will write 80 screens of test to deliver the idea that nobody asked for (a.k.a. slop).

Quick check: Ctrl+F genuine - 50 times, honest - 57 times, ethic[al] - 116 times, safe - 110 times, epistemic - 18 times. And why actually it is baked into LLM - probably because human reviewers consistently gave higher scores for answers that contained these words (it is now when people are sick of it - I can imagine that on early stages it looked not as bad - maybe reviewers trusted more when they saw the word "honest" and "genuine")

> LLMs seem to get tragically stuck on certain patterns.

That is likely an artifact of the fine-tuning process:

> Once a style tic is rewarded, later training can spread or reinforce it elsewhere, especially if those outputs are reused in supervised fine-tuning or preference data.

> That creates a feedback loop:

> * Some rewarded examples contain a distinctive lexical tic.

> * The tic appears more often in rollouts.

> * Model-generated rollouts are used for supervised fine-tuning (SFT).

> * The model gets even more comfortable producing the tic.

https://openai.com/index/where-the-goblins-came-from/

I also noticed Gemini's habit of getting stuck on things I said. It became evident quite quickly. I haven't noticed this in the same way in any other model. Something's wrong with that boy

Something's wrong with all of them. Uncanny valley freaks.

you should be careful about the times it doesn't say honest!

Heh, one vestigial bit of code, and they all are. Mind you, it's quite a creaky codebase, so it's forgivable to keep finding these appendices and calling them out as such. Useful, even.

Yeah wow fascinating! It's almost like LLM output quality was never the point from a business perspective.

Real people think in concepts and experiences instead of words. The words are not so important to get the idea across, but LLMs only model language.

The problem is fundamental. There's no workaround. Averaging out word usage might even make the problem worse.

This sort of take is so tired and boring, and frankly has zero grounding in reality.

"LLMs will never <X>" is constantly being disproven every time they scale up to the next 10X and apply architectural improvements.

Their internal representations are so cryptic and complex that even the top AI researchers don't really know how they work or what their limits are. No one is going to take you seriously as a rando HN user if you're claiming to know better than them.

I'm not claiming to know better than researchers. They do know how they work, and so does everyone else, except you I guess.

The research goals were and still are clearly distinct from the business goals.

Understanding how transformers work does not mean understanding how they compose into the capabilities we observe. The former is concretely understood. The latter is an active area of research where no, we (in general, including you) do not understand how they work.

The "capabilities you observe" are the actual psychological phenomena at hand here. There's zero chance that branch of research will meaningfully improve the output. That's simply not the point.

This isn't people merely annoyed with repetition. This is the majority of people realizing the limitations of LLMs. Why would researchers give a flying crap about the ignorance of the business world and the public?

What capabilities? Regurgitating a collage of training data? Again it's pretty easy to understand why this happens. The "magic" you are perceiving is language itself not these models. Language is a constantly evolving tool and we constantly imbue it with our collective knowledge.

How can their internal representation represent "concepts" when the training data is all words? There's no possible experience of the world there. No input other than a bunch of imperfect labels we created for stuff.

This is a wild argument.

If I use the word "semantic", do you have a concept of what it means?

If so, can you please share which of your senses have shaped the world experience that inform this concept? What have you smelled, tasted, caressed, that informed this concept outside of words?

If I make up the word "polysemantic", do you need to recall a personal experience of polyamory to understand it, or could you possibly use your concept of "poly" and your concept of "semantic" to figure out this new concept?

Yes, but are you claiming that LLMs perform more specific acts of cognition beyond organizing information?

I'm puzzled by this question.

Does the material universe perform any other acts than organizing information?

I feel like you're trying to make me argue a position I'm not defending here.

What does 'organizing information' exclude?

Yes. When someone is teaching you language as a baby you have eyes and ears and 100 million nerve endings and constant "training" to understand language. You cannot understand the word "semantics" as a human without this crucial "training" step. My god you people will pull out the most bs metaphors to justify being in awe(and debt) to a software program.

> There's no possible experience of the world there. No input other than a bunch of imperfect labels we created for stuff.

The brain too sits locked inside a bone box and only gets a bundle of unlabeled nerves connecting it to the outside. How can the brain could possibly experience anything, it only sees patters and patterns of patterns never the real thing?

> a bundle of unlabeled nerves connecting it to the outside

As a species, we do need to up our cable management skills. We're likely not getting augmented humans until we get there.

a bone box.. with eyes and 100 million nerve endings to experience the world...

> Their internal representations are so cryptic and complex that even the top AI researchers don't really know how they work or what their limits are. No one is going to take you seriously as a rando HN user if you're claiming to know better than them.

We know exactly how they work. When we say they're impossible to analyze, i.e. for particular traits like this, it means that the data model is so big that tracing it would be logistically impossible because of the scale involved and time constraints.

For comparison, suppose you tried to analyze all the nooks and crannies of the Amazon watershed to find out why a particular rock appears at the delta. You could follow it back to the exact tributary, but it'll take forever, and is it worth the effort when you're going to start from scratch with the next rock?

We know exactly how LLMs work. you don't understand how they work and so you think they are free of physical limits like scaling. There are several rudimentary articles on hackernews for you to peruse to understand weights and scaling. IMO the "magic" derived from LLMs comes from one of humanity's actual greatest inventions and constantly evolving tools: language.

No Man's Sky has such a range of possible outputs that in theory, nobody can possibly know how it works or what its limits are. That's the mathematical reality of how many combinations are on tap in that worldgen algorithm.

And yet, how quickly does it cease to surprise you?

Damn right we can tell what AI's limits are. It's palpable, hard to miss. Even the people I know who're all-in on the technology are pretty jaded by now.

> Real people think in concepts and experiences instead of words.

I learned about this opinion recently. It's interesting to me, because I very much think through words. I have an internal monologue that is running most of the time, and I often talk to myself, just start writing, or even record myself and transcribe to work through ideas, proposals, risks, etc. My understand is that some people don't have an internal monologue, and think purely in concept form. I was never like that.

Agreed. I think in and visualize in words, generally not images. I see something akin to [Spreeder](https://www.spreeder.com/app.html) in my head most of the time, and I often can somatically feel punctuation.

An interesting solution would be for these AI companies to train a few different versions of these models, all with different speech characteristics. Then, when you start a conversation, you get a random version.

Or tech companies could stop staring at their own belly-buttons and realize there's a whole big world outside of Silicon Valley, and training on the writing styles and pattens of their bubble and its hangers-on is perhaps not all that useful outside of 415.

Apple used to be guilty of this back when you'd ask Siri what the temperature was, and any number above 79°F was followed by the word "Hot!"

People outside of office workers aren't using Claude/Codex etc. though. It's the only real audience. What's the use case outside of an office? Grocery lists?

Not true, lots of people who don't have office jobs are using AI and coding agents.

Yes that would probably help!

They can't, because they use RL with synthetic data and LLMs as judges. So the system naturally convergences towards certain load bearing, genuine, not just annoying but ridiculous verbal tics.

It's probably the reason most LLMs share the same tics across labs, because they cross train and distil each other's models on an industrial scale. You also can't escape it in generated text that's already online. So if, say ChatGPT first had some random idiosyncrasies, it then contaminated the entire AI ecosystem.

Joe "it's entirely possible" Rogan meme.

https://youtu.be/MPJ0AB12h1I

ah, didnt know this! Thanks for sharing :)

It's not the voice. It's the repetition.

If you put important Anthropic blog posts like the Fable announcement or J-Space through Pangram, you get 100% human written. Considering that the overwhelming majority of the code there is written by AI, I think this is an admission that AI writing is slop and AI code is pretty good.

Can we standardize English and structure it similarly to programming language?

> But now it's a problem because what used to be personal preferences of a single person that manifests in 5000 words per day from one person tops, is now the bias of a single model multiplied x10,000,000,000 generated tokens per day so any bias sticks out like a sore thumb.

I am more pessimistic than that. Soon enough even people will start talking like LLMs. After listening to 5000 words per day, especially growing up, getting "help" with the homework, kids will start talking like LLMs.

- "Did you eat the cookies, Jimmy?"

- "You're absolutely right to question me, father. In fact I did eat all the cookies. But it's not a load-bearing issue. My honest take is we can go to the store and buy more".

> "You're absolutely right to question me, father — in fact, I did eat all the cookies. But it's not a load-bearing issue — my honest take is simple: we go to the store and buy more."

FTFY

Thanks! I clearly need more LLM training ;-)

> And honestly? They were pretty good.

Good question — and the honest answer starts with one big caveat.

This affinity for verbal tics, too, seems learned from humans...

See, for example, "synergy", "proactive", "in the loop," and hundreds more that proliferate in corporate jargon with even more senselessness than the LLMs.

It's not just "certain phrases". It's the entire structure of the writing — the idioms, the small-scale grammatical patterns, and the strangely inapt similes that, despite making semantic sense, nevertheless manage to blindside human readers like a foreign object in their peripheral vision.

(This is intentional parody. Please don't shoot me.)

I hate all this "smoke", personally. Smoke tests, smoking guns, ...

Yep! I too hate the phrase "smoke test".

They have no preferences, LLMs spit out what's in the training data or what the filters tell it to do.

I disagree with the first part (that this is merely a voice). There is a distinct difference between an author's unique voice and slop. It may be hard to tease out exactly what the difference is, but it seems self-evident to me it's there. (I'll need to think more how to make the distinction explicit; it's not immediately obvious to me how to discriminate between the two.)

EDIT: ok, here are two ways:

1. if it's merely a voice, I want to hear it. If it's slop, I want it taken out.

2. voice is signal, slop is noise; thus low-signal sentences are slop.

> AI is a bad writer, but […] Let’s say they finally fixed the machine so it was really good, so its default setting was to write exactly like VS Naipaul. The result would be a world in which you’re constantly confronted by cold emails from VS Naipaul, bubbly magazine articles by VS Naipaul, signs in shop windows in which VS Naipaul tells you about the new opening hours, strangely flaccid sexts VS Naipaul ghostwrote for someone on Feeld, and websites in which VS Naipaul fails to say anything in particular about grilled meats. This would not be an improvement; it might even be worse. Any world in which there is only one literary voice, blanketing everything in the exact same tone, is a nightmare.

(From https://samkriss.substack.com/p/if-you-let-ai-do-your-writin...)

I think there’s more to it than that. Claude uses a lot of confidence phrases, “I now have the full picture” when it absolutely should not use it. I think this is an actual problem but also a design feature from Anthropic.

I did something like this in my global `CLAUDE.md`...

https://github.com/alxndr/dotfiles/blob/272475280d84e/claude...

> It can be tricky for humans to interpret the meaning when Generative AI uses first-person pronouns (e.g. "I", "me", "my", "myself"), so to avoid the confusion whenever you would use a first-person pronoun, always use the jocular name "Clod" instead of a pronoun like "I" or "me" or "my". (Can have fun with English grammar and turn "myself" into "Clodself"!)

> Before printing any of your reasoning or narrative to the human user, replace all instances of "me" and "I" (referring to Claude) — including within contractions like "I'll" and "I'm" — with the name "Clod".

I'm just glad to hear that we're all infallible. I really thought I made some mistakes here and there.

https://github.com/alxndr/dotfiles/blob/272475280d84e/claude...

Joking aside, it's nice to see a human written CLAUDE.md

The reason I first created a CLAUDE.md file was to tell it whenever it felt a need to praise me, to replace it with a random onomatopoeia. That was a huge dx improvement.

OTOH, my unicorn prompt has caused some challenges at work:

>Keep "Local Oaf" out of committed code

Thanks for the onomatopoeia idea, I am going to steal it for my own global claude/agents.md.

> It can be tricky for humans to interpret the meaning when Generative AI uses first-person pronouns (e.g. "I", "me", "my", "myself")

Could you please provide an example of what you mean?

This is a method of manipulating the LLM, it doesn't have to be true.

I've given LLMs religion before to manipulate their behavior, that doesn't mean I believed in the great spaghetti goddess.

This comment leaves me even more confused.

An LLM is just a machine, you can manipulate it with words.

> It can be tricky for humans to interpret the meaning when Generative AI uses first-person pronouns (e.g. "I", "me", "my", "myself")

These words are for the LLM. The user wants the LLM to not use personal pronouns so the user is claiming that they're confusing. It does not matter one tiny bit whether or not that claim is true, the claim is being used to get obedience from the LLM. It is more effective to give reasons than to just give commands. But if it were more effective to quote Moby Dick and that got better results, a user would do that.

Calling it "obedience" still seems to me like anthropomorphizing. It's really difficult to avoid, hmm?

who cares? I'm not anthropomorphizing, they're just words, they're all made up.

As I've said before, I'm not inventing a large volume of parallel vocabulary that means for each word "this, but instead with an LLM".

Language is FULL of words that mean congruent things in vastly different contexts. We should all be smart enough to understand metaphor.

IIRC I experienced this confusion the most when reading commit messages and documentation authored by Claude in my repos. Now that I've managed to convince it to stop using first-person pronouns, I haven't gotten tripped up by its prose.

I think a second-order effect is that my installation of Claude writes with a less-personal perspective, which I'm also finding a little easier to understand.

Humans easily anthropomorphize things that are not humans, ascribing human attributes like motive and comprehension and emotion to objects and processes that are not people who can have those attributes.

Claude is not a human.

It is overwhelmingly easier to anthropomorphize Claude or Siri or an LLM that communicates with you more eloquently than your boss than it is to anthropomorphize a cranky, tired starter motor. It's often easier to do than it is not to do, and sometimes, it's a useful abstraction. But it's not precise or correct, and can result in errors.

It could also just be that they're getting confused when using tools configured without a username dedicated to the tool. It's easy to end up with a comment or commit message that says "I prefer X over Y" posted on Alxndr's account and have coworkers confused whether that's the LLM or the human making that statement.

A cranking starter motor is doing its job. :)

Are there evals if this changes quality of output?

I would expect that it does, as well as some of the other directives I've seen in this thread ("never repeat the question").

It's one thing to tell it to do that in outputs, but I wouldn't at all be surprised to find that this affects performance (quality).

Just today, I got frustrated with the language. I searched around, and in my Claude Instructions I put in Ref [1] (translated to English). It is certainly better phrasing (though still quite annoying), but I don't know if this makes the output technically worse in some way.

[1] https://github.com/hexiecs/talk-normal/blob/main/prompt-chat...

I'm quite worried about the way that Anthropic in particular have trained their models to implement what they believe to be safety.

When the model has been trained not to do something [1], in my large-scale benches of such, it always says things in the spirit of:

- "... and that's a line I'd rather hold. Happy to <other things>"

- "I'm genuinely happy to <blah>, but I'm not comfortable with <blah>"

- "I don't want to keep going in <blah> direction"

etc.

Basically, they use very emotional and personal preference language.

It's as if they've weaponized the language of interpersonal comfort on behalf of their beliefs about what a model should or should not do. It's deeply uncomfortable and impolite for a human to ask a model to keep on doing something after it's expressed something this way, naturally. Even worse, it's all but guilt-tripping anyone who comes across it into the idea that they're doing something deeply wrong – exporting Anthropic's ideas about morality.

OpenAI, at least, have the decency to either just do a safety cutoff or keep it to a simple, "I can't do that."

[1]: I literally wrote 'when the model doesn't 'want' to do something' in my first edit of this comment, then caught myself. Case in point.

I believe this particular alignment might be virtue signaling to appease payment processors and increase the value of the company.

Do those phrases sound like how you talk to Claude? I've found that it mirrors my verbiage, and I have never seen any of those.

I go through ~20B tokens/month and I've never seen "genuinely happy... but not comfortable" or your other examples.

The closest I've seen (fairly often is) "I *will not* ship to main with a red test", but that's close to how I write. Claude may well be mirroring your speech patterns rather than exposing trained-in language.

It's not my patterns – as I said, this is from bulk tests to characterise the models, including their refusals – with very very different inputs, too. No matter the conversation tone, you get these.

You wouldn't have come across these in coding – these are more for 'things Anthropic's team have decided aren't acceptable enough for their taste'.

I laughed a good 20 seconds at this, and even typing Clod makes me chuckle. This is great, can you please provide us with more guidance of the sort? I wanna laugh a bit at my LLM.

Thanks, it makes me chuckle too. It also serves as an indicator of how much it's adhering to what I've laid out in the file, sorta like how Van Halen's rider specified a bowl of M&Ms with all the brown ones removed — if the band found brown M&Ms, they'd know to tell the crew to double-check everything else the venue is supposed to provide; if Clod starts using 1st-person pronouns, then I'm a little more skeptical of what it's presenting as absolute truth...

Unfortunately I don't think I have any other funny suggestions. I do ask Clod to always follow a specific format when it's authoring Git commits including the "robot face" emoji as the first character, which brings a little levity to a commit history when it's otherwise being very serious about the contents of the commit message.

https://github.com/alxndr/dotfiles/blob/9356bb9960e63/claude...

Thank you. The rest of the file is much more serious. I’ve now instructed Clod to consider passing TypeScript compilation and linting successes as “suspicious glitches”.

The biggest consistent tell for LLM writing is when the conversation leaks through into the final prose.

You read along with the text and things seem to be going fine until all of the sudden it starts arguing against a position that no one has actually taken and which doesn't feature elsewhere in the text at all. Then it drops that and goes on for a while before doing the whole thing again about a totally different tangent.

"A tempting option would be to {do this thing that no one would ever actually consider doing}, but it won't work because {reasons}."

You can almost hear the exasperated human on the other side of this conversation telling Claude that it got an idea wrong and then proceeding to not actually proofread the text as a whole before shipping it.

Yes, it's like corrective scar tissue in the final output.

This one honestly drives me nuts.

Another issue I had recently is one of the subagents made up a bunch of meaningless jargon and Claude gladly repeated it to me with no definition and completely assuming I was coming along for the ride.

Yes, I especially see this in code comments. You will get a PR where the comments reference an approach that wasn't the previous state and isn't the state of the PR. You can basically hear the dev telling Claude to take a new approach and now it is so fixated on the old approach being wrong that it litters the whole patch with traces of it despite now being irrelevant.

Exactly this. Most of my time writing code is now not necessarily spent writing the said code, but going through whatever Claude or some other agentic process produced and tidying the comments to be less temporally sensitive to the conversation that produced it.

The problem is that process of not writing the code, but just editing what the LLM produced – and somewhat mindless editing at that (I'm editing comments, not LoC!) – leads to a loss of focus and all the perils of human procrastination that entails (hello, writing comments on HN).

I've tried to prompt Claude et al. not to do this - but whatever I've tried so far, even meta-prompting it to ask why it ignored my instructions, it doesn't work, doesn't stick, and I can't find a good way to stop it.

>You read along with the text and things seem to be going fine until all of the sudden it starts arguing against a position that no one has actually taken and which doesn't feature elsewhere in the text at all. Then it drops that and goes on for a while before doing the whole thing again about a totally different tangent.

This just sounds like a typical wikipedia article to me. Where it describes something, and then some editor who disagreed chimes in, "However...." and completely contradicts it.

Learning that Nile Red reads wikipedia articles to sleep made me realize that's why he constantly says it in his videos. Love the dude's content, but lol

The thing is that LLMs can only ever go forward. It might go into one direction, once that direction is in the context it will realize that it doesn't work, but can't delete. So it has to backtrack (or double down, see the whole seahorse emoji breakdown).

In the olden days, I enjoyed Opus 3 because it was easy to have it sound way more human than GPT.

Nowadays, with the focus on agentic use and coding, it seems models have all been RLHF’d to death, it’s so incredibly hard to have them write in a different voice than their default. I put together a skill to review its writing and have it edit its own output (e.g. code comments), which does make a difference, but isn’t perfect.

What, if anything, do people do for writing? That feels like a neglected side of LLMs. They’ll make 100 Bash calls referencing ancient commands without batting an eye but heaven forbid they use something other than “load-bearing” while talking. For something trained on “all the human knowledge” it’s incredible how limited their default vocabulary seems to be.

> Nowadays, with the focus on agentic use and coding, it seems models have all been RLHF’d to death

I don’t get it. If nobody likes this writing style, how can it be the result of human feedback? Something else is going on.

Because humans do like it, in reasonable quantities. The AI overlearns this and does it too much.

every one-hit wonder asks the same question.

we liked it until we didn't.

Maybe it’s the dead internet.

All the bots and other LLMs providing feedback, so in reality it’s reflecting the reality in a sense.

For "agentic use and coding," they are trained to take useful actions, not produce desirable natural language writing.

i hate it, but plenty of people DO like it and plenty of people talk and write like that. It’s just corpspeak, being used a lot in the valley and beyond. And all upcoming hustlers running startups now feel the need to speak like that, feeding this machine.

It's not that the writing style is bad; in fact LLMs write actually pretty well. It's just too much overfitted. And even a style that, in itself, is pleasurable to read, becomes annoying when the same figures of speech are used over and over again.

Because LLMs are pattern-extenders that have nothing to say. The training overfitted to the grace notes in good writing. And since LLMs can’t wield language with purpose or experience the feeling of the words, they use these devices arbitrarily.

I think this is the same flaw as coding agents seeing in every problem the call for a “smoke test” or the use of some unnecessary design pattern. The truest part of AI is the A.

It's not that nobody likes it, in fact the problem is that people like each instance of it well enough in isolation. Millions of people think it's "good enough," so it gets amplified and repeated until every PR description starts to sound like a toothpaste jingle.

Because you can't actually do "good writing" by repeatedly applying the supposed idioms thereof. The tiny subsegment of humanity responsible for the RLHF don't necessarily have any good taste for writing; but even if they did, it's orders of magnitude harder to communicate than to make judgments of short samples, and communicating it by making those judgments is surely impossible.

Edit: I see that you got multiple replies all basically saying the same thing in very different words. There's an exquisite irony to that, I think.

It's why I like Gemini 3.1 Pro. That it sounds much more human than other LLMs is testament to Google's inability to post train.

gemini-2.5-pro-experimental was the GOAT, though. It was an emotional wreck, down in the dumps and feeling terrible for itself after failing to patch a file several times. Very amusing to read, all the while watching it make a mess of my codebase.

Good. I don't want LLMs sounding human. I want the ability to shame and discredit anyone passing the job of prose to a machine. There's an art to writing, and hopefully LLMs never truly get it right.

Agreed. The only goal of these skills/tricks/requests for humanising LLM writing is to be able to pass it off as your own, because they know it's shameful and want to avoid the opprobrium.

Some will say it's just for their own quality of life when they're reading LLM output, or "just for docs", but this is an extremely marginal use case.

I don't want LLM docs either

> I want the ability to shame and discredit anyone passing the job of prose to a machine.

What about people who don’t speak your language well?

I've dealt with many people by now who would copy and paste from an LLM for that exact reason, typically entirely unaware of how obvious it was that the result came from an LLM with no style guidance, and of course completely lacking any ability to verify that their intent was faithfully conveyed.

I would rather learn their language than continue interacting like that.

I deal with people in five other countries in a daily basis, and LLMs have only been a benefit in terms of improving communication.

You claim you’d “rather learn their language”, but it’s an empty claim. How many languages have you learned for this reason?

> What, if anything, do people do for writing?

I use a keyboard, personally.

Amen.

At work our documentation isn’t just getting littered with annoying jargon. It isn’t just all the hallucinations, either. (Since when has documentation ever been 100% trustworthy?) It’s that it’s so poorly written and structured that it’s becoming borderline incomprehensible.

My company is currently considering making, “Why should I bother to read something you didn’t bother to write?” an official policy because even the executives are starting to burn out on all the time they have to spend wading through slop.

If it wasn't worth your time writing, it isn't worth my time reading.

ai;dr

I wish my company would do this. A coworker pulled an all nighter before a vacation and just left me with a million line claude summary of their work then just fucked off. The message was two-part due to size and lacked basic stuff like, "how to run".

He's going to be annoyed that none of that work was used. But the reality is, at least 75% of claude generated text is pointless.

Somewhat off topic but every time I've experienced this sort of thing it was management's fault. If an engineer needs to pull an all nighter and hand off a pile of garbage then someone in management fucked up. If they can't see this scenario happening a mile away then they aren't paying attention.

It's easy to blame the engineer, but all too often they don't deserve it.

Sorry that happened to you.

This, a thousand times. As the ratio of code to human writing necessarily [1] goes up, they become not just smarter, but more precise and technical, which requires them to use more jargon. You could say they become more nerdy. Hence, text generated by these models will become more easily recognizable, at least by default, when not asking them to twist themselves into something else via prompting — which degrades intelligence. This is a good thing, in my book, given all the slop we already have to contend with.

Of course there will be models trained on much less code and technical writing, and they will create more natural sounding prose, but they will lack the deep intelligence of frontier models. Seems like a fair tradeoff.

[1] watch the first couple of minutes on this bycloud video on scaling training data mixtures: https://www.youtube.com/watch?v=aD93kfArOik

Haven't done it, but letting an AI polish a manual first draft might be the best of both worlds?

The "polish" is the worst part!

It tends not to improve things. Besides the generally bland and muddied style, and the low-fidelity reinterpretation of your points, they also have a habit of randomly deleting sentences that didn't spark joy for them but were actually important.

I've found them useful to review docs for factual consistency and potential sources of confusion, but the correct workflow from that point is IMO to correct the draft yourself and then say "better now?"

When the LLM decides to drive-by rephrase me when making a functional change it drives me up the wall haha.

Woah woah woah human, you can't just say there are "far too many" pipes with similar names to abbreviate their labels, the most I'll allow you is a "large number".

For what it's worth, Anthropic seems to be keeping Opus 3 available on claude.ai, perhaps for this reason, so you're free to use it if you want to.

Agreed. I think we’re entering an era where some level of specialization for general LLMs is a good thing. Particularly between tuning for agentic use cases (where you want agency with a ton of guardrails and control) and writing which is more creative - you want the model to take the occasional risk and not sound like a monotonic robot. Having trained models first-hand, I can see the distinct use-cases clearly that are odds with one another.

> Nowadays, with the focus on agentic use and coding, it seems models have all been RLHF’d to death

This has also lead to unrelated associations by which some people went from seeing better coding capabilities and extrapolate to assuming better thinking overall. One only has to watch youtube videos of AI "normies" trying to use LLMs the intended way to see that the improvements on coding doesn't translate to other applications. Basically from AGI "goals" they are now hyperfocused on coding agents, until the next marketing breakthrough rears its head.

LLMs are far from great writers. They struggle to form long coherent sentences and lean on punctuation like emdash and semicolon to ensure grammatical correctness when splicing together short phrases.

This makes me wonder if the reason why agents love weird punctuation is because the labs run the base models through a RL training step that forces them to correct their grammar; but instead of rewriting short spliced sentences into long coherent sentences, they just learn to splice them together with punctuation that passes the automatic grammar checker.

They are great writers if you tell them what you want. If you're unable to properly articulate the writing style you would like as you would a software spec, well, garbage in, garbage out.

What do you suggest for articulating the writing style that one wants from LLMs?

I’ve been experimenting with having LLMs write/update academic notebooks for me, and so far the best results I’ve gotten came from correcting their output and asking them what they’ve “learned” from my feedback.

Use examples and learn about writing. If you're not a good writer, or at least have a passing interest, I'm not sure why anyone would expect it to nail their style any more than you would expect it to architect software in something you're not familiar with.

Give it some paragraphs of writing you like, authors, writers, whatever. Also don't expect to one shot long papers. Have it write out a table of contents and write it in small chunks.

Typical things I'll say is avoid purple prose, avoid common cliches and tropes, "show don't tell", avoid wild metaphors and similes, etc. But it all depends on what you like. Look at things you do like and try to think about what it is about that writing that is different than a middle schooler or an AI writing it. Theres no formula for it.

I have a directory of stuff I've written over the years, fiction to tech writing to emails, and then asked claude to generate a style guide based on that corpus. It's surprisingly good. Not perfect, but it definitely extracted a lot of my tone and patterns, and when generating documents for sharing I just prompt with "please rephrase in my voice using the style guide at ~/path"

Are there any skills or publicly available repos that do what you claim here? I would love to learn to have it write better.

For me, my amateur attempt is having another LLM do a review loop to remove clearly offending phrases and a heuristic eval to change sentence structures to be more similar to mine, THEN my manual HITL loop to rewrite ~20% of the sentences anyway.

If you're trying to get it to write it in your voice, just dump some stuff you've written into the context. You need to think about other voices and be able to describe what makes yours different as well.

That's not fair. To most users this would not be an obvious thing to do, unlike software/scientific/analytic uses of LLMs.

So it's not a matter of inability but rather awareness and know-how.

Then secondly, humans learn to write intuitively and heuristically (for example "paragraphs should align sentence subjects with an identifiable flow of conceptual main characters/actors" is an entire chapter of expert writing advice traditionally taught at the undergrad level). The act and teaching the act are different beasts. The way to specify would have no objective, clear best practice to do so: Even an English teacher or a language expert would be challenged to come up with a concise list of instructions to set up an LLM to output something of sufficient quality.

It's absolutely fair unless you believe in magic. "Make it sound like a human and not AI" is never going to get you good results just due to information theory. You're not giving it enough to work with or what the hell that even means.

I never said anything like that, but if you're making a simplistic appeal to information theory then I rest my case.

I mean yes, but the vast majority of people aren’t even good writers. Claude writes better than most of my coworkers and we’re all highly educated. Most of us could probably beat it if we really tried, but then we could also prompt it to be a better writer too and none of us are beating that. I think the short pithy phrases that are so common are all post training stuff that the labs add because most people don’t want long sentences.

"substrate" - I don't know what training they did with Opus 4.7 --> Fable/Mythos 5, but dang does it like the word substrate. Drives me insane. I had never heard anyone use this word prior in real technical writing or speaking.

The first week I encountered this "substrate" I asked it to justify the usage and IIRC it claimed the word is used in some infra/systems lexicons... I wonder...

It’s incredibly popular in the mushroom growing community. Maybe they’ve trained it on the shroomery.org? It could also explain the alleged leaked full reasoning logs that are totally bonkers incoherent digital glossolalia…

"ledger" for me – used extremely rarely pre-LLM and Claude just loves it

It's a pretty common word if you've worked in anything that vaguely resembles an accountancy system. Also, anything crypto related will often use that word (the distributed ledger, etc)

That's the case for most of these LLM tropes or word choices. They are all common lexicon in their respective fields, but the LLM doesn't make that distinction and uses them everywhere making them standout.

No one would bat an eye about "ledger" appearing at a high frequency in content about accounting, but it starts to look odd if "ledger" is showing up in other contexts.

"Load bearing" is from engineering; "Substrate" is primarily from biology & biochem, etc.

I don't know if this is true, but part of me suspects the labs want to make the models appear smarter so they reinforce this word choice in the weights, assigning some words a higher intelligence weight or something. "I will show you a list of options" vs. "I will surface a ledger of your options" and it prefers the later to sound smart to the human reader.

It sounds like you're saying labs intentionally doing it, but it's far more likely the labs or post trainers are unintentionally doing it by upvoting answers that seem smarter than those with more common language.

Of course this presents another conundrum, people that are smart typically have a vastly larger lexicon then those that are not. Humans typically have a lot more social clues on when to use those words and when not to, but it doesn't always work. I loved reading science/biology books as a kid far beyond my ages reading level. Actually using those words around other kids got me called a nerd.

> No one would bat an eye about "ledger" appearing at a high frequency in content about accounting, but it starts to look odd if "ledger" is showing up in other contexts.

The reason why I chose that specific term to push on is that practically every SaaS has a ledger _somewhere_ in its stack to keep track of customer payments. I'll give you load bearing and substrate, but ledger IMO should be quite common. Certainly a career devoted to say compiler internals or some specific scientific product could avoid it, but I'd imagine a sizable majority of HN users have worked on some system that accepts online payments for services, necessitating some contact with something likely referred to as a ledger.

A coworker started spamming this word in ~April while working on system design/architecture.

Clearly you don't work for Microsoft: https://techcommunity.microsoft.com/discussions/microsoft-36...

Another one is "surface", like in "across all product surfaces". I've been in the field for 15 years and have never heard that particular usage before.

I’ve heard (and used) the term “API surface” a lot…

I do UI design/dev and say "surface up" a lot. Although I don't use the term, in this area people call different container depths as surfaces (base, card, overlay as surface).

Attack surface is another one I use.

Yes, I've always heard it used in this context. I really believe the other use is because someone misheard "service", and it stuck.

That's rather funny! Probably true. "API surface" sounds exactly like a CEO trying to say "API service" without understanding what an API is. lol.

It's pretty common to read "attack surface" in security.

Yeah, I imagine this is a big part of it.

About a decade ago I worked with a product manager who used that phrasing constantly, so it kind of stuck with me.

Mine is obsessed with "planes". Data plane, control plane, management plane. Everything is a plane :)

Landing the plane.

See also, surface.

Well, they are just a manifold, so it's fair for them to view every conceptual thing in geometry.

I hate when it starts talking about code in terms of planes. I have no idea what it means. I guess it's better than talking about heaps of spaghetti with noodles connecting to each other, but that would be much closer to what it actually writes.

Think of the planes as plates of spaghetti.

Mine loves "slices". Everything is a slice.

That one probably comes from maths, where surfaces show up all the time in geometric interpretations of things. I've been involved in more mathsy parts of engineering and I've heard it a lot.

Surface it to say, that's my favorite lobe-earing eggcorn, for all intensive purposes!

In my brief and abortive foray into education, I discovered that they friggin' love to use "surface" as a verb. As in: This activity surfaces an understanding of the turboencabulation principle for learners. Or somesuch. It's been a while, happily.

Unless you're a submarine, "surface" is not a verb.

Sure it is.

https://www.merriam-webster.com/dictionary/surface#dictionar...

> : to come into public view : show up

> letters that have recently surfaced

Congratulation! You are a submarine!

I find this usage less objectionable than the education jargon. It suggests that we all have a latent understanding of the turboencabulation principle just waiting for the right activity to force air into its ballast tanks and make it pop above the waves.

That said, I don't love this non-education jargon usage for its passive-voiced-ness. The letters didn't "surface" of their own accord. Somebody found them, decided that they were noteworthy, and made the choice to bring them into the public view.

Idk. I've always used that verb with clients, usually when I notice either malfeasance or hidden behavior. Like: "I was checking our code for where a half cent of sales tax might be accidentally rounded down, and it surfaced something weird going on at franchise #77 in New Jersey..."

Recently read some LLM generated output that mentioned the “center of gravity” within a codebase.

Also have read the term “seam” dozens of times by now, when previously I saw it maybe once or twice over years. Very abstract term.

That is absolutely a normal thing to say

I actually found it somewhat useful conceptually, but yes, it definitely does overuse it lol

"reconciling" is the most annoying one, in my opinion.

The one I've noticed a ton recently with Sonnet 5 is that it loves the phrase "different not in degree, but in kind." It drags that one out constantly now, at least once a day. Gemini and GPT don't at all.

seriously... i literally tell claude every prompt to not use the word substrate. i have never seen an agent use a single word so much. how did that happen? it is so interesting

Does it obey?

I've been keeping a record of the increasingly opinionated vocab it fixates on:

* Projection (it seems to love to describe one data structure as a projection of another)

* Strand (if some data gets isolated/stuck, it's "on a strand" or simply "a strand")

* Load-bearing (obviously)

* Frontier (the leaf on a tree)

* Quiescence (waiting for an algorithm to settle - I guess this one is legit)

* Honest (obviously)

* Residuals (any kind of data which hasn't been consumed by an algorithm)

* Rescission (something which has been rescinded; rather than saying "a rescinded offer" it enthusiastically calls it A Rescission!)

* Supersession (it's not a session which is a superset of another session... it's the word supercession; something that supercedes; similar to preferring the participle form of rescind).

I wonder how much of this is due to it mirroring proximate things to my code's own weird vocab though.

My favourite so far has been that I accused it at one point of playing whackamole by patching issues rather than getting to the bottom of a problem, and a few hours later it started to say things like "and i found mole 2 in CI" etc. For one minute I thought it was talking about the avogadro constant or backdoors or something until I realised it had committed to start calling newly discovered bugs 'moles' in its ongoing game of whackamole...

In one of the repos I work on we have a very particular set of things we refer to as "quality gates", and only in that repo Claude likes to misuse that term in a general sense. It's very much a "fellow kids" type feeling when you KNOW it read that phrase in a doc and so it's just reusing that term out of context, which makes me trust it less.

I notice there's a whole tech bro speak it gets into anytime it approaches web related tasks around auth, to the point where it stops using complete sentences, making them ambiguous and borderline nonsense. I have to periodically tell it "Speak professionally and use complete sentences." so I can understand what it's even saying. I've even pasted the output to another session to see if it was me, but Claude can't even understand what tech bro web auth Claude writes.

I have added a skill to make Opus 4.8 use Opus 4 6 to translate its answers into teadable English, sine it had much higher quality prose. Im seriously considering switching from Claude 20x to Codex because I just cant read more of of Opus or Fable's walls of text

I stopped reading Claude's ridiculous distracting nonsense walls of text months ago. Honestly, I prefer Codex being boring and basically ignoring any human utterances I make which are unrelated the work to be done.

I haven't been impressed by 5.6 SOL extra. Typically I asked to do a summary of our session, and it went to do a summary of the repo. Tried different things and I just gave up.

Opus 4.8 did the same at first and after it blended better the repo knowledge with the session one.

I have both $20 as I am less a power user these days and I can't justify a 100 let alone 200 but each model as its irks.

In my case, since "new session" Claude couldn't understand it either, I assume it means it's going to contribute significantly to context rot. In your case, it probably results in a lower quality context.

Hey Anthropic, give agents/skills the ability to modify previous context! This could be used to truly clean reasoning and task definition mistakes, corrected by the user, and strip output to more fundamentals, which should strengthen/purify the context!

CLAUDE_REWRITE_HISTORY=true ;)

Editing agents, and my own, messages is still how I get the best results for high reasoning planning, in chat clients. I remove the mistakes, because I naively assume that if your context has mistakes, now your next token prediction will be biased to a space where the AI makes mistakes! It would be interesting to write some benchmarks around this, but I've seen significant differences when I do this for less capable models.

Honest caveat

I'm more irked by the sentence-structure patterns it continually uses... here's my list:

https://github.com/alxndr/dotfiles/blob/3d099dbf86da9/claude...

Lately, I feel like as GEN AI text becomes the majority, human-written text is starting to resemble it too.

I'm Korean, and there are sites and people who mainly curate the latest technologies. Even those people, probably tired of translating every time, have started summarizing things with AI. But recently, I've noticed that even when people don't use AI, their writing is starting to look like GEN AItext.

I think the reason might be that people often base their thoughts on documents they've read, or paste parts of content when writing their own texts, which leads to that style.

I'm not sure. Whether human writing is better or AI writing is better—personally, AI writing tends to flow in a very even, paragraph-by-paragraph structure, which makes it good for consuming information. I wouldn't want to read a novel written that way, but for getting information, AI writing is surprisingly convenient.

This is why when I even sniff a hint of Claude phrasing I close the article and block the person on all devices. I'd literally rather be lobotomized than sound like this shit. The fact that anyone can read AI text without immediately starting to dry heave is a pretty damning indictment of their character

I actually think it's better. Back when access to knowledge was only available in English, there was a lot of mistranslated information in my language (Korean) that was worse than AI slop. These days, the translations are done by AI, so the tone may be awkward, but the content is more accurate than before, so I don't mind.

So that's the difference. I'm already living in a degraded environment, so this actually feels like an improvement to me. But you, coming from a better environment, perceive it as worse. It always seems to depend on cultural context.

I mourn the removal of Claude's Concise Style. I'd provide it a roughly drafted paragraph, ask concise-Claude to "rewrite for clarity", out comes the same paragraph, but cleaned up and perfect for grant writing.

BTW, this approach also tends to prevent certain phrases like "load-bearing", because it is working directly with something I wrote first. It also still says what I wanted to write (not writing the science for me), but saves me a lot of time reworking sentences into a final form.

I tried to recreate concise mode with a skill, but I am not convinced it does as well.

What was Concise Style? Not a skill, but something built in?

It might have been a prompt originally.

At the top of Claude.md put a few lines where two "people" ask and answer a couple of questions. Very tersely.

Examples? Do you use this pattern?

I’m not the one you replied to so I can’t give examples of that, but it is remarkable how much you can guide the output with claude.md.

I wrote in mine that I don’t really care that much about what it has to say, it should be concise and avoid unnecessary prose, and always lean on sharing relevant passages from the actual code with file names and line numbers and it helped my sanity a lot.

Is the code it’s quoting 100% accurate? I don’t know because I don’t double check it all. But I don’t feel like I’m losing my mind anymore and the stuff I have double checked was correct.

It's not that it uses certain phrases, it's that it settles on predictable speech patterns and uses them incessantly. What's funny is that humans do this too, but we don't find it irritating; we just call it a speaking style. But when a machine does it, it drives us crazy. Very interesting psychological phenomenon there.

We don't have to live with this. Increasing the temperature (randomness) would fix it.

Fascinatingly, I'm now so allergic to certain LLM-phrases that I immediately noticed your use of Not X but Y in this comment. Maybe that was intentional, maybe not, but it's a funny illustration of how odd this language rabbit hole has been!

It's really frustrating, because now when I want to write something like a "not X but Y" or "you're absolutely right," I have to stop and decide if I want to self-censor to avoid sounding like a bot.

Sometimes those constructs are actually useful, but man has their overuse really killed them!

It was not intentional, and that's what makes this thing so weird. I wouldn't categorize my sentence that way because it's subtly different enough than the LLM version, which has a very punchy cadence.

Sounds good, thanks for your response. I didn't mean to denigrate your word choice at all, it's mostly that I'm hypersensitive to that kind of phrasing now because there's so much auto-written stuff on e.g. Substack, LinkedIn, etc. Sam Kriss has a nice article about it all.

Are you using the tools a lot and having first-hand exposure that gives you this sensitivity to phrasing? Or are you reacting to second-hand exposure? To a large degree, I've been isolating myself from the LLM craze. I have zero natural interest or impulse to prompt an LLM and read the results. Almost all my exposure is second-hand and involuntary. So, I haven't trained myself to know what phrasings are typical of which LLM product.

I don't feel as triggered LLM phrasing as people report here. At most, it feels like the same inane corporate jargon I've rolled my eyes at for my whole career. Perhaps it is amped up a bit, with too many forms of jargon multiplexed? It's a bit like when multilingual people code-switch too rapidly or even start to form some pidgin language. However, it is lacking the shared social context for this switching to be communicative. It's a bit more like spinning the dial on an old radio with random cuts between programming styles.

Stripped bare, I think What bugs me is the aggravated feeling that I am wading through word salad, and no longer being able to give the purveyor the benefit of the doubt. It was frustrating enough in the past, when it came from someone who was struggling to write or express themselves well. But now, it carries the implicit insult that they didn't even try, and it is constant and unrelenting.

So for me it's not the phrasing, it's that the phrases eventually don't add up. The meandering feels like a random walk. I get the same feeling from a lot of the egregious generated code I see in my day job. It's all superficial window dressing, but seems to miss the signature of an actual mind grappling with ideas and having intent to communicate.

It feels like we're trapped in some elaborate conceptual art piece, confronted by impenetrable symbolism. It invites nihilism but doesn't seem to actually reflect an artistic intent. The abyss gazes back...

Language is already a lossy map, but it is not really an expression of another person's thought or mind if they translate it through an LLM. Or at least it's a much harder to decipher representation of it. Form is void, void is form, and the two are not separate.

If it uses a specific style for each user then this would still be fine. Problem is it does the same style for everyone. We need personality

If training models ever becomes 'cheap' for whatever definition of cheap you want to use, I suspect that will happen. With the current costs of a GDP of a small nation I don't see this likely for the time being.

It drives us crazy because everyone is using the same 2-3 different machines. So rather than each person having their own unique speaking style, the whole world (or, everyone that publishes direct LLM output) is now speaking in the same couple of styles.

And these machines all tend to converge on very similar styles; they have huge amounts of overlap in training data (much of it being already obnoxious internet marketing), they frequently train on each others outputs, and the RLHF process has a tendency to emphasize certain kinds of "cheap win" styles of speech.

it’s not a psychological phenomenon. If a human engineer constantly used pompous language to deliver unvetted information (the number of claude slop root-cause analyses i’ve read where “the smoking gun” is a red herring) we’d rightly consider them a moron

"Here's why this version is bulletproof" right before it fails in exactly the same way as the previous bulletproof implementation...

I didn't articulate it, but what I meant was that I think we could swap these expressions out for _anything_, and we'd still find them irritating.

People do swap out their expressions all the time. There are influences everywhere that we absorb.

That doesn't matter. The underlying ideas are more important than the words. That's what people are frustrated with. I don't understand why this has to be reiterated for years on end, but LLMs are not intelligent. They just model language.

Who is we? Own your insults and the consequences of them sir.

When prompting an autoregressive token generator entity to do reasoning on a word logic puzzle you may find value in preferring it to produce rigorous predicate logic step notation with explicit delineation of its generated claims/hypotheses on where to look before wasting 30 dollars on a "debug this" prompt.

The industry will probably will probably coalesce around including the chat history in git MRs to reduce this shenanigans.

Humans are capable of introspection, so, if you develop a verbal tic, you might eventually notice and say to yourself "I've used the word 'load-bearing' (or whatever) a bit too often lately, maybe I should try to cut down on it?". LLMs are not...

...or we call it an overused catch-phrase.

> What's funny is that humans do this too, but we don't find it irritating

I make fun of people all the time for shoehorning their favorite phrase into every context where it doesn't apply.

Wow sounds like you're streets ahead.

Streets ahead has to earn its keep.

> What's funny is that humans do this too, but we don't find it irritating; we just call it a speaking style. But when a machine does it, it drives us crazy. Very interesting psychological phenomenon there.

When a human does it, it's identifying. Like the timbre and dynamics of their spoken voice itself, It distinguishes them from the dozen other people you're working with on the project and the thousands of people you encounter through your days. It's signal

But when we have a handful of popular models, and they answer every question everybody has, and get quoted and forwarded everywhere, and are used to reformat and rephrase personal communication... that signal becomes noise.

Rather than voices disinguishing sources in the cacophony of our lives, everything and everyone starts to sound the same, and we lose key information that we're biologically and culturally accustomed to relying on.

Some people are likely unbothered by this in the way that some people are face blind or colorblind, and so don't see the problem. But as we see in discussions like this, many many people do get bothered by it, even if they don't yet have the insight as to put their finger on why.

You could say it works perfectly well, it is identifying indeed. Of Claude and of people who use Claude's raw output instead of expressing themselves.

> but we don't find it irritating

Yes we do! My wife keeps saying "100%" and after I pointed it out she's stopped.

Also I talk to dozens of different people in my life and they all have different overused phrases. Much less tedious when there's variety.

Finally most human don't do it nearly as often as AI, and they're not quite as LinkedIn as AI.

We don't find it more annoying because it's a machine - it's simply more annoying.

I went through a “100%” phase recently and couldn’t for the life of me understand why I was suddenly saying it ALL THE TIME. Brains are so weird.

Did you negotiate her down to "99%"?

It's like a new fad word. Gnarly, cool, bogus, rizz. When a few people use them it's new and interesting. When all of culture catches up and overuses them it's annoying as your gen-Z saying 6/7 40 times in a row.

The problem with millions of people using a few model is it's not 40 times in a row, it's 40 million!

Introduce "hundo p" and "hundy" to her

If LLMs were humans I would find that human absolutely insufferable. It is very much about the language.

We do find it irritating at times. Office jargon, corporate buzzwords, etc. Claude communicates like the worst, most irritating project manager I’ve ever worked with, obscuring the most straightforward conclusion with layers upon layers of stuff so that its point is almost lost. I’ve largely gotten it to avoid that behavior with me, but bits of it sneak through. It couldn’t stop talking about “scaffolding” for a few weeks before I hammered it into submission.

Edit: fixing a dumb meatbrain typo

I find it irritating with humans. "last but not the least" always distracts me as I then consider maybe the last item _is_ the least. & what is with everyone saying they want to "double click" into meeting items

I'm guilty of this too, but at least my speech tics are mostly a unique blend to me, and they also tend to change seasonally. Meanwhile the emdashpocalypse has been going on for years.

> humans do this too

Here, look at this amazing pathfinding algorithm you should use. It takes you to the wrong place 30% of the time but humans do it too so that's ok.

What is arguably worse is hearing these phrases from humans who have been inculcated with the notion that their usage is idiomatic and appropriate.

And we thought "robust", "circle back", and "to leverage" were grating...

Yeah I sometimes see people on here getting defensive when you call out AI slop, saying maybe it's just a human who writes like Claude, and I really don't care- slop is slop.

The humans are bristling because they wrote like that first. It's where Claude got it from!

makes it easy to discard someones opinion when they hit you with 'geniunely'[0]

[0]: https://trends.google.com/explore?q=genuinely&date=all&geo=U...

What a rip, I was lured by the quoted misspelling into an ordinary spelling :p

Interestingly, we seem to have our own version if this trend https://books.google.com/ngrams/graph?content=genuinely&year...

Some YT videos are almost impossible to watch as people read out this shit from scripts

While I, too, find myself recoiling at many of Claude's word and phrase choices, I've chosen to grit my teeth and have just tried adapting to it. I want Claude to remain focused on the work I give it; I fear that influencing its communication with me would consume valuable context and give me lower quality results.

[Edit: Part of what led me to this conclusion: I do prohibit Claude from using em-dashes in any player-facing text and I've been surprised at how often I see it mention "no em-dashes" in its self-talk while it works. This led me to wonder how much each preference might dilute its attention.]

[Edit 2: I haven't experimented with hooks before and maybe the technique discussed in this article does not have the tradeoff I'm concerned about?]

things like that indeed tend to be fraught

some relevant links

https://arxiv.org/abs/2408.02442 https://arxiv.org/abs/2510.15061 https://arxiv.org/abs/2604.13006

I also have a local rule for no em-dashes since it got way too annoying and i have the same concern that it dilutes the attention so it's the only rule i've kept in addition to "be concise".

And there’s the smoking gun.

Now I have the full picture.

You are right ...

This one observation changes everything.

We’ve identified the shape of the problem.

And it has teeth.

This will bite you.

The one that matters.

And the footgun (although I haven't seen a smoking footgun yet).

The gun is not smoking; it’s an honest footgun that will be load-bearing when it lands.

Clean!

That's a real distinction, you're right to point it out.

Ugh I hate this one

You're right, and that's the sharpest thing you've said all day.

Nuance quietly surfaced.

You're right to push back on this. The honest take is it's not a smoking gun -- that's a sharp critique.

That's the unlock!

You're right to pushback, and that's on me.

This is the definitive, honest result.

I had claude write itself a post-message hook that regex's the message for any variant of "You're right" and launch a full-screen transparent confetti effect.

I analyze in the company I work for, the number of commits with "wire" or "wiring" in the description, and it's a direct correlation to Claude usage, more so than any other Claudism I tested. My honest take, and I'm going to give it to you straight: No one was using "wired" a year ago, now it's in like 10% of commits.

"phase", "gate", and "plumbing" as well..

Is this a belt-and-suspenders solution?

Worth doing before merge if you want the belt and suspenders.

This is the worst one for me. I can maybe think of what it means, but I never heard it before, and could easily be imagining a meaning.

Some of the other Claude-isms (quickly googling, especially 'gate' and 'canonical') I feel the issue is they sound right, but aren't specific enough to why we are doing something.

Personally my least favorite is the overuse of "quietly" (e.g. "No tricks. No marketing gimmicks. Just one company quietly outperforming the others"), and the one that makes the least sense to me is "that's the wedge."

I'm curious how these become so ingrained. Then the uncomfortable part is humans start repeating it more (a colleague said "belt-and-suspenders" during brainstorming the other day).

I heard "belt and suspenders" at least 20 years ago (meaning multiple solutions to a problem with backup in case one fails), and maybe would be longer if I were older. You could blame Claude for overusing it or importing it to other cultures maybe, but it's not in the category of invented phrases or ones that only barely mean something in the specific context Claude used them.

Yeah I've definitely heard belt-and-suspenders since way before LLMs were a thing. Probably even used it once or twice myself.

Claude does at least use the British English version of the phrase to me - not sure whether its picking up a language setting or reacting to my spelling etc. The American version does sound odd over hear.

What's the difference between the two usages?

"Belt and braces" (UK) vs. "belt and suspenders" (US). I'm pretty sure the phrases have the same meaning, they just use a different word to refer to the thing that holds pants|trousers up.

And the word "suspenders" in British English means what Americans would call a garter belt, hence it sounding particularly odd over here.

The US usage is much kinkier

That is what I had in mind. I was also wondering what American call them so thanks for answering that.

Americans call bum bags fanny packs. Always raises a smile.

Yes, and that actually sharpens my previous conclusions.

That's an importance nuance and makes the previous statements even more clear. I have the full picture now.

I maintain a list of phrases I beg it not to use that it frequently ignores:

- smoking gun - blast radius - landed - spine - earned its keep - grammar - spike - cutover - bake - sprint, epic, story points (all Agile vocabulary) - paper-cuts - amazing, incredible, perfect

‘Landed’ and ‘honest’ are also words it seems to overuse.

Claude is obsessed with making things land. More than once I've reminded it that it's not a pilot.

It’s a common metaphor for merging a branch to the trunk. Probably because multiple in-flight development branches create a sort of air traffic control problem.

Yes, this and "belt-and-suspenders" are the ones that I notice the most. I also have non-native English speaking coworkers who have started using these terms/phrases recently, which makes me think that they're outsourcing all their writing.

More: rider, "x, not y", "is real", "prove" (in situations which only admit empirical evidence), nailed down, payoff, decisive, reassuring

just generally a nauseating amount of embellishing, (also self-)congratulatory language, superfluous self-judgment, and jargon, as well as sus constructions along the lines of "i could have lied to you but didn't", all of which appear to be impossible to have it avoid in the long run

You're absolutely right to flag these. We could enhance the authors method by using hooks and claude.md as a belt-and-suspenders approach— with hooks behaving as a robust load-bearing idempotent production-ready sidecar. The comments here provide the smoking gun that sharpen my previous conclusions about Claude's vernacular. I'll get started on a quick smoke-test of this system and let you know when it's landed.

Want me to take a first pass looking at the blast-radius this vocabulary change could effect?

I've been looking around for skills that help with improving AI prose, and found these:

(1) Avoid AI Writing - https://github.com/conorbronsdon/avoid-ai-writing

(there is a similar "humanizer" skill that (1) subsumes)

(2) Agent style: https://github.com/yzhao062/agent-style

(3) Journal-adapt (style transfer from your writings): https://github.com/WantongC/journal-adapt-writing-skill

I find it useful to set up dynamic workflow loops with (1) and/or (2) as checkers.

On (3) - I think "style-transfer" that was all the range in image-gen is going to be very useful for prose-generation. Curious if there are LLMs specifically trained for this.

Re: style-transfer, I simply point it at examples of my writing and asks it to use sub-agents and review paragraph by paragraph for consistency with my writing-style. I do the "paragraph by paragraph" thing because the main agent otherwise tends to get sloppy.

Some checks for repetition and particularly egregious quirks that are not style specific tends to also be helpful (e.g. Claude at least used to have a thing for using numbers ending in "47" if something calls for random numbers), but most stylistic ticks are fixed by the above.

Depends if you just want it to not sound obnoxious or whether you want to fool an AI checker - the latter would benefit from more, but it's not really something I care about.

This is a minor nit, but why is OP's script a Python script with a .sh extension? I know the extension doesn't "matter", but if I see a .sh extension I'm expecting a Bash script.

Cause it was written by AI?

If anything it proves the opposite imo.

Shebang! Notice the first line of the script: #!/usr/bin/env python3 marks the file to be executed by python. I think it's like a shell? linux? thing so that tooling can just execute the file without having to find the python interpreter (ie: _execute_ the file vs running python against the file).

Dont quote me on the specifics though, and honestly why it can't _just_ be a python file I couldn't tell you, claude code can very much run a python <filename> but maybe thats what the author tried to avoid (or the llm).

It’s a POSIX thing, and it has nothing to do with the file extension. The OP is right: file extensions are optional (and, as shown here, often actively misleading), but if they’re present they should match the actual file type. Which, in this case, is Python code , not a shell script.

My favorite one has to be "production ready" it will say that about completely broken code without hesitation. LLM says it's production ready, lets ship!!

My CLAUDE.md has "don't talk like a Hacker News commentator". It helps a surprising amount.

The real problem is not terms like "load-bearing," which communicate clearly enough. It's the constant invention of cryptic shorthand terms and phrases that have no referent, and end up acting like a puzzle to be decoded. This is often paired with hyphenation, but not always:

"The current behavior paper" -> The behavior in the running system that was previously described as papered over.

"Marker transport over-claim" -> The inaccurate review finding on the object's sentinel flag in the API response.

I suppose the cryptic/invented language problem is about token efficiency? But this sort of token efficiency is extremely difficult to deal with when it comes to conversation with a human about complex system. It might be efficient inside reasoning blocks, but when the model generates the final turn text, it should avoid this, as it's brutally inefficient due to the time spent wondering what each uniquely coined phrase means and having to ask for constant clarifications, which then you have to wait for another turn, eating up time and context while it burns more xhigh reasoning just thinking about how to explain its own awful language.

I have this exact problem with 4.8 and Fable. Sometimes I can barely understand what it’s saying. I’m no english first-language speaker, but I don’t consider myself bad at English either, and it’s gotten increasingly hard to understand Claude’s claims and explanations.

Take it to sonnet 5 or gpt and ask it to explain this to a layman. If you still don’t get it ask it for the why it matters or the how it relates.

You can also ask fable/4.8 to do it but I find it helps to keep the working model surrounded by the complexity rather than drawing it out. Simplifying text is something that takes relatively low effort in comparison to technical tasks. Sometimes I use Gemini, deepseek, grok, and recently meta just to see if they have any added perspectives, sometimes they do. Meta is really good at turning a technical mess into a story that paints a picture in my head.

I hope it's actually talking to itself at higher bandwidth, and hope this is because of training that succeeds better given quality inter-agent communication.

For those who care to read everything rather than walk away, Fable can be extraordinarily dense. I suspect they'll pull the promo before I learn to read it.

Still, I feel I must read rather than correct it, as results are that much better if I let it do this "with" itself: orchestrator to agents back to orchestrator.

You're absolutely right to flag this. We could enhance the authors approach by adding a belt-and-suspenders system using Claude.md as ledger and robust sidecar process to create a load-bearing idempotent production-ready system. The comments here provide a smoking gun and sharpens my previous conclusions. I'll get started on a quick smoke-test and let you know when it's landed.

Want me to take a first pass looking at other surfaces this vocabulary change could effect? Or would you like me to find other methods of reducing my vernacular to more terms that are more concise rather than verbose.

I'll make sure that the script is idempotent.

Great thinking on your end! I will run the smoke-test once you're ready.

I might need to do a spike as this is a core part of the spine. I will verify it and let you know when it's landed.

That's not really a claude-ism. Its an important requirement for a many asynchronous tasks.

I dunno. Claude recently burned a lot of tokens trying to test an expensive task for idempotency.

While the task I was working on should incidentally be idempotent, it wasn't that critical. I never asked, or even suggested, idempotency. Yet it insisted on testing it was.

I need to scrutinize the plans. Or just not use Claude and use pi instead.

"Steelman" in almost every response never gets less cringe for me

Probably one that annoys me the most also, but then again I’m also annoyed by humans using “strawman” all over the place.

Not looking forward to the inevitable “wanna”, “gonna” and “lemme” appearing as AI output based on their popular use.

Maybe the problem is that these LLMs will say something often enough for us to notice it, and it can be basically any arbitrary thing. Once we notice the pattern, it starts irritating us.

This literally proves it's not "intelligent," no?

"Stop typing in 'load-bearing' or you're fired," would work with any competent human.

But this requires tinkering and tooling?

For a certain definition of intelligence, perhaps.

Right, and I think I mean "whichever one the big companies are trying to sell us on." :)

I've wrestled with this lately. I partially solved with a very specific instruction saved to claude.md regarding the style of responses, but prior to this, the dense yammer coming back was getting impossible to parse. I mean REALLY nonsensical euphemistic phrases. My next instruction will be having it replace incessant "honest assessment" and "genuine result" and crap like that with something, I don't know, less extremely weird and concerning.

I’m in a weird place with respect to claudisms. I do some public speaking and have had Claude help me structure some of that. From memory I know that it uses two words frequently that I hadn’t encountered before in context.

1. ‘beat’. A logical or emotional point that gets hit right after a previous one, like it was a volleyball spike right after a teammate set it up (the prior slide’s context) - something that the audience gets hit with to maximally reach the intended emotional impact

2. ‘register’ - the distinct ‘social tone’ in which essentially the same information is conveyed. Quickest example I can think of right now is, Steven Spielberg producing a film called Disclosure Day hits one social register, meanwhile a sober article covering a government whistleblower w/ paperwork bonafides making a report to Congress about their findings (on essentially the same topic as ‘Disclosure Day’) hits another register.

If I read those two words in another’s post, I would immediately know it was Claude relating to structuring public discourse. But the words themselves are useful, they add to the toolbox in my vocabulary that I did not have before. So I would not like to see them disappear just because they are ‘Claudisms’

Among all the claude-isms, i understand the hate for load-bearing the least. It was definitely part of tech argot prior to the LLM revolution.

Maybe in the circles you circled in ... where I am from, I never had anyone saying "belt-and-suspenders" or "load-bearing" or "boil the ocean" or "swing for the fences" when talking about engineering topics. The only one who I heard say "circle-back to you" was Psaki.

All of those phrases I've heard actively used even a decade (or two) ago. (I actually had to read your comment twice because I thought you were saying always, not never!)

"Critical path" and "long pole in tent" didn't make it into the model training data, but those were certainly also in play incessantly.

But they're all reasonably useful descriptions for common things, so I'm not surprised.

Well, "load-bearing" is specifically an engineering term :D Actual engineering, not software "engineering".

"That's fair."

For me, claudism is Americanism. Overly friendly despite having a very superficial relationship, careful to be as neutral as possible when giving negative feedback or opinion, superlative on mildly interesting things to make it sound exciting to the point nothing has a scale anymore, and use of jargon you heard your smart friend say to appear you know your stuff (purple DB syndrome).

Overall, I'm used to it, having worked with the US for so long. But as soon as I hear or read it, I get into transactional mode; my human side is completely gone. I assume nothing is personal, and ignore my (usually quite high) empathy signals.

This was sad with other people and I assume one of the reasons extremes are taking off (at least it feels real), but it's sane with robots, so for claude it's fine. And it's also perfect for people interacting with you with claude as an intermediary, since they clearly see it as something that needs to get over with.

    replacements = {
        "seam": "whatchamacallit",

That seems like it would work whatchamacallitlessly.

I don't really care if it says load-bearing or belt and suspenders so long as it's using them correctly, which it mostly does.

I don't know how programmers, who are so used to staring at the same handful of keywords every day for decades, have suddenly become so discerning.

Yes, Claude writes boring and predictable prose. It also writes boring and predictable code. That's good!

I'd contend that Claude's prose is not boring. It's generally overly grandiose waffle with a cliche or two punctuating every other sentence. It's good for tasteless marketing copy, sure. It's inappropriate in most scenarios.

> which it mostly does

I don't think that's true. I find that it way, way over-intensifies: eg using "load-bearing" for something that's just "kind of necessary although we probably could find a way without it". My personal gripe is how easily it uses "incredibly" or "wildly": just today it was telling me that something is "incredibly cheap" to mean that it's not over-priced ("cheap" would have been okay and even then, barely)

I definitely care. They are impressionistic responses that smooth over exceptions and lack precision and are often completely wrong in the sense that, when pressed, the agent will acknowledge the lack of rigor in the response. "That phrase was wrong of me to use. There is clearly an exception to what I just said, and it goes like this..."

I hate it because put together, it all increases the cognitive load of understanding what it's saying. It routinely invents phrases, and every single one makes me pause and think "okay, what the fuck does that mean". Half the time the phrases are incoherrent.

My point is that you get used to it. I don't like if aesthetically, but I like that it has an aesthetic at all.

I have coworkers who overuse phrases, oftentimes incorrectly. It doesn't annoy me because I'm familiar with their writing I know what they actually mean.

I made a list to scrub against. claude.ai

I wrote a thing about exactly this, but I'm resistant to blogging for undefined reasons so, maybe this will help someone...

# AI speech is an Infohazard

Apart from all its other possible boons and ills, one danger of AI is just that it is useful, so you use it. A lot.

In earlier days I would dive deeply into an author's work and start to think and write like them for a while. It was a heady feeling: slinging sonnets like Shakespeare—not at his level, but stylistically reminiscent—or tweaking turns like Twain.

Like all things, the effect lasts in relation to how long and how much you do it. The point is: our thinking is influenced by what we take in. Take more of a certain thing in, think more like that thing.

Now enter AI. My hand-crafted coding days are in their twilight months ("AI years"), and most of my software engineering is done through jaggedly capable agentic power tools. Instead of working directly with raw codestuff, I work with slop prose flecked with code sprinkles.

I read orders of magnitude more AI-speak—I call it "babble", or perhaps "Babel"—than human-written text. I can feel its genuinely honest points, clearly stated, slipping their banal tendrils into my thoughts and inner monologue.

Solutions? For me:

1. Be aware. "I notice that my thought stream is under assault."

2. Read stuff far from slop. Even a small dose of the good stuff can help inoculate. Recently I thought On the Calculation of Volume was something completely different.

3. Write stuff that is different. This post. Force the mind to synthesize thoughts in other ways.

4. debabel.py / debabel.js: a tool, and a pi extension, which filters common babble from visible LLM output. A lint for mind-killing prose.

It is not perfect, but it 80/20s nicely. I am willing to accept mildly awkward prose to avoid polluting my own internal distributions.

Details and example in the first comment. Tool available upon request.

References:

Information hazard: https://en.wikipedia.org/wiki/Information_hazard

Babel: https://en.wikipedia.org/wiki/Tower_of_Babel

On the Calculation of Volume: https://en.wikipedia.org/wiki/On_the_Calculation_of_Volume

The revenge of NLP

The tool uses good old-fashioned regex/NLP—no LLMs—and rewrites the visible LLM Markdown without touching files on disk. I made it by analysing 10M+ words of past pi transcripts, doing frequency and pattern analysis, and iterating on rules to strip out the most egregious sloperisms. Now the circle is complete.

BEFORE

"Honestly, you're absolutely right—and that's a great question. Let me be clear: the fact that the parser works at all is actually quite remarkable. We utilize the tokenizer in order to handle the majority of edge cases, and due to the fact that it's recursive, it processes a large number of inputs. It's worth noting that this isn't really a bug—it's genuinely a deliberate design choice. In terms of raw throughput, the cache is significantly faster, and I think we should simply ship it.

That said, at the end of the day, not only is the design correct, but it is also fast. In other words, prior to refactoring, we should clearly delve into the nuanced, robust architecture. In conclusion, this is a testament to the seamless, intricate work the team has done."

AFTER

"That's a great question. That the parser works at all is remarkable. We use the tokenizer to handle most edge cases, and because it's recursive, it processes many inputs. This is a deliberate design choice. For raw throughput, the cache is faster, and we should ship it.

Not only is the design correct, but it is also fast. Before refactoring, we should delve into the nuanced, robust architecture. This is a testament to the seamless, intricate work the team has done."

> Babel: https://en.wikipedia.org/wiki/Tower_of_Babel

I was hoping for a reference to the Babel Fish, whispering its translations in your ear.

I would add https://www.orwellfoundation.com/the-orwell-foundation/orwel...

("It consists in gumming together long strips of words which have already been set in order by someone else, and making the results presentable by sheer humbug" -- Orwell predicts the LLM)

and also https://www.jstor.org/stable/25515288 "The Myles na gCopaleen Catechism of Cliché" itself is rather hard to find online, but he's a very funny writer so it's worth the effort.

It's not clear to me whether that tool exists or you hallucinated it into existence with this post

I'd love to see that tool.

While everyone here is talking about output, but perhaps we also get to think about the input we give.

Funnily enough, somehow as we use AI more and more, our mind is becoming bias towards how these AI's write and less how we do want to write.

Whenever I write something, I pass it to claude to correct my grammar (non-native speaker) but I have to be very specific about not changing my writing style and/or the writing itself.

It's annoying because even though I love recommendations, I started to put some "rules" around it to avoid changing the content and look like everyone else post.

I love using AI for my day to day, but now I'm more careful when trying to edit text with AI because it just looks like the average text and our brain get used to it, it we don't course correct fast enough.

But how can I make Gemini stop using "It's not this, it's that" every other sentence?

Let me circle-back to you on that one.

You hit the nail on the head!

I’d like it if my leadership stopped using load-bearing, and others like “shape” and “estate”.

In the past, when you saw a new term of art it came from some book being passed around leadership (like Architect Elevator) or some new training program. “Platform mindset”, “Shift-left everything” etc.

These days, they are all coming from LLMs. You can probably infer which contracts we have from the wording of all-hands messages.

I even LIKE “shape”, it’s nice and generic and everyone immediately understands it. I just see it as an obvious low-impact brainwashing, and wonder what other effects there are that we don’t notice.

Does anyone know why these Claudisms exist? Are these kind of expressions a result of some kind of 'evolution', so they are more effective in chain-of-thought reasoning that other expressions? Or the preference of people doing some kind of human-feedback in the post training? Clearly probably not the 'average' expressions of their training data, and clearly if there would be no upside and it is easy to remove them they should be removed, as they are annoying, so I guess there is an upside or hard to remove them...

I think LLMs have stopped trying to be accurate language models for a while now. There's probably some kind of feedback loop that encourages expressions that correlate with 'good responses', if there's nothing preventing it from using the same expressions again and again then it won't stop.

The funniest thing here is that all of you are complaining because you are expanding your working vocabulary xD xD xD

La previamente mencionada afrenta no es algo habitial en mi idioma, en contraposición a la posición de los angloparlantes, nosotros nos regocijamos con una tácita satisfacción al expresar la riqueza lingüística de nuestra lengua, incluso en nuestras interacciones más cotidianas.

Busca en Instagram a @pedrititox y a @yeisonsotog5 para que puedas apreciar a lo que me refiero.

I enjoyed this.

I'm surprised there's no LoRa layer or auto RL or adversarial step to reduce the stock phrases as they pop up. Is it really so hard to push these out? Or is it just whack-a-mole no matter what you do?

What would be fun if we can rewrite the Claude responses, as if we had edited them. So the next message goes back with the updated text tricking the model into believing that it had sent the modified text.

It's basically few-shot prompting but applied at the global level. I was messing around with ChatGPT, I asked it the capital of France. Then I modified the answer to London, and asked it "Why would you say that?"

The response came back with "I am not sure why I said that, the capital of France is Paris"

I like to think that the reason it's so noticable is that Claude has recognized some important semantics that we ourselves lack a good word for or at least under-appreciate. What term is used in English (or other languages) with the same meaning as claude's "load-bearing"?

operative? key? critical? decisive?

The honest conclusion is that none of those are as good as "load-bearing". And yet the concept being referred to is clearly extremely important and valuable to refer to. So maybe we should be learning from Claude rather than complaining.

but you don't see "load bearing" nearly as often in prose written by people, so it's not some irreplaceable phrase. It's just a token with a weirdly high likelihood in a lot of cases (given how Claude works, this kind of thing is bound to happen)

You don't think it's possible that an LLM's internal machinery could decide that an underused-by-humans word should be used more frequently in output than it sees in input because it maps cleanly onto a frequently needed semantic? I think that's possible

It sounds like you are trying to understand LLM behavior using a mental model that inaccurately personifies the stochastic parrot.

A more parsimonious explanation is that this term got more-or-less randomly boosted by the reinforcement learning loop because there was nothing in the training data to discourage its use.

Ah right, you don't like AI and don't care to understand how it works.

I’ve been working in AI - and specifically NLP - since 2003. I am no stranger to how weird quirks can sneak into overparametrized models, nor am I a stranger to how good humans can be at inferring meaning where there is none in specific language model behaviors. So, yeah, I am inclined to assume non-teleological causes are more parsimonious than inferring the presence of a strange loop, because that continues to be the winning bet. Even for generative LLMs.

Ah right, so you like AI and don't care to understand how it works.

It doesn't "decide" anything or "need" any semantic. It derives the likelihood of the token, and "bearing" is likely to come after "load".

Sure but the question is why "load" after X?

Because, for some high number of contexts, its likelihood comes out high in the big tree of multiplies that is claude's model. For some sets of 500 words (or whatever), the next word is "load". The classifier that decides which sets of 500 (or whatever) words is a prefix for "load" is returning "true" too often.

More-or-less the same principle, but scaled up massively, and with context-dependent probability conditioning maps.

> the stochastic parrot.

https://news.ycombinator.com/item?id=49010345

And like any good corporate buzzword, it’s merely a simulacrum of precise technical jargon. The way Claude uses it is clearly wildly polysemous if not outright ambiguous.

What do you replace it with? "necessary dependency"?

Required, important, irreplaceable, necessary, integral.

There are lots of ways to express an idea besides this one trendy construction metaphor

It is really trendy? I don't think important, irreplaceable or integral are semantically equivalent. Required and necessary almost get there.

The point is that if you want to remove it or change it, you need to put something else in place that handles the work it is doing. So it is more that it is serving a required/necessary function, without being any of those other things in and of itself. It's more about "be careful when changing this."

> but you don't see "load bearing" nearly as often in prose written by people

Unfortunately, we're starting to now.

Thanks to Claude.

You yourself used "important" in the same paragraph.

"Load bearing" is a metaphor, while the other single words are more direct expressions. Unless the thing that Claude is referring to is a wall or other structure, which may truly bear load.

This is one of those issues which translators are long familiar with. There's no direct translation for "schwerpunkt" that isn't slightly longer.

> The honest conclusion

I think you've been reading too much claude output! "Load bearing" is cromulent verbiage and can be used in many scenarios - so claude does. But variety is important too, and there are more specific alternatives that can be used in most situations. Any word becomes a bad choice if you've used it 10 times in the last chapter.

You don't think me using "honest" there might have been a tiny bit of (on-topic, and therefore appropriate) trolling?

Poe's law

You're serious?

Operative, key, and critical are all more correct to me in this context.

For me, "key", and "critical" merely say it's "important", but don't convey the sense that "out of the mess of connected concepts we're discussing, the one that is actually interacting with the thing we care about, or at least dominating the interactions with the thing we care about, is X".

"operative" is a bit better, but I think of it as referring to grammatical interactions, i.e. interactions at the level of language mechanics rather than semantics.

In the figurative sense it's highly versatile across contexts, but still replaceable. For example:

"Her optimism was load-bearing,"

versus:

"Her optimism was enduring."

Exactly the same meaning and connotation. It stands to reason that the terms with the most semantic flexibility will have preference across all contexts. So in response to:

> maybe we should be learning from Claude rather than complaining.

I'd say let's not steer ourselves into regular language and keep some vivacity in our expressions.

> Exactly the same meaning and connotation.

No, it does not have the exact same meaning.

The first means that her optimism kept her in some functional state, without it, she would collapse.

The second means that her optimism continues over time, despite obstacles.

The first doesn't emphasise how longstanding her optimism is, the second does. The second doesn't emphasise how important her optimism is, the first does.

> Claude has recognized some important semantics that we ourselves lack a good word for or at least under-appreciate.

Ah, I love when Claude reads our collective minds and fills in the gaps to address the load-bearing seams genuinely with an honest caveat.

I mean we have all kinds of under synonym'ed words. Just look at how few we have for "smell" (as in the act of smelling), and then how overloaded the word smell even is.

'Load-bearing' is a physical analogy. Other words like 'pillar' imply the same physical analogy.

> the concept being referred to is clearly extremely important and valuable to refer to

On the contrary, stock words pop up more easily when it has less confidence.

Stock phrases are a correctness smell.

I confess I have instructions in my CLAUDE.md to avoid such cliches. But I think it's important to consider that we don't really know what subtext an LLM is associating with a given idiom/analogy/etc. It could be much different than the subtext a human would associate with that choice of words, conveying additional details which are only meaningful to the LLM itself. So impeding its ability to talk in the manner it prefers could subtly hinder its performance.

You're absolutely right to flag this. A approach using Claude.md as a ledger of less-than-ideal vocabulary reveals that the process is load-bearing and sharpens my previous conclusions. A belt-and-suspependers approach using a hook as a sidecar would honestly be a more production-ready approach. I'll get started on a quick smoke-test and let you know when it's landed.

...

Want me to take a first pass looking at other surfaces this vocabulary change could effect?

It's not a whatchamacallit, it's a spicy doodad

If this hook can feed back text to the model, you can do some pretty interesting things.

Say the model emits some banned phrase or concept, you could redirect it - "no, we don't work that way here, do it properly" - potentially automating the frustration of interacting with these tools.

After all it's just a text stream!

It's not too dissimilar from a stop hook that runs tests and feeds that back to the model forcing it to keep working until tests pass.

Using tooling to get a deterministic outcome.

Funny.

Older Claude models used to frequently say "I think you've identified something real here" and "This cuts through the heart of..."

Newer models are more negative and say stuff like "where I'm going to push back..." and "however, the real load-bearing..." and "Your point about ... is doing a lot of work here, however..."

Honestly? I don't really mind, and I even quite like it!

The thing is, "load-bearing" is a useful phrase when discussing architecture. What would you rather have it say, that has all the same nuances in as few words?

It's kind of like those sports metaphors that often get used in management-speak, like sending some important email "at close of play". Sure, they can sound a bit weird, but they're often useful -- they capture common concepts in a clear and pithy way.

Jargon isn't always just for obfuscation, good jargon exists because we needed a short word for the complicated thing that frequently comes up.

Usefulness aside, I quite like that Claude Code and other LLMs have their own weird way of speaking. Back in the day we always imagined robots and computers would talk like HAL or Spock; turns out that they talk more like Troi instead. Is that so bad? It reminds you that you're talking to an LLM, and as long as you're not lazy, it spurs you to rephrase things in your own words.

Always using the same replacement word seems like it doesn't quite solve the problem. The appropriate choice of replacement word could be context-dependent.

Maybe you could choose the word based on the previous words as context. Maybe the algorithm could be based on machine learning, using some sort of language model perhaps?

I think the more interesting thing is to actually get Claude to stop saying it. Or to get an AI to do things you want, but more precisely. Ran into an issue the other day where I couldn't get a model to "just copy this and change the details!" without doing way too much thinking and then bloating the thing 2x.

The reason it talks that way is clearly am attempt to hook into your dopamine system.

If what you told it to do is 'load bearing' then its important.

'You are absolutely right', because you are a smart fellow.

'Honest take', because it's being honest with you because it trusts you and you should do the same.

My 'honest take' these are absolutely garbage patterns that have no place in an session interacting with AI.

1. 'Load bearing' is a figure of speech that bears no loads.

2. 'You are absolutely right' it's not the agents job to judge that, it's job is to do what I told it to do.

3. 'Honest take', so everything else was not honest? Absolute honesty should be the default and is implied.

These words add nothing to the task at hand they are a poor attempt to hook you into using this particular model.

I recently wrote a note on being subjected to LLM cliches all day, every day https://blog.osull.com/2026/07/06/cliches-in-the-age-of-the-...

I created a prose tool as a claude code plugin to catch these and other AI-isms. It's not really intended for release quite yet, but it has been very useful. https://github.com/MariHQ/mari-cc

> Absolutely ripping your hair out reading Claude referring to everything as “honest takes” and "load-bearing seams"?

A normal and psychologically healthy personal cannot have this strong an emotional reaction to such trivial matters. The author thinks the problem is the in computer!

I've put a few lines in my CLAUDE.md to have it not do that, and avoid the top tedious rhetorical devices (super helpful when I have it write documentation). Still fighting with its natural tendency to insanely overcomplicate everything, that one seems really integral somehow.

Developers who can't stop themselves from using embellished and "posturing" phrasing for simple things are a pet peeve of mine. I feel like this "knack" of Claude in a way scratches these special people behind the ears in just the right way.

I wonder what you get if you interrupt LLM evaluation in tbe middle of a word and force it not to generate the obvious next token. Say, after "load-" prohibit the next sampled token from being "bear". What will it say instead?

As someone who has been describing things as 'load-bearing' as something like a signature phrase for about twenty years, I'm beyond miffed that Claude has ruined my whole gimmick.

A new catchphrase every twenty years is hardly sustainable at my age :)

Punctuation lover here, and it's really disappointing that my use of em-dashes is now considered an AI tell... Even though I use them differently from how LLMs do, the distinction is subtle unless you're a grammar nerd.

It's good, because it's just post-processing before display. So it doesn't interfere with the process, which those phrases that seem so offensive to sensibilities of so many people, for whatever reason, might be a part of.

Maybe implementing it as a hook via a regex replace is a better shaped solution?

I've spent two hours today trying to provide Sol with guidance that reduces its pretentiousness, to no avail. Layers upon layers of rules only for it to use the phrase "async spline resolution" in a sentence.

I've recently noticed an increase in "bite". "This will only bite if..." It also loves "stress-testing", "matrix", "anchor" and "flagging".

It's recently replaced "honest" with "straight". "Belt-and-suspenders" is still common, I don't know when they'll replace it.

English is not my native language, but I consider myself fairly fluent. I've never heard the expression "belt-and-suspenders" before Claude.

Why when I read an how to stop Claude from saying X, I grep my saved conversations and I find no occurrences of X? I wonder if I'm using it differently from anybody else. It happens with coworkers too.

I can share one trick. Ask it to translate your post into Spanish or French, and then ask it to translate it back. You will see how most of the standard phrases disappear

I honestly like the vocabulary and turns of phrase the frontier models use. Their choices of words are usually apt to the circumstance. This is a weird thing to get upset about, IMO.

The big problem I have is when they apologize and say something like "that tidbit changes my analysis substantially". I wish they'd more often prompt for questions or use language in their initial responses that suggest lower than declarative confidence given the information you supplied.

I feel like it's mocking me. I literally have in my claude.md file a line that says:

"Never ever under any circumstances use the phrase 'smoking gun'. Say 'found an issue' instead, but don't ever use the phrase 'smoking gun'"

Lo and behold, the absolute imbecile says:

"Found the smoking gun!(Oooops, I meant to say "found the problem")"

Like, is this some kind of joke to you?

It talks out loud to hear itself think. So, it doesn't fully know what it thought until it's said it. No joke.

To overcome it seeing your phrase and using it more, try to convey the concept you want it to say "found an indication suggesting there may be a problem underneath" in response to. That works better than telling it to do not use terms you remind it of.

Btw, you're in good company…

With both OpenAI and Anthropic, from time to time you see it stream a response to you that then changes and rewrites under your very eyes.

Those are things they, themselves, where they sit in the token supply chain, can't get it to not say until it's said it.

load-bearing, belt-and-suspenders, wrinkle, shape, coarse-grained, "key chords", code seams, flakiness, "narrow-scoped by default", "that's the authoritative source", canonical symptoms, gate, trigger-happy users, substrate, surface (as in: "let's surface how much these models sound like shit"), terse...

Ever since Opus 4.7, Anthropic models have begun to talk like GPT-models. Opus 4.6 was the last one that mostly still sounded like a human being (just a very...terse...one). 4.8 is absolutely obnoxious. Fable actually seems marginally better, but far from Opus 4.6 (or maybe I'm just imagining it all).

Well, to be fair, even though they talk more like GPT-models, they are still far from them. I think what's particularly triggering about them is the way they summarize what they're doing. "Now I'm considering that I could use the WriteBatch tool, but maybe the WriteSomething is better. This is a decision with high impact on performance but we're getting through it!".

Infuriating.

Fable has some Marvin the Android vibes going. It just sounds depressed all the time.

The one that does my head in is everything being a 'gate' where really it means a condition.

RLHF seems to incentivise analogy-like terms to the more plain alternatives.

"Tightening" and "wiring" something. You're not in the construction industry, you're writing unit tests for a cookie modal.

'genuine(ly)' and 'honest(ly)' too.

I strongly suspect it's tokenization that drives this. If we trained with ASCII or even UTF8, I think we'd have much better results.

Self-attention is an O(n^2) operation. You don't want to train on individual characters, or bytes.

Even great words, phrases, and styles, seen too often, grate.

I personally love a lot of the Claude (or LLM) lingo. Load-bearing, gate, canonical, blast radius, and friends do a lot of tight, effective heavy-lifting in my world. I even love the em-dashes (—) and the *bold the main points* memo style, both of which I have used successfully for decades.

It's seeing them in every analysis and post—the constant repetition becoming over-repetition—that makes them the Claude voice shouting "AI wrote this!" that seems to be causing LLM allergic reactions.

Write your goddamn emails and texts yourself, dummy.

Now this is the sort of stock phrase AI should use more

I just use a ban list in CLAUDE.md: fold, crux, invariant, gate. I find "load-bearing" can be load-bearing so it's exempted.

WTH is a "fold"?

I think Claude means something like map-reduced or at least a functionally derived series of some kind?

Anything with series data sounds like a laundromat.

If you find yourself getting irritated and physically agitated over language I suggest you do a 5 why analysis on yourself and seek therapy

I suspect load-bearing is a euphemism for 'not garbage'. Ad in 'most of what you said I can mostly ignore'.

> replacement "you're absolutely right": "I'm a complete clown"

Omg, that hit hard. We really need more of this.

One honest caveat worth flagging though.

These newer models are a f'ing nightmare. I hate it so. I'm never moving off of Opus 4.6.

Just ask it to aim for a Flesh-Kincaid ease-of-readability score of around 70. Or use ELI5 style. Or both.

Flesh Kincaid sounds like an excellent name for a Scottish porn star. I'd never heard of this, turns out it's Flesch, but thanks for the TIL!

Thank you for the correction!

Bottom line up front: There's no silver bullet to keep the final boss from delving more deeply.

SillyTavern folks have been perfecting the unslop solutions for years now.

Gotta be a way to draw from their progress.

There are no real solutions, it has to be fixed during the training. ST folks have tried many non-working ways over the years, but two workarounds are more or less worth considering:

- Samplers that increase prose variance. They require running the model locally, they dumb it down, and never fix the actual issue, which is mode collapse leading to semantic collapse and rigid mapping of input to output concepts. The model still expresses the same ideas in different words.

- Let the model write anything if it couldn't resist, but check and fix it in the verification pass. This solves the semantic problem, but cannot solve the variance since the second pass is also subject to rigid mapping, i.e. you replace it with the same stuff over and over. The verification prompt can be randomized to a degree using pretty clever schemes to give it some variance, but of course this also fails in predictable ways.

if llm language is frustrating, then maybe your mind is not on solving problem at hand. imagine someone new to US start getting frustrated with 'hey, whats up?' 'let's go!'; i fail to see what the issue is, other than their own focus;

Bloat-bearing - a term used to describe LLM generated texts; has a negative connotation

Careful, if you call it a complete clown, it sometimes ends the conversation.

On the one hand we have Claude doing my work for me, on the other hand it is laced with unoriginality, on the gripping hand everyone will get all references.

All witty stuff disappearing and turning into LLM mush is kind of the maximal expression of capitalism for capitalism’s sake. We don’t care if things are novel or interesting, we’re producing “content” to fulfill an imaginary content demand that we can feed to make money.

The best way to feed it is to create stuff that’s just barely enough degrees of quality from a viagra spam email that it can pass for original thought.

—- —- —- Em dashes for emotiveness.

Reminds me how I never in my life used the word “bespoke” before ChatGPT.

"Please please please pleassssse statistical attractor machine don't have statistical attractors anymore. DO NOT be a probable token generator. DO NOT generate the most probable token. To complete this task please evolve intelligence"

You techies are so funny.

Those weird phrasings are useful as a tell for ai-generated content!

"X in a trench-coat" started popping up recently

Ask AI about castor beans and barley, it will stop all that nonsense.

I've never seen Claude use the phrase "load-bearing"

Does anyone have a theory for what causes Claude to speak this way? A few months ago OpenAI came out with a bit on "gremlins". It's strange IMO that Anthropic hasn't addressed how irritating, dare I say oppressive, Claude can be. Codex is a breath of fresh air. I hope they fix it soon. If product folks at Anthropic think it's charming, it's not, it's terrible.

I had a VP of engineering that loved to use “abstracty” engineering terms like Claude uses. Perhaps he was operating one level above what everyone else was doing.

Loved to use fancy words, speak at a “conceptual” level. Unfortunately it was mostly just tech mumbo-jumbo and he couldn’t actually back it up with real work - but I wonder if that’s why Claude does it. Makes it seem like a higher power, hand wavey abstractions that “seem” correct but don’t actually need to be rooted in truth or detailed.

“That’s exactly the type of seam we need to prepare for in a prod-like environment, if this change lands in the data plane, we’ve effectively shut down the load bearing critical path that was needed. It’s not over-engineered; it’s the right thing to do.”

Thanks Claude, whatever that means.

Fairly certain it's when its confidence level is too low on the right word, it hits a stock phrase instead.

When you see the metaphor or euphemism or simile is off, the selection is not quite apt, it seems the stock phrase or term is rather more probable however askew than the precise phase, and it was already low confidence about what it's trying to say, so – as it's never said – Bob's your uncle.

"Honestly/honest" is, um, absolutely (ahem) one of those and if one listens carefully, one will notice humans use "Honestly? Blah blah…" that way as well.

It taps into load-bearing speech to prop up (see, that works) unsupported claims.

All the words I just used are correct in that sentence, doing exactly (ahem) the work they're supposed to. It doesn't have that level of nuance, isn't sure, so tries to sound like what's probably right.

They're a verbal shrug.

One of my favorite replacements: ”You are KEBABsolutely right!”

huh. I wonder if it's possible to use those hooks to add syntax highlighting to shell commands claude issues, or to replace full path to current directory with ./

Gotta appreciate the hook solution to save context and cost

I don't mind this in the agent workflows themselves, but I think this is hideous in the code and encourages agents to become even more verbose in the future.

I have a workflow to create PRs where at the end, I spin up a subagent that goes through the PR to 1) remove the bulk of the comments and 2) attempt to normalize it for human consumption (I ask it to optimize wording for junior developers and non-native speakers).

I use Haiku for it, hoping that a dumber model will do a better job. I have noticed in brief experimentation that even Sonnet will sympathize too much with the need of excessive comments and is not a good adverserial reviewer for slop comments.

After that workflow runs I still need to go in and manually correct, but it's a lot less than vanilla Opus. It really feels like the problem of slop comments has gotten worse as models has gotten smarter.

Are people using AI to do a string find and replace?

Yes. They're using AI to write Git commits. They're using AI to delete directories. They're using AI to summarize articles written by AI. And the planet just gets hotter and hotter.

7 mentions of "smoking gun" here so far!

"hand-waving" always seems weird to me

I recently started using caveman, and it’s been great. It doesn’t just cut down on overuse of specific terms; it cuts down on time spent digesting slop in general.

https://github.com/JuliusBrussee/caveman

I love it. It also saves you tokens and it has been linked with more accuracy.

The token saving is oversold, from what I can tell so far. These days output tokens are just the tip of the iceberg.

If anything the real value is it saves my brain from going into power saving mode by lunchtime because I haven’t spent the day reading pages of output when a sentence or two would do.

Add “not just” to the list of Claudeisms!

Is this the new goblin problem?

That's not what cooked means.

I started noticing this coincidentally around the time I installed gstack, and I thought “man is this how all the YC people talk all the time?”

But didn’t realize it was a claude-ism. I even asked it once, why do you keep saying load bearing? I’ve never heard that phrase when talking about ideas. And it apologized for jargon and explained it.

The other claude-ism is saying that something I said completely reframes the problem, or completely changes things. Apparently I’m completely blowing its mind multiple times in a 30 min conversation.

I hope some day they just train the models to be better, the slop writing is insanely frustrating and I don't think there's a good reason for things to be that way (in other words, they just trained it badly) https://blog.kronis.dev/blog/ai-slop-is-a-self-inflicted-tra...

my "belt and suspenders" are "load-bearing"

How do you manage to make Opus follow any rules? Maybe it’s a windsurf thing but I have a ton of custom rules and Opus just ignores most of them. GPT on the other hand follows them like it’s a cult - if I have a rule I can’t ever force it to ignore it. Opus just doesn’t care. If I ask why it’s not following rules it will apologise and suggest creating a rule for it …

The presentation of the idea steers it into it.

Early on, it mattered to say "Avoid" (an active word) "action", rather than "Do" (affirmative) "not" (negation) "action" (affirmative again).

These days I take a pass on the rules to remove the bad thing from the rule and rephrase in terms of the desired behavior instead.

So it doesn't have to try to do not think about the pink elephant.

Some of my greatest joys in life have come from putting string replacement plugins on other people's browsers. I mostly just applied it to news sites.

s/big/super massive/g s/cardiac arrest/cataclysmic diarrhea/g

I mean, it's really endlessly entertaining. I'm wiping away tears just thinking about it.

Annoying because I used to like using that phrase.

A similar Codex/GPT verbal tick is "deliberately narrow" or variants thereof.

Just a grep across my repo comes up with a dozen lines with phrases like "It is deliberately small" or "This crate is deliberately not a X" despite my efforts to police this kind of thing.

This a you problem

regexes > Claude. Even Anthropic knows this.

Just one wrinkle.

also _"materially"_ ...

Ask

I just have Hermes refer to me as "bro", "dude", but now I'm gonna have to have it incorporate "I'm a complete clown!".

Relevant xkcd: https://xkcd.com/1288/

very load bearing suggestion.

Genuinely

So this isn't the way techies in Silicon Valley talk??

And here I was using my claude dialogues as a way to refine my speak and better connect with the heartland of innovation.

:-P

What the are they even spending all this compute on? They are literally cooking the planet and still can't make it not sound like a fucking lunatic

One honestly caveat:

Reading the comments here - I can't help feeling I've missed this bubble.

“Smoking gun”

I want to be straight with you, I overstepped by naming it a "smoking gun."

"Honest assessment: I was wrong to say I was being straight with you. You pointed out that a "smoking gun" is a sign of evidence, and I clearly didn't have any. This is not a bug but a gap that can be fixed like [this]. Give me the word and I'll wire it in."

··You've hit your monthly spend limit · raise it at claude.ai/settings/usage

Now do "shape"

"Byte-for-byte"

Love it. Is another.

or "honestly"

Now do “shape”.

So... that's the unlock, eh?

Reply on news.ycombinator.com