This would sound insane to me from two years ago but I have recently started insisting on writing all my own commit messages and pull request descriptions. I do usually have an agent review them for factual accuracy, but not rephrase them.

It slows things down a bit, but in the best possible way. It has helped immensely to improve the depth of my understanding of the agent-generated code. When agents are doing everything its way too easy to “skim” diffs and not really absorb them.

I always prided myself on my technical writing, and commit messages and PRs were a great place to hone that skill. I found that I missed it and my work is better now I’ve reclaimed that part of my old job back.

I fear in the future people will read even less because of how often people cheap out and use AI to write. I know the author of this article tells you never to use a single word it suggests, however I still, while reading, am checking for any amount of AI used. This makes reading less enjoyable, and even a bit stressful.

If you can't spend the time to write it, why should anyone read it?

Everything I post is from my own finger tips, only to get a reply from someone using a bot. I'm tired.

Even if you don't want style advice from an LLM (I generally don't, not for blogging), for a technical blogger "please check this post for factual accuracy" is a good way to not look like an idiot. Often it complains about hyperbole, which I tend to ignore since there's no point blogging unless you have some firm opinions, but on at least one occasion it pointed me towards some documentation that undermined my entire post - and led to a much better one a week later once I'd processed and applied that new information. Thanks, LLM.

I sort of agree, but I also think the advice here is circular. To look at the style suggestions produced by an LLM and decide that you agree with some of them in spirit, but want to put it in your own words, you already need how to know how to write well.

In other words, you need to read a style manual or two, and you need to read and have opinions about other people's work. You must have taste. Otherwise, you can't really distinguish between good and bad advice. You'll anchor to the suggestions made by the LLM and it will turn your writing into LLMese. I guess that works - it sure gets clicks on HN - but it won't be your voice anymore.

There's still value in getting a second pair of eyes on what you write. But here's another cool trick that gives you a similar benefit: just let it stew. Go for a walk, grab a lunch. Re-read what you wrote. You'll be surprised.

I've found the only good way to write with LLMs is to absolutely not use it for the first few drafts, the using it as a tool to list specific kinds of problems in text (word repetition, unnatural hard to read sentences etc) and approaching those points with caution and skepticism. Even copy pasting a single sentence, I've found, is a slippery slope.

Step one: open up the LLM

Step two: close the LLM

Step three: pick up a pen and write

Writing with LLMs is a great way to make yourself stupid: https://lambdaland.org/posts/2026-08-07-ai-writing-stupid/

I'm glad the OP acknowledges that you have to have to do the writing yourself—though they say this more as a means to avoid getting pegged as having used an LLM, whereas I see the problem as being cognitive surrender.

"LLM paragraph will register to much of your audience not as writing but as output"

The best advice for writing for other humans is: don't use LLMs.

If you're writing for machines, liking coding, then fine go for it. If you're writing for processes with formal highly structured content like manuals, specifications, form content, procedures, information, that sort of thing, then also ok to use LLMS. But if you're writing for a human mind to ingest and extract meaning from, then LLMs are poison.

> Present the original and new writing to the model and ask it which is better.

LLMs have absolutely terrible taste when it comes to writing. I don't find their feedback useful at all, beyond trivial spelling/grammar mistakes, which you don't really need an LLM for in the first place.

Proofreading is all you need.

Edit: I do sometimes ask an LLM for a fact-check, though.

It's extremely refreshing reading good human prose isn't it! It's like when I would read a newly-written Paul Graham essay 20 years ago.

He also uses rule of three and the phrase "load-bearing", just for fun, which elicited a wry smile from me.

Thanks OP for a pleasant start to my day!

What this quite reasonable article has to endorse may be useless. I've read some decent articles both obviously and possibly authored with LLM-assistance and while they were well-structured and carried at least the gravitas of objectivity I couldn't help but feel like their technical fastidiousness belabored their messages.

This article was so much better than what I though it would be

Does anyone know about "Style: Lessons in Clarity and Grace" ? Seems like an interesting book

> Your voice stays intact, but your work is faster, better, and less painful.

This sentence specifically seems like an LLM. It's hard to quantify that. They just take that sort of tone when they're needlessly summarizing what they already told you. "This thing contrastive thing remains important, but we got XYZ benefit."

The advice of "use your voice" only works if you have a good writing voice and comes off like Henry Cavill's dating advice or famous writers who said that you shouldn't use a thesaurus.

Paying close attention to posts written by folks who are not using LLM's the comments here regularly complain (sometimes extensively) about the writing quality. Doesn't matter what people are saying or suggesting about avoiding LLM writing, the real lesson I am seeing is either to naturally be gifted at writing or somehow figure out how to have LLM's help with your writing. "Rule Number One: You may not use a single word an LLM suggests to you." isn't helpful for probably the majority.

I’ve used llms in two ways while writing fiction. First I’ll have a story idea and use the llm to explore what has come before that may be related; then I’ll go read some stuff. This has worked great for seeing where my ideas fit and flushing them out. Second I’ll describe settings or plot devices and use the llm to brainstorm consequences or further ideas. This fails miserably. The llms suggestions are devoid of creativity. This is all moot though because I’m really not a writer, I’ve just always thought I might like to be one someday.

> Rule Number One: You may not use a single word an LLM suggests to you.

I think there's a use case where this rule can be ignored. Sometimes I ask LLM to help me pick a single word or short phrase in some language, that has specific meaning, and connotations. Or doesn't carry some connotations or associations. Especially when I have an idiom from some language in mind, but I'm writing in some other language and my brain struggles to find a good replacement.

LLMs are super handy as fancy vibe-based dictionary search engines. They're all about words that go with other words, so when I'm struggling for a good term, I see no harm in asking and picking an option that I feel closest to what I want to express. My choice, my voice, just a very fancy thesaurus machine.

I think it's good advice. I find it a bit funny the author said to not use a single word coming out of the LLM, but they probably did just that because I never heard a human being say "load-bearing" before but now it's all over the place, also used here. Or I don't get the irony.

I've seen this in action as well. The biggest value by far is seeing the overused phrases and tics in your writing style. The prompt I like is to tell it to highlight and flag passive voice and run ons since those are my major pain points. I can also tell when its getting out of "real" advice since it'll offer the same "trim 10-15%". Also important to tell it to highlight specifics otherwise it'll give vague editorial advice that isn't anchored. Another fun exercise is to submit other authors writing as well. Say a short story from a magazine. Some of the edits are legit, but do it enough times and you'll start to see the same generic writing advice gets applies to just about any input.

another iteresting thing models can do is to highlight and notice themes and motifs in a piece. I find that helpful when thinking about a piece of work. Might be obvious ones, sometimes it can be subtle themes. Good for stirring your own brain.

> Those rethinks are load-bearing parts of your voice.

I really hope this was intentional.

LLMs seem overpowered for the use the author suggests, finding and marking problems in prose. Actual human-written programs exist for that purpose, things you don't have to bludgeon into not praising you or semantically ablating your work. Start with langtool or the like and start adding others. Even art-bollocks mode for Emacs can be useful.

> frontier models are supernaturally good at selecting pleasing turns of phrase

It's not even the suggestions in isolation, but it tries to convince you that your writing is stellar and then slips in a word or two, and you accept it because otherwise you'd be skipping greatness. I guess in some way one has to have a bit of hubris about one's writing to tell the model to f off.

And OP, assuming this, from your prompt, is your language:

> "… First get the bones up…"

Congratulations, you are now doing LLM-speak :-)

For commit messages and other technical writing, I use a "narrative coherence" checklist to help ensure that a change is clear and understandable:

- Establish a single thesis an external reviewer could recover from the diff.

- Unify vocabulary across code, comments, tests, and commit message.

- Use the same vocabulary consistently to refer to the same concepts.

- Keep every hunk that serves the thesis; consider the removal or deferral of the rest.

- Introduce abstractions at the point of need.

- Align tests to narrate the same story as the implementation.

- Reconcile the commit message and the diff.

- Order changes expositorily, not chronologically.

- Explain what the change does and why. Do not explain the details of the development process.

- Prefer to edit subtractively.

- Recompile and run all relevant tests after edits.

- Iterate.

This is very good advice. I hate when llms turn into yes-men. I am surprise that the author doesn't use ai to construct the skeleton of the written piece. In my opinion, they excel at this, with all the books they've guillotined and digested, our biological brain cannot match their content retention. I would do the old trick of "asking for a friend". In this case, a friend of mine has mental block and he needs to write a paper on (insert topic here). I want to help him by designing the skeleton, and give the conditions/parameters of your piece.

> use them like a copyeditor rather than a ghostwriter

This is what you need to know. Seriously, after 25 years in news (writing), you realize that you need “second eyes”, not the first hands.

“Rule Number One: You may not use a single word an LLM suggests to you.”

So far so good…

“Those rethinks are load-bearing parts of your voice.”

Oops.

I use LLMs for nailing down the narrative in bullet point form. It proposes a lot of inappropriate points that I tell it to remove, but makes some useful proposals as well. It’s actually fairly efficient for narrative exploration, I then keep it away from prose for as long as possible since whatever it writes is pretty bad but once I start editing, it will usually clobber my edits if I make any narrative changes. Humans are still better at actual prose writing, especially when it comes to word rhythm.

undefined

I agree with the central thesis, but disagree with some details. I wrote a blog response: https://writing.brlewis.com/posts/2026/llm-copyeditor/

A workflow I've recently discovered for myself, that may be useful to anyone who speaks at least two languages: Ask the LLM to write a first draft in a language that isn't the one the piece should ultimately be in. Then use the LLM's first draft as a blueprint, to write a first draft yourself in the intended target language. That way, you won't be tempted to shut off your brain and just meatproxy anything from the LLM's first draft. Then hand your actual first draft back to the LLM for editorial suggestions. At that point, feel free to ignore suggestions, push back on bad ones while adding a rationale about why you don't like them, etc. Work in the best suggestions, then iterate. If you're in “writer block” mode, force yourself to keep going by asking the LLM for five suggestions on how to phrase something, form an opinion about what you like and don't like about them, draft it yourself by frankensteining from among the suggestions, without using any one suggested phrasing literally. Optionally, provide feedback to the LLM explaining your thinking about why you did or did not like certain suggested phrasings.

For objective/factual writing, asking for a fact check is a widely-used practice. And there's a useful counterpart for writing subjective things and opinionated hot takes, namely to ask the LLM: “Anticipate how this might be misunderstood in ways that might cause offence or trigger people.” Sometimes, you might want to say something that will cause offence, and there might be no way around it. In this case, making a conscious choice to let that be might be the right thing to do. But I find it surprising, how often an LLM will see an angle that I myself did not see, where something might cause offence in a way that I genuinely missed and genuinely want to fix. And then that gives me a chance to re-write before posting something.

Read Ong's "Writing is a Technology that Restructures Human Thought".

Emitting words is not really what "writing" is.

https://twinada.wordpress.com/wp-content/uploads/2012/09/708...

Great article. Thanks for sharing.

If English is your second language and you intent to write in your mother tongue, my advice is to use English to communicate with the LLM.

This makes Rule 1 straightforward ("You may not use a single word an LLM suggests to you") and helps avoid the horrible feeling that the machine is taking over your personal voice.

I guess more often than not we are thinking about what to write rather than how to write :-D

For blogging I just don't use an LLM at all. I just use a spelling/grammar checker. I get to sound like me even if that is imperfect.

For docs I have used an LLM for the convenience of having them kept up to date.

One thing that you should absolutely never do: have an LLM review and “improve” the text repeatedly in some closed loop.

This might work well for some tasks (coding), but for writing it will absolutely take reasonable text and turn it into a pile of incoherent garbage no human would ever write.

Maybe I’m the only one keeps trying this (more often than I’m willing to admit), but I suspect it’s a common cope engineers reach for when having to deal with the not-so-fun task of writing prose.

Till today, I was doing the complete opposite of what the article suggests. Definitely going to try this approach out, pretty helpful.

I have a specific ChatGPT chat that rewrites my posts really well every time; I usually just give it a rough draft and let it clean things up. But it was pretty interesting to realize how good these LLMs are at actually finding mistakes rather than just rewriting things. The point about not letting the model influence your voice was also very, very relevant.

Writing is thinking. You are outsourcing the process of thinking through your thoughts so you can communicate and understand them.

>So give the options to a model that doesn’t have the context of your editing process.

> I conjured a bit of software to manage this for me[...]

I don't understand how the annotation application shown helps with the stated goal of preventing the model from praising you. What's the actual software assisted workflow?

undefined

The frontier live voice models (especially GPT Voice/Live or whatever they call it this week) do not have the typical AI-tells. So I wonder about a few things:

Why is this?

Why couldn’t whatever went into post-training these voice models be applied to the text-gen models to get rid of AI smells?

Could we set up a pipeline that leverages the voice models to clean up text that reeks of AI ?

This is similar to how I've come to use them but it's focused too much on the mechanics or writing. I've found them more valuable at role-playing as a specific profile of reader and asking it how they would receive the arguments advanced by the writing and what counterarguments they could present. After some rounds of this process, you either close all the gaps in the argument or realize it's not a strong one.

I worry that reading too much LLM output will make me tend to write more like an LLM. Seems like it would be inevitable, to some extent. (Here's why it works)

"Streamline and improve" is redundant. Pasteurization is great. "jacking it with corn syrup" is nonsense.

I'll get my writing tips somewhere else.

At one point in my graduate college studies, I tutored qualified (DEI) undergrads 'for free' (for them) as part of a work-study program which paid me a bit for that work.

OTOH, an AI could not have qualified as such a student. But even if it did, before we began I would certainly bargain for considerably more than work-study tutoring paid.

Not letting your llm write is good advice and letting him spot problems also is if you want good writing.

For anything beyond "good", we're still on our own and I love that for us.

If you write with an LLM it will taint you voice. Slowly but surely.

Thought I'd get an article with tricks how to stop an LLM from sounding like the most obnoxious Linkedin influencer. Instead it says to use it as a copyeditor.

Ok - but that's not very useful.

Sounds like solid advice. Those typical Claude turns of phrase are really grating on me. And I now see them everywhere. Many YouTube channels that are presented by a real human, still sound fake because their script is blatantly written by AI. I hate it.

And yet, I'm stuck having to write an important message to someone that has real stakes riding on it, so I use Claude to refine it, and I see myself succumbing to Claude's pressure to use its words instead of my own. I hate it, but my words aren't very good.

And finally Claude proclaims the message perfect and I let my wife take a look at it, and she rejects the entire thing, restoring some semblance of sanity again.

> Readers can detect LLM words in the parts per trillion. However much work you put into scuffing up and humanizing it, an LLM paragraph will register to much of your audience not as writing but as output. So, first the bad news: you have to write for yourself.

---

and then

> Those rethinks are load-bearing parts of your voice. Readers won’t put their fingers on what’s wrong, but they’ll sense that you’ve become artificially-flavored.

---

Are they wrong in the first paragraph? Or are they simply not following the advice they dispense to others?

I mostly use LLMs for coding, so this is an interesting use case. I’m curious how long an article it can handle in practice before the context starts becoming an issue.

But why would you want to write with the help of an llm, you should be moving in the other direction. Don't let the creativity die!

> Readers can detect LLM words in the parts per trillion.

Oh, we do. We can also detect hackeneyed, LLM-made sentences where just one word had been replaced. "That's where it's going to f*** you".

Indeed. I have been experimenting with using frontier models for creative writing.

Using adversarial models to evolve ideas. I have a whole novel of material, and the feedback has indeed been useful. But the actual novel? Not even AI models like their own writing once you give the simple prompt of "account for model sycophancy".

The resulting structure and plot beats are good with AI as copy editor. Only the least discerning reader might not care or notice the actual non-human writing. And hey - maybe those are your readers. Probably not.

Can we all appreciate for a moment the sci-fi dystopia which snuck up on us all? I just used to phrase: " non-human writing". 50 years ago, that would send a chill up your spine.

I think style is overrated. Yeah if you write commercially, you are a journalist or a copywriter, sure, you need to write in certain style. There’s rules to this shit. I think it was Hemingway who said that. Then again, the good ones break the rules.

Today even if you don’t use LLM’s following certain formal style you end up sounding like ad copy or wannabe Atlantic -hack. The. You might read some kid on Tumblr writing beautiful prose with typos and errors and all.

First of all you should have something to say. I write a blog just to make tutorials for myself on stuff I need to regularly check out. I don’t really care if anyone reads it. I don’t use LLM to make the posts since, well, pride, but also because I learn better when I need to write the stuff down and check my facts several times. I don’t much care if I make a typo or grammar error here and there. If I want to write it as Jeeves from Wodehouse book, fuck it, I will.

If you have a story you have something to write about. It doesn’t have to be fiction, it can be a story on how multicast works. And it needs to be your story. Then the words come by themselves and writing is relatively easy. To have a story you have to live and think. LLM’s don’t help in that.

undefined

iA Writer's "Style check" feature can do 90% of this in real time.

undefined

How to write with an LLM?

Don’t

I was seriously reading this article right up until the word "load-bearing", and then despite my best efforts, my brain just sort of disconnected from the piece. I just skimmed to the end quickly before realizing what happened.

“Here’s how this is going to fuck you“ - this article was either written by a human, or the most aggressively tuned llm.

"Those rethinks are load-bearing parts of your voice."

Loved the (possibly not deliberately ironic) use of the LLM'S favourite analogy, load-bearing.

Nice wee article though. I particularly agree with "Don’t take all of the model’s copyediting advice." I've been using LLMs to review teaching material and often it just doesn't get the point of teaching. It's good for prompting reflection on your writing but it's still an idiot savant.

Don't. There, easier

Just like you do everything else with an LLM. You ask it for something, if that doesn’t work you just ask it in natural language to change it until it outputs exactly what you want.

It's more effective in my experience to have an LLM emulate your own writing examples than to apply lists of rules in the hope it sounds human.

Why are people not learning to write at school? In most developed countries don't we spend like a whole decade or more learning to do stuff like writing? Yet people still feel like they can't write?

Writing is something we really, really need to learn from other people, because it's with people we are trying to communicate. There should be no need for "writing with an LLM". None at all. If what you are writing is important, ask a friend to proof read it. If you can't write, learn.

This article should just be one word:

Don't.

It's darkly amusing that while the author advises us not to let LLMs pick our words, their post includes this chunk of AI slop:

"Those rethinks are load-bearing parts of your voice. Readers won’t put their fingers on what’s wrong, but they’ll sense that you’ve become artificially-flavored."

The metaphor at the start is AI-like weirdness too: "without pasteurizing and jacking it with corn syrup." Pasteurising is good, corn syrup is bad - what the heck am I supposed to make of it? It's nonsense.

It may even have been handwritten by the author, but it's slop nonetheless.

You may be telling yourself that you're disciplined and you're just using the LLMs for review. But while you're reading their output, the slop makes its way directly into your brain anyway - and then it drips out into your writing.

How I write with an LLM: for each page “suggest up to 10 word changes to increase clarity.”

> How To Write With An LLM

Simply don't.

> How to Write with an LLM

TLDR: Don’t.

[flagged]

[flagged]

[dead]

[dead]

[flagged]

I would have to be an utter moron to use LLM to write anything directed at other people.

just don't use it for writing

Simply do not.

Ironically, "load-bearing" appears in the main text of the article that says never to use a single word the LLM suggests. Womp womp.