HN

Need help?
<- Back
LLM Writing Tropes.md

LLM Writing Tropes.md

walterbell

Comments (50)

  • Jordan-117
    Wikipedia also has an exhaustive guide, though it's not fun finding tropes you use yourself (I'm very guilty of the false range "from X to Y" thing):https://en.wikipedia.org/wiki/Wikipedia:Signs_of_AI_writingAnother one that seems impossible for LLMs to avoid: breaking article into a title and a subtitle, separated by a colon. Even if you explicitly tell it not to, it'll do it.
  • capnrefsmmat
    I work on research studying LLM writing styles, so I am going to have to steal this. I've seen plenty of lists of LLM style features, but this is the first one I noticed that mentions "tapestry", which we found is GPT-4o's second-most-overused word (after "camaraderie", for some reason).[1] We used a set of grammatical features in our initial style comparisons (like present participles, which GPT-4o loved so much that they were a pretty accurate classifier on their own), but it shouldn't be too hard to pattern-match some of these other features and quantify them.If anyone who works on LLMs is reading, a question: When we've tried base models (no instruction tuning/RLHF, just text completion), they show far fewer stylistic anomalies like this. So it's not that the training data is weird. It's something in instruction-tuning that's doing it. Do you ask the human raters to evaluate style? Is there a rubric? Why is the instruction tuning pushing such a noticeable style shift?[1] https://www.pnas.org/doi/10.1073/pnas.2422455122, preprint at https://arxiv.org/abs/2410.16107. Working on extending this to more recent models and other grammatical features now
  • joshvm
    No mention of Claude/ChatGPT's favourite new word genuine and friends? They also like using real and honest when giving advice. Far as I can tell this is a new-ish change.> Honestly? We should address X first. It's a genuine issue and we've found a real bug here.Honorable mention: "no <thing you told me not to do>". I guess this helps reassure adherence to the prompt? I see that one all the time in vibe coded PRs.
  • CharlesW
  • awakeasleep
    If this bugs you, open chatGPT personality settings, choose “efficient” base style, and turn off the enthusiasm and warmth slidersIt makes a tremendous difference. Almost everything on this list is the emotional fluff ChatGPT injects to simulate a personality.
  • oooyay
  • newAccount2025
    Great list. Invented Concept Labels is the one I think I get most frustrated by. When exploring new areas, I’ll read its paragraphs of acronyms and weird words and think I just don’t know some term of art, and as soon as I ask for a definition it’s like, “I just made that up, that’s not a formal term, blah blah blah.”
  • carleverett
    "The "It's not X -- it's Y" pattern, often with an em dash. The single most commonly identified AI writing tell. Man I f*cking hate it. AI uses this to create false profundity by framing everything as a surprising reframe. One in a piece can be effective; ten in a blog post is a genuine insult to the reader. Before LLMs, people simply did not write like this at scale."This one hit home... the first time I ever saw Claude do it I really liked it. It's amazing how quickly it became the #1 most aggravating thing it does just through sheer overuse. And of course now it's rampant in writing everywhere.
  • layer8
    As the article points out at the end, these aren't bad per se. The issue is that LLMs overuse them, and we're all getting the same(-ish) LLM. It's not so different from how people sometimes have their idiosyncratic phrasings they use all the time.
  • mvkel
    Weirdly, LLMs seem to break with these instructions. They simply ignore them, almost as if the pretraining/RL weights are so heavy, no amount of system prompting can override it
  • veqq
  • FartyMcFarter
    The article has been slashdotted so I don't know if this one is in there but:One I've seen Gemini using a lot is the "I'll shoot straight with you" preamble (or similar phrasing), when it's about to tell me it can't answer the question.
  • 1970-01-01
    What we really need is a browser plugin underlining these patterns, especially for comments.
  • anon
    undefined
  • layer8
  • bryanrasmussen
    I sort of think the whole middlebrow angst thread about Bourdieu going on right now applies to LLM writinghttps://news.ycombinator.com/item?id=47260028
  • anon
    undefined
  • bryanrasmussen
    This makes me think of the attractiveness of overly bad writing to writers, as a challenge, the most obvious example being the bulwer-lytton award, or the instinctive ignoring of instructions from fiction magazines that might say "we don't want any stories about murderous grandparents, French bashing, bestiality, bank robbers from the future, or kind-hearted Nazis - and especially do not try to be super brilliant and funny and send us your story about kind-hearted Nazi bank-robbing french-bashing grandparents that like killing people and having sexy fun times with barnyard animals! Because every original thinker like you thinks they are the first to have come up with that idea!" and then as a writer you feel challenged to do exactly what they say they don't want because what a glorious triumph if you manage to outdo everyone and get your dreck published because it's dreck that is so bad it's good!It does not seem like there are lots of people who are perversely inclined to write a story with all these tropes and words in it, but surely there must be some, because if you make something that beats the LLM (by being creatively good) using all the crap the LLM uses, it would seem some sort of John Henry triumph (discounting the final end of John Henry of course, which is a real downer)
  • roywiggins
    Don't forget "The Ludlum Delusion"- every header in an article or readme reads like a Robert Ludlum novel title, ie "The [Noun:0.9|Adjective:0.1] [Noun]".
  • xgulfie
    If only we could fix how it writes like garbage
  • netsec_burn
    Another trope: longer README.md's than anyone would make, or want.
  • bitwize
    You know how no one ever wrote their own software and then generative AI came along and suddenly we could have app meals home-cooked by barefoot developers? (The use of such cottagecore terminology for a process that requires being an ongoing client of a hundred-gigabuck, planet-burning megacorporation rubs me in many wrong ways.)If AI finally gets rid of the thing that drove me nuts for years: "leverage" as a verb mean roughly "to use"—when no human intervention seems to work, then I shall be over-the-moon happy. I once worked at a place where this particular word was lever—er, used all the damn time and I'd never encountered something so NPC-ish. I felt like I was on The Twilight Zone. I could've told you way back then that you sounded like a bot doing that, now people might actually believe me and thank god.I will stick by the em dashes however. And I might just start using arrows too. Compose - > → right arrow. Not even difficult.
  • charlieflowers
    This list reads like, "AIs are not your typical braindead person on the street. They actually use a decent but not crazily advanced vocabulary."I mean, "tapestry" is a great word for something that is interconnected. Why not use it?
  • tiahura
    Many of these are standard fare in legal writing.Negative parallelism is a staple of briefs. "This case is not about free speech. It is about fraud." It does real work when you're contesting the other side's framing.Tricolons and anaphora are used as persuasion techniques for closing arguments and appellate briefs.Short punchy fragments help in persuasive briefs where judges are skimming. "The statute is unambiguous."As with the em dash - let's not throw the baby out with the bath water.
  • cyanydeez
    This kills the headline baiting tech blogger.
  • agnishom
    > (let's play cat and mouse!).No thanks, I hate this large scale social experiment
  • xpe
    > ... But prose? That's from human to human, it's sacred and meant for other people. Using AI for that is deceitful.I understand the sentiment. Meaning I think I understand some of the underlying frustration. But I don't care for the tone or the framing or the depth of analysis (for there isn't much there; I've seen the "if you didn't write it, why should I read it" cliché before *, and it ain't the only argument in town). Now for my detailed responses:1. In the same way the author wants people to respect other people, I want the author to respect the complexity of the universe. I'm not seeing that.2. If someone says "I wrote this without any LLM assistance" but do so anyway, THAT is clearly deceptive.3. If you read a page that was created with LLM assistance, it isn't reasonable for you to say the creator was being deceptive just because you assumed. It takes two to achieve deception: both the sender and the receiver.4. If you read a page on the internet, it is increasingly likely there was no human in the loop for the article at all. Good luck tracing the provenance of who made the call to make it happen. It might well be downstream of someone's job. (Yes, we can talk about diffusion of responsibility, etc., that's fair game -- but if you want to get into the realm of moral judgments, this isn't going to be a quick and tidy conversation)5. I think the above comment puts too much of a "oh the halcyon days!" spin on this. Throughout history, many humans, much of the time, are largely repackaging things we had heard before. Unfortunately (or just "in reality") more of us are catching on to just how memetically-driven people are. We are both individuals and cogs. It is an uncomfortable truth. That brainwashed uncle you have is almost certainly a less reliable source of information than Claude.6. The web has crappy incentives. It sucks. Yes, I want people to behave better. That would be nice, but I can't realistically expect people to behave better on the web unless there are incentives and consequences that align with what I want. The Web is a dumpster fire, not because of bad individuals, but because of system dynamics. Incentives. Feedback.7. If people communicate more clearly, with fewer errors, that's at least a narrow win. One has to at least factor this in.8. People accusing other people of being LLMs has a cost. Especially when people do it overconfidently or in a crude or mean manner. I've been on the receiving end. Why? Because I write in a way that sometimes triggers people because it resembles how LLMs write.* I want to read high quality things. I actually care less if you wrote it as bullet points, with the help of an LLM, on a napkin, on a posterboard ... my goal is to learn from something suited to some purpose. I'm happy reading a computer-generated chart. I don't need a human to do that by hand.The previous paragraph attempts to gesture at some of the conceptual holes in the common arguments behind "if you want a human to read it, a human should right it": they aren't systematically nor rigorously "wargamed" or "thought-experimented"; they are mostly just "knee-jerked".I am quite interested in many things, including: (1) connecting with real people; (2) connecting with real people that don't merely regurgitate an information source they just ingested; (3) having an intelligent process generating the things I read. As an example of the third, I want "intelligent" organizations that synthesize contributions from their constituent parts. I want "intelligent" algorithms to help me focus on what matters to me. &c.If a machine does that well, I'm not intrinsically bothered. If a human collaborates with an LLM to do that, fine. Whatever. We have bigger problems! Much bigger ones.Yes, I want to live in a world where humans are valued for what they write and their intrinsic qualities, even as machines encroach on what used to be our biggest differentiator: intelligence itself. But wanting this and morally shaming people for not doing it doesn't seem like a good way to actually make it happen. Getting to that world, to my eye, requires public sense-making, grappling with the reality of how the world works, forming coalitions, organizing society, and passing laws.Yes, I understand that HN has a policy that people write their own stuff, and I do. (See #8 above as well as my about page.)Thank you to the approximately zero or maybe one person who made it this far. I owe you a beer. You can easily find me. I'm serious. But then we have to find a way to have a discussion while enjoying a beer on a video call. Alas.I expect better from people -- and unfortunately a lot of people's output is lower quality than what I get from Claude. THIS is what pisses me off: that a machine-curated output is actually more useful to me than a vast majority of what people say, at least when I have particular questions to ask. This is one or many uncomfortable realities I would like people need to not flinch away from. As far as intelligent output is concerned, humans are losing a lot of ground. And fast. Don't shoot the messenger. If you don't recognize this, you might have a rather myopic view of intelligence that somehow assumes it must be biological or you just keep moving the goalposts. Or that somehow (but how?) humans "have it" but machines can't.