I'm Kenyan. I Don't Write Like ChatGPT. ChatGPT Writes Like Me

> There were unspoken rules, commandments passed down from teacher to student, year after year. The first commandment? Thou shalt begin with a proverb or a powerful opening statement. “Haste makes waste,” we would write, before launching into a tale about rushing to the market and forgetting the money. The second? Thou shalt demonstrate a wide vocabulary. You didn’t just ‘walk’; you ‘strode purposefully’, ‘trudged wearily’, or ‘ambled nonchalantly’. You didn’t just ‘see’ a thing; you ‘beheld a magnificent spectacle’. Our exercise books were filled with lists of these “wow words,” their synonyms and antonyms drilled into us like multiplication tables.

Well, this is very interesting, because I'm a native English speaker that studied writing in university, and the deeper I got into the world of literature, the further I was pushed towards simpler language and shorter sentences. It's all Hemingway now, and if I spot an adverb or, lord forbid, a "proceeded to," I feel the pain in my bones.

The way ChatGPT writes drives me insane. As for the author, clearly they're very good, but I prefer a much simpler style. I feel like the big boy SAT words should pop out of the page unaccompanied, just one per page at most.

A lot of training data was curated in Kenya[0]. I would imagine if LLM data was curated in Japan our LLMs would sound a lot like the authors of their most popular English text books. Maybe other common Japanese idioms would leak in to the training data, like "ね" or "でしょう", ChatGPT would say "Don't you agree?" at the end of every message.

[0] https://www.theverge.com/features/23764584/ai-artificial-int...

I had a similar experience. We were talking about a colleague for using ChatGPT in our WhatsApp group chat to sound smart and coming up with interesting points. The talk sounds so mechanical and sounds exactly as ChatGPT.

His responses in Zoom Calls were the same mechanical and sounds like AI generated. I even checked one of his responses in WhatsApp if it's AI by asking the Meta AI whether it's AI written, and Meta AI also agreed that it's AI written and gave points to why it believes this message was AI written.

When I showed the response to the colleague he swore that he was not using ant AI to write his responses. I believe after he said to me it was not AI written. And now reading this I can imagine that it's not an isolated experience.

It’s not just the diction and phrasing, it’s the overall emptiness of the writing that is the biggest tipoff that chat gpt is used. It’s using the (ironically) “it’s not just x, it’s y” constructions when they feel tacked on, it’s the empty adjectives everywhere (e.g. intricate) that convey no additional meaning.

The same writing style is prevalent in lower-end consulting and precocious student essays where people have been falsely taught that more words are better and complicated words are better.

I do see people unfairly judging some good writing as being gpt written, presumably the same sort of people who in another situation might think gpt writing is good. But for the most part, if your writing looks like ChatGPT wrote it, you should take this as useful feedback and try and cut out the fluff.

Finally, the “let me explain” and “here’s the thing” in the article I don’t associate with gpt but with LinkedIn broetry posts, these are inappropriate in any circumstances.

To my eyes, this author doesn't write like ChatGPT at all. Too many people focus on the em-dashes as the giveaway for ChatGPT use, but they're a weak signal at best. The problem is that the real signs are more subtle, and the em-dash is very meme-able, so of course, armies of idiots hunt down any user of em-dashes.

Update: To illustrate this, here's a comparison of a paragraph from this article:

> It is a new frontier of the same old struggle: The struggle to be seen, to be understood, to be granted the same presumption of humanity that is afforded so easily to others. My writing is not a product of a machine. It is a product of my history. It is the echo of a colonial legacy, the result of a rigorous education, and a testament to the effort required to master the official language of my own country.

And ChatGPT's "improvement":

> This is a new frontier of an old struggle: the struggle to be seen, to be understood, to be granted the easy presumption of humanity that others receive without question. My writing is not the product of a machine. It is the product of history—my history. It carries the echo of a colonial legacy, bears the imprint of a rigorous education, and stands as evidence of the labor required to master the official language of my own country.

Yes, there's an additional em-dash, but what stands out to me more is the grandiosity. Though I have to admit, it's closer than I would have thought before trying it out; maybe the author does have a point.

I can only dream of writing english as well as OP. Kudos for mastering the language!

The formal part resonates, because most non-native english speaker learnt it at school, which teaches you literary english rather than day-to-day english. And this holds for most foreign languages learnt in this context: you write prose, essays, three-part prose with an introduction and a conclusion. I've got the same kind of education in france, though years of working in IT gave me a more "american" english style: straight to the point and short, with a simpler vocabulary for everyday use.

As for whether your writing is ChatGPT: it's definitely not. What those "AI bounty hunters" would miss in such an essay: there is no fluff. Yes, the sentences may use the "three points" classical method, but they don't stick out like a sore thumb - I would not have noticed should the author had not mentioned it. This does not feel like filling. Usually with AI articles, I find myself skipping more than half of each paragraph, due to the information density - just give me the prompt. This article got me reading every single word. Can we call this vibe reading?

It's an arms race between human writers and AI. Writers want to sound less like AI and AI wants to sound more like writers, so no indicator is reliable for long. Today typos indicate a real writer, so tomorrow LLMs will inject them where appropriate. Yesterday em dashes indicated LLM, so now LLMs use them less.

Beyond these surface level tells though, anyone who's read a lot of both AI-unassisted human writing as well as AI output should be able to pick up on the large amount of subtler cues that are present partly because they're harder to describe (so it's harder to RLHF LLMs in the human direction).

But even today when it's not too hard to sniff out AI writing, it's quite scary to me how bad many (most?) people's chatbot detection senses are, as indicated by this article. Thinking that human writing is LLM is a false positive which is bad but not catastrophic, but the opposite seems much worse. The long term social impact, being "post-truth", seems poised to be what people have been raving / warning about for years w.r.t other tech like the internet.

Today feels like the equivalent of WW1 for information warfare, society has been caught with its pants down by the speed of innovation.

It's the curse of writing well. ChatGPT is designed to write well, and so everyone who does that is accused of being AI.

I just saw someone today that multiple people accused of using ChatGPT, but their post was one solid block of text and had multiple grammar errors. But they used something similar to the way ChatGPT speaks, so they got accused of it and the accusers got massive upvotes.

Ironically, mistakes and idiosyncrasies are becoming a sign of authenticity and trustworthiness, while polish and quality signal the opposite.

Earlier today I stumbled upon a blog post that started with a sentence that was obviously written by someone with a slavic background (most writers from other language families create certain grammatical patterns when writing in another language, e.g. German is also quite typical). My first thought was "great, this is most likely not written by a LLM".

Ironically OpenAI used Kenyan workers[1] to train its AI and now we've come to the point where Kenyans are being excluded because they sound too much like the AI that they helped train.

[1] https://time.com/6247678/openai-chatgpt-kenya-workers/

Correlated but kinda off topic: I don't mind the style so much, I mind the verbosity. The amount of words spit out effortless by the writer which then need to be comprehended and filtered by every reader.

Seeing a project basically wrapping 100 lines of code with a novel length README ala 'emoticon how does it compare to.. emoticon'-bla bla really puts me off.

I read about 4 paragraphs of the blog post, it does not at all read like it was written by ChatGPT!

Some people are perhaps overly focussed on superficial things like em-dashes. The real tells for ChatGPT writing are more subtle -- a tendency towards hyperboly (it's not A, it's [florid restatment of essentially A] B!), a certain kind of rhythym, and frequently a kind of hard to describe "emptiness" of claims.

(LLMs can write in mang styles, but this is the sort of "kid filling out the essay word count" style you get in chatgpt etc by default.)

I feel this a bit, since I'm a voracious reader and a constant writer across a few languages (but mostly English), which over the decades has led to my converging on a certain (if imperfect) degree of polish. Plus my multiple concurrent and often fragmented simultaneous trains of thought while writing lead me to use parentheticals very often while drafting, which then means I often need go back and re-introduce structure.

And guess what, when you revise something to be more structured and you do it in one sitting, your writing style naturally gravitates towards the stuff LLMs tend to churn out, even if with less bullet points and em dashes (which, incidentally, iOS/macOS adds for me automatically even if I am a double-dash person).

I'm not sure I've read any of Marcus' previous writing, but there's no way that essay could have been written by an AI. It's personal and has a structure that follows human thought rather than a prompt.

For sure he describes an education in English that seems misguided and showy. And I get the context - if you don't show off in your English, you'll never aspire to the status of an Englishman. But doggedly sticking to anyone's "rules of good writing" never results in good writing. And I don't think that's what the author is doing, if only because he is writing about the limitations of what he was taught!

So idk maybe he does write like ChatGPT in other contexts? But not on this evidence.

I have seen people use "you're using AI" as a lazy dismissal of someone else's writing, for whatever reasons. That usually tells you more about the person saying it than the writing though.

AI / LLMs, including ChatGPT, can already be made to sound (almost) any way you want, just by telling it to. The usual tells that something was written or created by AI are changing monthly.

Just recently I was amazed with how good text produced by Gemini 3 Pro in Thinking mode is. It feels like a big improvement, again.

But we also have to honest and accept that nowadays using a certain kind of vocabulary or paragraph structure will make people think that that text was written by AI.

I always thought the whole argument was about explicitly using em dash and / or en dash. Aka — and –.

Because while people OBVIOUSLY use dashes in writing, humans usually fell back on using the (technically incorrect) hyphen aka the "minus symbol" - because thats whats available on the keyboards and basically no one will care.

Seems like, in the biggest game of telephone called the internet, this has devolved into "using any form of dash = AI".

Great.

The fact that everyone is now constantly forced to use (oftentimes faulty) personal heuristics to determine whether or not they read slop is the real problem here.

AI companies and some of their product users relentlessly exploit the communication systems we've painstakingly built up since 1993. We (both readers and writers) shouldn't be required to individually adapt to this exploitation. We should simply stop it.

And yes, I believe that the notion this exploitation is unstoppable and inevitable is just crude propaganda. This isn't all that different from the emergence of email spam. One way or the other this will eventually be resolved. What I don't know is whether this will be resolved in a way that actually benefits our society as a whole.

BS. ChatGPT writes in the sterile and boring manner of the average graduate of business, marketing or journalism: it is dull, safe, somewhat pompous but professional, the ideal style for corporate communication.

Basically, for two reasons:

1) A giant portion of all internet text was written by those same folks. 2) Those folks are exactly the people anyone would hire to RLHF the models to have a safe, commercially desirable output style.

I am pretty convinced the models could be more fluent, spontaneous and original, but then it could jeopardize the models' adoption in the corporate world, so, I think the labs intentionally fine-tuned this style to death.

This post doesn't read anything like ChatGPT. Correct grammar does not indicate ChatGPT. Em-dashes don't indicate ChatGPT. Assessing whether something was generated using an LLM requires multiple signals, you can't simply decry a piece of text as AI-generated because you noticed an uncommon character.

Unfortunately I think posts like this only seem to detract from valid criticisms. There is an actual ongoing epidemic of AI-generated content on the internet, and it is perfectly valid for people to be upset about this. I don't use the internet to be fed an endless stream of zero-effort slop that will make me feel good. I want real content produced by real people; yet posts like OP only serve to muddy the waters when it comes to these critiques. They latch onto opinions of random internet bottom-feeders (a dash now indicates ChatGPT? Seriously?), and try to minimise the broader skepticism against AI content.

I wonder whether people like the Author will regret their stance once sufficient amount of people are indoctrinated and their content becomes irrelevant. Why would they read anything you have to say if the magic writing machine can keep shitting out content tailored for them 24/7?

Funny how sci-fi always envisioned AI to speak in a rigid, hyper-rational terseness, whereas reality gave us AI which inherited the worst linguistic vices of "human" voices.

Also Kenyan, I once recently spent 10min explaining a technical topic via chat, and the response I got was "was this GPT?". I took a few minutes then just linked an article of how underpaid Kenyans trained ChatGPT for OpenAI [1]

1: https://time.com/6247678/openai-chatgpt-kenya-workers/

Bang on. The self proclaimed detectives have never had to take TOEFL where you'll get marks deducted for not using connectors like furthermore.

Well, his writing style is too good. The sentences flow too beautifully, he uses rich vocabulary and styling. It's unusual to see that style of writing online. I definitely don't poses that power.

I don't know the author of this article and so I don't know whether I should feel good or bad about this. LLMs produce better writing than most people can and so when someone writes this eloquently, then most people will assume that it's being produced by LLM. The ride in the closed horse carriage was so comfortable it felt like being in a car and so people assumed it was a car. Is that good? Is that bad?

Also note that LLMs are now much more than just "one ML model to predict the next character" - LLMs are now large systems with many iterations, many calls to other systems, databases, etc.

The internet been the same for a long time, it's just the wording that changed. As someone who apparently thinks differently, the amount of time people just end up saying "Well, you're just a troll, no one actually believes something like that, so whatever" since I started frequenting the internet in the early 2000s is the same as always. But some people try to be trendy and accuse you of using AI for writing the replies instead, but it's the same sentiment.

Besides, of course what people write will sound as LLMs, since LLMs are trained on what we've been writing on the internet... For us who've been lucky and written a lot and are more represented in the dataset, the writings of LLMs will be closer to how we already wrote, but then of course we get the blame for sounding like LLMs, because apparently people don't understand that LLMs were trained on texts written by humans...

I wouldn't usually use the 'non-native speaker argument', but thank you! Just yesterday I was accused of sounding like AI - https://news.ycombinator.com/item?id=46262777 - my default mode is that I oscillate between sounding too boring/technical, or when trying to do my best, sounding like AI

I’m having a similar problem. Spent way too much time on the internet starting in my preteens and it shaped the way I write - which not surprisingly - is a similar way to how an AI - trained on the online data - writes

We shouldn't need to have people bearing false witness. Anyone who uses AI tools to produce published works should offer a clear disclaimer to their audience. I share the same concerns as the author: "Will my written work be used to say that I plagiarize off ChatGPT?"

All the toil of word-smithing to receive such an ugly reward, convincing new readers that you are lazy. What a world we live in.

If you used a calculator to do a calculation, would they say the answer looks like created by calculator and not done by-hand?

I think the only solution to this is, people should simply not question AI usage. Pretence is everywhere. Face makeup, dress, the way you speak, your forced smile...

The author uses dash (-) not em dash (—), there is a big difference in that everyone has a dedicated dash/undersocore key on their keyboard, but nobody has a em dash key. You can use word processing software etc, but using em dash consistently throughout a text is very unnatural in casual written texts.

It's pretty rude to "accuse" someone of using AI. Would you yell "Dictionary!", "Grammarly!", "Reference manual!", "Newspaper quote!" at them? Maybe "Harvard!" or "Tutored!" ? You don't know who they are or what their life is like. Maybe they're blind and using it as an assisted device. Maybe their hand is injured and they use it to output information faster. Maybe they're old, infirm, a non-native English speaker, etc. Maybe they're just a regular person who feels insecure writing, and wants to use new technology to give them the confidence to write/comment more. Or, maybe they just talk like that.

Let's say you happen to be lucky, don't accuse someone unfairly, and they are using ChatGPT to write what they said. Who cares?! What is it you're doing by "calling them out" ? Winning internet points? Feeling superior? Fixing the world?

I don't mind the "normal" text so much, where you aren't sure if it was written by an AI or not. What's really getting annoying is the flood of bullet points and emoji that is flooding LinkedIn in particular. Super obnoxious!

Love this, everything about this - I still teach the foundation, 3 columns, roof, of the persuasive essay - except one bit:

Perplexity gauges how predictable a text is. If I start a sentence, "The cat sat on the...", your brain, and the AI, will predict the word "floor."

No. No no no. The next word is "mat"!

What was the "dead giveaway" referred to in the pasted tweet? Was it the dash, that people assume for some reason regular folks never use? Or was it something more interesting?

While author is correct in general, I would like to add a counter-point regarding em-dashes specifically. Yes, many people use them like this - and many website frameworks will automatically replace a keyboard not-really-a-minus symbol with em-dash. So that is not a sign of the LLM generated slop.

What LLMs also do though, is use em-dashes like this (imagine that "--" is an em-dash here): "So, when you read my work--when you see our work--what are you really seeing?"

You see? LLMs often use em-dashes without spaces before and after, as a period replacement. Now that is only what an Oxford professor would write probably, I've never seen a human write text like that. So those specific em-dashes is a sure sign of a generated slop.

Thank you for writing this. I too was a heavy user of the em-dash until ChatGPT came along. Though my solution has been to eschew the em-dash or at least replace with triple hyphens.

This must be infuriating:

> You spend a lifetime mastering a language, adhering to its formal rules with greater diligence than most native speakers, and for this, a machine built an ocean away calls you a fake.

This is :

> humanity is now defined by the presence of casual errors, American-centric colloquialisms, and a certain informal, conversational rhythm

And once you start noticing the 'threes', it's fun also.

> For my generation, and the ones that followed, the English Composition paper - and its Kiswahili equivalent, Insha - was not just a test; it was a rite of passage.

OK but come ON, that has to have been deliberate!

In addition to the things chatbots have made clichés, the author actually has some "tells" which identify him as human more strongly. Content is one thing. But he also has things (such as small explanations and asides in parentheses, like this) which I don't think I've EVER seen an instruction-tuned chatbot do. I know I do it myself, but I'm aware it's a stylistic wart.

the initial rlhf training evaluation was done by kenyans specifically

Systemic discrimination, happens all the time. I am blind. I regularily fail the "tell computers and humans apart" test. You imagine, that feels very much like the dehumanisation it is. Big tech couldn't care less. After all, they need to protect themselves against spammers. Much like the guy who was on the HN frontpage just a few days ago, arguing that he is now trashing accessibility because he doesn't want to be web scraped. If you raise these issues with devs, all you get it pushback, no understanding at all. Thats the way it is. If you are amongst a minority small enough and without a rainbow coloured flag, you end up being ignored, stepped over, and pushed aside. If you are lucky. If you are unlucky, and you raise your voice, you will be critizied for pointing out the obvious.

Honestly, people assuming I'm using ChatGPT to communicate with them and liberally using that suspicion as a filter sounds like a great meta-filter.

Looking forward to the deliberately abstruse and illogical essays of the future. Everyone will have to write like a second-rate French philosopher.

I can't wait until we reach the point of AI adoption where genuine content is suspicious.

Wanna submit a proof in a criminal case? Better be ready to debunk whether this was made with AI.

AI is going to fuck everything up for absolutely no reason other than profit and greed and I can't fucking wait

This resonates with me. LLM output in Spanish also has the tendency to "write like me", as in the linked article.

On that regard, I have an anecdote not from me, but from a student of mine.

One of the hats I wear is that of a seminary professor. I had a student who is now a young pastor, a very bright dude who is well read and is an articulate writer.

"It is a truth universally acknowledged" (with apologies to Jane Austen) that theological polemics can sometimes be ugly. Well, I don't have time for that, but my student had the impetus (and naiveté) of youth, and he stepped into several ones during these years. He made Facebook posts which were authentic essays, well argued, with balanced prose which got better as the years passed by, and treating opponents graciously while firmly standing his own ground. He did so while he was a seminary student, and also after graduation. He would argue a point very well.

Fast forward to 2025. The guy still has time for some Internet theological flamewars. In the latest one, he made (as usual) a well argued, long-form Facebook post, defending his viewpoint on some theological issue against people who have opposite beliefs on that particular question. One of those opponents, a particularly nasty fellow, retorted him with something like "you are cheating, you're just pasting some ChatGPT answer!", and pasted a screenshot of some AI detection tool that said that my student's writing was something like "70% AI Positive". Some other people pointed out that the opponent's writing also seemed like AI, and this opponent admitted that he used AI to "enrich" some of his writing.

And this is infuriating. If that particular opponent had bothered himself to check my student's profile, he would have seen that same kind of "AI writing" going on back to at least 2018, when ChatGPT and the likes were just a speck in Sam Altman's eye. That's just the way my student writes, and he does in this way because the guy actually reads books, he's a bonafide theology nerd. Any resemblance of his writing to a LLM output is coincidence.

In my particular case, this resonated with me because as I said, I also tend to write in a way that would resemble LLM output, with certain ways to structure paragraphs, liberal use of ordered and unordered lists, etc. Again, this is infuriating. First because people tend to assume one is unable to write at a certain level without cheating with AI; and second, because now everybody and their cousin can mimic something that took many of us years to master and believe they no longer need to do the hard work of learning to express themselves on an even remotely articulate way. Oh well, welcome to this brave new world...

Everyone thinks they are great at detecting AI slop, but they usually aren't. For art, there are certain giveaways, but for text?

I regularly find myself avoiding the use of the em-dash now even though it is exactly what I should be writing there, for fear of people thinking I used ChatGPT.

I wish it wasn't this way. Alas.

Sorry but using the emdash is just a shitty, over corporate way to write, and it instantly rubs some spot in the brain for some people; it doesn't matter if it was generated by an llm or not.

False accusations of AI writing are becoming absurd and infuriating.

The other day I saw and argued with this accusation by a HN commenter against a professional writer, based on the most tenuous shred of evidence: https://news.ycombinator.com/item?id=46255049

as a researcher, writing ended up being my job, and more specifically, writing in english. i never developed any sentimental link to the english language, to me it always felt bland, because i had to use it in bland environments, to write texts that had to be bland and manneristic.

chatgpt revolutionized my work because it makes creating those bland texts so much easier and fast. it made my job more interesting because i don't have to care about writing as much as before.

to those who complain about ai slop, i have nothing to say. english was slop before, even before ai, and not because of some conspiracy, but because the gatekeepers of journals and scientific production already wanted to be fed slop.

for sure society will create others, totally idiosyncratic ways to generate distinction and an us vs others. that's natural. but, for now, let's enjoy this interregnum...

Actually, there's a sweet solution to the writing and art crisis we are inflicting ourselves with in our AI craze. I call it "the island". Just find a nice tiny islet somewhere, make a few houses, and rent them by the week to writers/artists. No internet in the place. Rent out sanctioned devices; glorified typewriters without Internet access nor GPU nor CPU fast enough to run an LLM. Bring a notary to certify stuff was purely human-made. Have fun with like-minded individuals.