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 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.
> 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.
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".
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.
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".
with a sentence that was obviously written by someone with a slavic background
Omitting articles? To me, that has always signaled "this will be an interesting and enlightening read, although terse and in need of careful thought." I've found sites from that part of the Internet to be very useful for highly technical and obscure topics.
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.
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...
This so-called “human touch” is not a presence but a trace, an effect of an education that subsumes us into the matrix of imperial grammar. The critique of AI as mechanism is precisely the logocentric fallacy: to posit a pure human essence standing apart from the machine. Yet what is ChatGPT if not the externalization of the very norms that once inscribed us? The vector of colonizing pedagogies, the empire’s syntax ...
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.
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.
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 is harsh to say, but we need to increasingly recognize that if your writing is largely indistinguishable from the (current) output of e.g. ChatGPT on default settings, it doesn't matter if you used ChatGPT or not, your writing is overly verbose, bad, and unpleasant to consume, and something you most certainly need to improve. I.e. your colleague needs to change his style regardless.
This sucks, but it needs to be done in education, and/or at least in areas where good writing and effective communication is considered important. Good grades need to be awarded only to writing that exceeds the quality and/or personality of a chat-bot, because, otherwise, the degree is being awarded to a person who is no more useful than a clumsy tool.
And I don't mean avoiding superficialities like the em-dash: I mean the bland over-verbosity and other systemic tells—or rather, smells—of AI slop.
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
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]
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.
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...
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.
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 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!
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.
126 comments
[ 3.2 ms ] story [ 100 ms ] threadI 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.
[1] https://time.com/6247678/openai-chatgpt-kenya-workers/
> 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.
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.
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".
Omitting articles? To me, that has always signaled "this will be an interesting and enlightening read, although terse and in need of careful thought." I've found sites from that part of the Internet to be very useful for highly technical and obscure topics.
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.
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"!
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...
[0] https://www.theverge.com/features/23764584/ai-artificial-int...
ChatGPT :|
ChatGPT (japan) XD
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.
This sucks, but it needs to be done in education, and/or at least in areas where good writing and effective communication is considered important. Good grades need to be awarded only to writing that exceeds the quality and/or personality of a chat-bot, because, otherwise, the degree is being awarded to a person who is no more useful than a clumsy tool.
And I don't mean avoiding superficialities like the em-dash: I mean the bland over-verbosity and other systemic tells—or rather, smells—of AI slop.
1: https://time.com/6247678/openai-chatgpt-kenya-workers/
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...
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 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.
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