The revolt of the reader

I recently read this William Zinsser quote that inspired a nickname for this: Clotted Claude [1].

> Nobody has made the point better than George Orwell in his translation into modern bureaucratic fuzz of this famous verse from Ecclesiastes:

> > I returned and saw under the sun, that the race is not to the swift, nor the battle to the strong, neither yet bread to the wise, nor yet riches to men of understanding, nor yet favor to men of skill; but time and chance happeneth to them all.

> Orwell's version goes:

> > Objective consideration of contemporary phenomena compels the conclusion that success or failure in competitive activities exhibits no tendency to be commensurate with innate capacity, but that a considerable element of the unpredictable must invariably be taken into account.

> First notice how the two passages look. The first one at the top invites us to read it. The words are short and have air around them; they convey the rhythms of human speech. The second one is clotted with long words. It tells us instantly that a ponderous mind is at work. We don't want to go anywhere with a mind that expresses itself in such suffocating language. We don't even start to read.

[1] https://blog.kierangill.xyz/clotted-claude

Always a pleasure reading Bryan's writing; it's like Bryan is sitting there with you and saying the words (hard to convey the feeling).

IMHO a big problem with Pangram in particular is that they market it as a reliable tool that can be used to catch students cheating. This can obviously have disastrous effects on young lives, because it is not as reliable as they suggest.

Per their own benchmarks, they do not achieve 100% accuracy even on text that is published on the Internet, and which is likely encoded into the models themselves.

There is validity to their goals, but that is overshadowed by the irresponsible way in which it is marketed.

(All of this, swirling in a context where students are being told that they absolutely must become proficient at using LLMs to do exactly this kind of work by the highest levels of state and federal governments, faculty leadership, as well as the leaders of the workforce into which they hope to graduate. The message to youth is extremely muddled at best.)

Also, there are relatively easy ways to prevent being recognized and I assume as now the use of LLMs changes our way of writing and speaking, it will get harder and harder for these detection tools. We will see more false positives (as for the future there will be no text 100 % authored by humans to train on).

https://www.lesswrong.com/posts/hrpQxfYvF6CBGWfJX/pangram-ai...

As English is my second language, I find the help of an LLM in editing text very useful. Yet, I agree with others that I don't want to read completely LLM generated texts.

More and more, I try to ask it for low prose responses because its writing just seems like such a low signal to noise ratio

I'm curious about why LLM writing fails. Particularly whether LLM writing is fundamentally flawed, or if it's just distinctive and since it often reflects low effort, that distinctive voice is associated with low quality.

I find its reliance on extremely consistent rhetorical patterns concerning. The fact that it always finds a way to talk about how "It's not the X, it's the Y Z" no matter what topic you feed it, makes me concerned that the tail is wagging the dog

If a person reads AI generated text and does not notice, they by definition will not know about it.

There have been numerous cases of people accessing human created content as being AI.

There are instances where it seems relatively uncontroversial that it is AI generated, but without knowing both the amount of AI content people are exposed toand the amount that they register I don't think you can draw a conclusion of the overall state.

No. I have good anecdata: readers cannot reliably distinguish my own prose from LLM-written one apart from cases where LLMs use odd metaphors or one of their specific patterns. I've been specifically experimenting with that.

What about answers that an LLM gave to a question that we ourselves asked? Should we “labor” to understand that answer?

I think the argument, as presented in this and other similar pieces of critique, is too simplistic.

I do understand the criticism, but I think it should be framed in a different manner. The problem, when we read a long form piece by an author, is that we imagine that there’s another “mind” at the other side. We imagine that we are following the reasoning within the mind of a fellow human being, the writer. There’s an implied sort of “intimacy” to it. And the breach is when we are fooled into thinking that we are engaged in human communication, only to discover that there is a machine on the other side.

When we ask questions to an AI, this problem does not exist, because we are fully aware that the entity on the other side is not a human being.

Yet there is no doubt that the reply from an AI can contain information that is very much worthy of our time, and of our “labor” and effort to understand it.

So I think this ultimately will be about disclosure. As long as we are being made aware of the percentage of AI use in a text, explicitly or implicitly, I think we will actually grow to accept it.

It's common enough that it's training me to recognize and recoil from AI tics through sheer classical conditioning.

Pretty much sums up the issue re: workplace lazy AI dumping on folks as well.

https://meatbag.atomic14.com/

https://www.atomic14.com/2026/08/18/detecting-claude-with-le...

It’s very hard to make reliable though. Different models have different characteristics and you can prompt your way out of being detected.

Even then, I would say that using an LLM is robbing you of the process of writing, a process that is crucial to developing and understanding your own ideas.

Think about the last time you wrote something for consumption and the sentence to sentence thought processes you’re going through. I bet a lot of that was “is that right?” Or “does that make sense?” Or “am I communicating this at the level of my reader?”.

All of that is fundamental to your readers understanding, but more importantly, its fundamental to YOUR understanding.

The big problem with the linked poll by Cynthia Dunlop is that it's all self-reported. The fact that the sample is not representative of the general population rather but might be closer of 'early adopter'/'power users' is interesting. But the idea that "I prefer authenticity to polished crowd-pleasing content" is something that people love to believe about themselves, but is hardly ever supported by facts.

As someone who spends a lot of time painting (maybe more time thinking about it than doing it, but still), it is quite evident that this is a fable. AI content is now everywhere in the streets -- just yesterday I went to a fair and food stands were divided into 2, the ones that hadn't updated their menu in 10 years or more, and the one that had generated it with ChatGPT. Museums are shameless at using AI images for their signs, and way too often even for their content. People do not prefer crappy human art, and those who self-report they do fail very hard at 'image Turing tests'.

I think the general point is true, but it does not give any timescale of when the dark age might end, when the tools will adapt etc.

I (am kinda forced to) use LLM to generate maybe 40% of the code at work, that is after my review and modifications. But I pretty much wrote all of the comments by myself. I can get into the flow by writing comments.

I liked your piece, and agree with almost all of it, but I'm surprised by your faith in the accuracy of Pangram at detecting AI writing. Is your faith based on testing it with lots of writing of known origins, or are you just saying that it reaches the same conclusion that you do as a talented human?

In particular, I wondered if you have tried running all of your own writings through it to verify that it thinks you are human. I was struck by Freddie deBoer's recent piece where he did this and said it often failed: https://freddiedeboer.substack.com/p/i-wouldnt-say-pangram-i...

What percentage of false positive rejections would you find acceptable? Would you accept this even if it forced you to change the way you write?

this goes 10x for all the slide decks and google docs and wikislop everyone's trying to pass off as an accomplishment lately

* If I need to learn something before I write about it, I rely on LLMs heavily to answer questions that I have about other source materials, e.g. to clear up ambiguities.

* I've recently started prompting it to find grammatical and spelling errors.

* And I've prompted it to find technical errors, places where I'm just wrong.

For all the prompting, I additionally tell it to not rewrite anything or offer any prose suggestions. It can keep all that to itself, thank you.

And I verify what it gives back for correctness.

(I'd encourage non-native speakers to use LLMs in much the same way. Don't sacrifice your human voice by letting the AI rewrite your words. Personally, I'd very much rather hear it from you, blemishes and all, than hear it from an AI.)

But if I could step back for a minute:

Why write anything?

If your writing goal is to flood the zone and make as much money as humanly possible from ads, then hell yeah, paperclip the everliving shit out of that.

But if your writing goal is to learn material or share material, then put that LLM on the back burner and don't use it to directly generate your text. It's bad for you, and the results are subpar.

When I'm learning something, I can go through reams of tokens and then, once I understand it, I digest that to single a paragraph about the topic. The paragraph is as concise and as helpful as I can make it. Now, I could just share the prompts that I went through with those pages of back-and-forth with the LLM... but wouldn't you rather just read the concise paragraph that gets the point across?

It's not hard to be better than an AI at writing for humans, so the minimum low bar to aim for is "better than an AI". And we can all get there with a small amount of practice. The real goal is to greatly exceed the LLMs' capabilities for sharing information.

Finally, I think everyone should write a lot. Blogs, morning pages, fiction, technical books, letters, whatever. Especially when it comes to technical content, nothing makes you do your research like putting your ass out in the ether to get flamed by 5 billion people. And teachers the world over know the best way to learn something is to teach it. Pick a topic, research, and write it up more clearly and concisely than anyone else ever has. You'll learn so much, and your readers will, as well. Writing fires up your brain. Don't give that up to an LLM.

I guess there is a kind of participatory element to the discourse where, if you want an audience, there is an editing process. Whereas in other cases, we wrote these as progress notes on an unknown journey, breadcrumbs or upturned stones to mark a path to the horizon.

Maybe it's the difference between writing as a mode of discovery, retreading the mental arc of a solution, and writing something honed to leave a mark.

To me, the glaring question is: What are we doing? The act of writing exists to 1) externalize and organize one's own thoughts for the purpose of considering and revising those thoughts; and 2) share one's own thoughts with other minds.

When we hand writing to a machine, we hand thinking to a machine, denying both our humanity and our role in the conversation.

I wish that were true, but I fear it may not be.

https://arstechnica.com/ai/2026/07/canadian-legislator-reads...

Well, I would rather not review AI code as well.

If an LLM helps you do that, then use it.

I dont care about llm use on any ideological grounds, or believe its a voiding of the social contract like the author of this piece does. I think its just a tool. However a lot of people use the tool really badly.

I would subscribe to a YouTube channel that did this.

> Why do people have this reaction? Beyond having to endure aggravating stylistic tics, when reading a piece that has had substantial LLM assistance, we — the readers — don’t know what is real and what isn’t.

This is well said. But, here too, I would pause and reflect on what it means to (think you) know what is real and what isn't in a pre-LLM setting. For example, authority bias predates LLMs, and can have disastrous consequences.

Another great day where Google only gives me 2 results on the first page.

Only saying "LLM writing" is honestly lazy writing. Specifically what?

I get the glaring cases, I get the idea that if the prose is generated then maybe also the idea, I get the feeling when reading a complete LLM authored piece.

But that doesn't help the piece, because - beside those glaring cases - most writing today is a mix between authors ideas and LLM prose.

https://github.com/anthropics/claude-code/issues/80015

It's tsunami of AI vomit drowning out a few human posters. I can 100% empathize with OSS maintainers banning AI submissions.

I dunno, man, according to Hardcover, I've read 76 fiction books this year, and I can't tell. All the "AI tells" fail the vibe check. I'm a writer and I get flagged by many of them.

And according to PhD linguists with expertise in the field, most AI tells are just the equivalent of old wives' tales. https://www.youtube.com/watch?v=ORgKY9AlybA

I vaguely recall that researchers were able to train people to tell, but only for a minority language that AIs likely aren't particularly good at mimicking, and after training.

This whole thing reminds me of how "you can recognize a vegan because they'll tell you." There, you have a ton of false negatives (i.e., since you aren't polling people to find out if they're vegan, you're only flagging the obvious vegans and missing all the regular people who happen to be vegan).

Except here, it's a bunch of false positives and negatives I bet. You don't really have a way of knowing, so you're accusing some people (without complete accuracy) and missing some people (without complete accuracy). But you have no way of knowing, so you're just like "hell yeah, my vibes tell me I'm right."

Research and experts disagree.

Of course I wrote an essay about this[1]. Here it is: https://rikverse2020.rikweb.org.uk/blog/how-not-to-train-you...

[1] - Trigger warnings: discusses poetry; touches on using LLMs in the research phase of essay writing and poetry review.

So this way there will be a lot of false positives like with school assignments.

I think a better solution would be to have a network of people that you trust manually read and label texts instead, so this way no machines are used and you don't need to read a text that 100 of your trusted friend/trusted readers flagged as artifitial.

By the way, this man does not care about AI slop. He cares about AI usage. In the same way there is very good code created with the help of LLMs, albeit a minority like Linus Towards says, there will be very good writing created with the help of LLMs.

Where I kind of disagree is that I don't think most readers will revolt. I think the mountain of LLM slop has actually changed people's behavior in more ways than one. Some are already relying on LLMs to summarize articles: then it doesn't matter to them who wrote it, they're just consuming machine-condensed content with no way to tell if a human or an LLM wrote the original piece. Or if their summarizer hallucinated.

This is an excellent point. It should apply to LLM generated code as well. If you send your colleague a vibeslopped PR, please also include all the prompts you used to generate it. Check that mess into version control right beside the code changes. Later, when we have to untangle all the spaghetti, then at least we'll have some archeological record of intent.

Pangram complains about "paraphrased" content though. Sigh...

> A compiler for the interface your agent already has: the shell. MCP makes an agent carry every tool's schema on every turn; declick compiles an API, an MCP server, or a database once into named verbs the model loads one at a time, and every verb returns one envelope with five exit codes. A team pushes a compiled adapter to a shared folder or git checkout once and every other machine pulls it. Ten engines, zero runtime dependencies, Node 24.

What the hell would it even actually mean to have a "compiler for the shell"? How do you "compile" a database into a "named verb"? What would this mean? These notions don't meaningfully cohere together. Why exactly is an "envelope with five exit codes"? Are we getting all of the exit codes at once? That wouldn't make any sense. LLM writing is full of this kind of arbitrary collision of concepts that sound like they plausibly go together thanks to frequent co-occurrence but that have no coherent logical meaning when you apply even a modicum of critical scrutiny. This is why I find it astounding that anyone believes these things are "intelligent" and not still just fundamentally a matrix machine completing words.

If you want to read something good, read a good book.

Is obvious AI-assisted writing better or worse than an obvious PR quid pro quo and/or cross-promotion?

Readers think they don't like LLM-authored text because they only recognize bad LLM-authored text as LLM-authored.

Blind trials have actually shown that readers generally prefer LLM authored books to human-authored ones on the same subject.

I don’t think this is remotely true. Sure, they’re legally obliged to let you unsubscribe, and sure, it’s not dick pills, but every US company will immediately send you a newsletter when you purchase something, review requests and, if they/you use Shop for checkout, expect an abandoned cart reminder.

PR pieces and software companies don’t write tutorials to be helpful, they are advertising to you. If the LLM can do it for cheap, they really don’t care.