Stealing wisdom you haven't earned is exactly how I think about it. It's a tool that allows you to embody a style of speaking that is not yours, embody a perspective that is not yours, basically cheat at the very concept of personhood.
And that is not only morally wrong, and completely embarrassing, but self-defeating in the most existentially important possible way: it's a literal concession of your personhood. It's a decision to give up on being a person. It's just incredibly, incredibly stupid.
And I agree - better to miss something than to use AI at all.
do you mean 'you can tell it's chat GPT'? There are 100s of AI LLM models (and AI itself is a research field, not a single capability btw) each with a distinct tone. How is a novel different to code in terms of creativity? And yet Claude has swept the board in Software Engineering. I'm not sure whether this post was a trolling, but you don't have to use an LLM to write. I don't buy books written after 2022 (when ChatGPT got good) because i want authenticity, but plenty don't, which is why LLMs will survive so railing against them is a bit tilting at windmills and life is short.
This is a good point, but this sentence: "I don't buy books written after 2022 (when ChatGPT got good) because i want authenticity" feels so unbelievably bleak to me. Welp, guess I'll just give up on knowledge and literature created after 2022. I dunno, I think I'd rather be out in the fields with Sancho Panza fighting giants...
Yes it is bleak but just the way things have gone - there are very 'good' AI platforms for authors now, and i imagine the authors banging out a manuscript on an old Olivetti typewriter are increasingly rare. My addiction to old Russian authors keeps me away from the post-AI stuff anyway, but post 2022 the probability of AI being involved is high. Personal choice though, read whatever you enjoy.
The Russians certainly wrote enough in the 19th and 20th centuries to keep you busy for more than one lifetime!
Personally, though, I suspect most truly gifted/experienced writers are not using these tools much at all. I'm an academic historian. I've found AI tools very useful in translation and transcription, though I've stopped using them for transcription for a variety of reasons. But for writing, in every experiment I've tried, I've found them to be a *huge* time suck. And that's not even accounting for the ways they might degrade my work or expertise in the long run (which are myriad)! If you really know what you're doing as a writer, they really just do seem to make things take longer. That's my own experience, anyway.
100%, i have a todo reading list several meters high 😄 I have seen some really cool uses for AI, the archeologists who used it (and particle accelerators so double cool) to interpret the Herculaneum scrolls, amazing. Yes they are slow, models are complex and prompting them is a tough job if accuracy is required.
The "hope you are well" tag saddens me, because I use it often in emails. I'm also fond of em dashes. Apparently my writing style is more congruent with AI - and so the rough average of most writing? - than I care to think about. I'm fine being average, but I hope I'm not embarrassing myself out there.
I appreciate your comment (and the essay as a whole). I'm more annoyed than anxious about my writing style's similarity to AI writing; I'm too old to change my ways, but would rather not have to worry about people drawing the wrong conclusions. Hopefully you're correct and my humanity shines through.
It will. I’m currently TAing a graduate class and have had to unfortunately flag potential AI on… quite a few papers. This is an online class and (not for lack of trying) I’ve only met a few of the students. But, AI usually sticks out, even without knowing the students’ original writing style. The AI heavy writing (I don’t want to suggest I’d easily find those that just use it to clean up their grammar) is uniquely empty. The sentences are too pretty, and they mean absolutely nothing. They’re vague platitudes. You could take the same words and read them in a context foreign to these students and it would still-very vaguely-apply. Genuine engagement and wrestling shows.
Thanks John for saying this. I adopted “hope you are well” a long time ago, perhaps out of laziness, perhaps because I couldn’t find anything else that was kind but not too personal for professional emails. And I, too, am too old to stop with the emdashes!
I don't like this idea that em dashes are supposed to be "too old" writing or Al. It's appropriate fucking grammar. Telling people not to use appropriate grammar to avoid sounding like Al is bullshit and is contributing to the dumbing down of society just as much as Al is. People need to learn things before they actually start talking nonsense.
I don't like this idea that em dashes are supposed to be "too old" writing or AI. It's appropriate fucking grammar. Telling people not to use appropriate grammar to avoid sounding like AI is bullshit and is contributing to the dumbing down of society just as much as AI is. People need to learn things before they actually start talking nonsense.
Unfortunately articles like this are the reason why people begin to curb their natural language and edit their writing styles because they’ve been told that starting an email with “hope you are well” (which is just proper email etiquette btw) sounds like AI. And even if it was AI, you are not that special or important of a person to feel personally offended because someone addressed you using an em dash. Please check your ego and be so for real. The AI witch hunting is honestly more embarrassing than the AI usage itself.
You're right here. I hate AI, I'm absolutely against it. I'm also Spanish, and very aware that my written English is not the one of a native English speaker. I'm not gonna change my use of "hope you've been well" because I just discovered from this article that's apparently also ai, same as compelling (maybe this one is more suspicious, since I've never seen anyone using it till now), delve, testament (really?), and the em dashes which are literally the only proper way to write narrative... and Emily Dickinson's favourite markings. At this point I'm gonna grab Emerson's essays and Dickinson's letters and copy on a paper every word they wrote... It's better to copy Emerson than to accidentally write like AI. Maybe if we keep reading online articles we won't realise that an important fraction of them is written with AI (unfortunately even here people use it), and that implies we can all end up writing like AI without noticing... So, big paradox here. Isolation is the only way to stop the AI influence, but that also means surrendering and not being able to fight against it, together. They (technocrats, governments)want to divide us and the internet is their weapon now.
Y'all gotta read the thing. There is no "tell" to avoid. That's not the point, nor is it "hating AI." It's that passing off what is clearly spit out by an LLM is embarrassing. That's it.
It’s not necessarily congruent with AI - as a technical writer, long before AI was a thing we used em dash, semi-colons, colons, all the good stuff available to writers. If AI mimics from the vast corpus of what humans have written, why should that suddenly become a writer’s problem, when it’s an AI mimic issue.
Don’t curb it. Why let AI and fear of suspicion by others dictate your writing style. Otherwise it’s like a game where AI feeds off scanning human writing on the internet, does a mash-up that it feeds back to people, who then modify their own writing for fear of sounding like AI of being accused of it.
I was thinking the same thing. I’ve started emails with “I hope you are well“ long before Ai. Now I have to change how I’ve always written to avoid sounding like Ai? Eventually we’re going to have to get sloppy with our spelling and grammar so no one confuses us with Ai.
I like to think that because average is not a single number but rather a spectrum that I am at the higher end of the spectrum. Of course I am also self-confident enough to never be embarrassed by what anyone else thinks of me.
I often begin with "I hope you're well" and I use em dashes like I got 'em on clearance, but the whole "not this, but that" construction is driving me nuts. It's everywhere! And not in an amusing or edifying way.
Nah, because you actually wrote what you wrote. Even if it’s insincere, it’s more sincere than AI because it’s still something you chose to write. And I like the em dashes, too. AI wouldn’t use them if there weren’t enough of us that do.
I use that often too, usually as a question? Though, rather than a statement. My main problem with receiving AI emails is that they are inordinately long and take me about twice as long to work out what they are saying, all while sounding ‘clever’. Brevity in emails and comms in general is often a sign of competency in my opinion.
How could someone shame you for something that’s not shameful? If you’re feeling shame, then you already know it’s shameful. I couldn’t shame you for not using a salad fork.
Yes, it’s the deception. I’m seeing how many people see it in the same way as I do which is rather comforting but too many people rely on it as if they’ve lost the ability to actually use their own minds. Whenever I say how I feel to some they point out the positivity. I don’t accept the lack of avoidance.
AI text is akin to the pre-written messages in mother's day cards. Yes, it was written by a professional, it's poetic in a way your own writing won't be. But no one reads that stuff and thinks "wow, what a moving card, my family loves me so much." Moms famously prefer the scribbled "I lov u" in their preschooler's own hand.
What’s funny about that cute example, in my opinion, is that it is perfectly the spirit of what’s wrong with these stupid things and what a hundred 50-page papers on AI will fail to notice. I guess that’s how these things go.
My great fear is that I'm wrong to believe "no one reads that stuff and thinks..." There seem to be many people for whom I have no theory of mind. People credulous enough to buy the things in advertisements, for instance. The world is a terrifying place: maybe there are people out there being deeply moved by the same Hallmark card 10,000 other people received the same day.
A Hallmark card is written by a writer who is addressing generic mom who is unknown to anyone and not the specific love of the writer. But a person who has a feeling that they consider might be better expressed by technology that has shown it can express those in a more pleasing way, that makes them what? A moral failure as a human being?
Right. Technology doesn’t express a feeling as a subjective matter. I’m using the word “express” more broadly, as in “a certain color blue can effectively express sadness in a painting” or “that music expresses happiness”. I see this use as not too different from a person who chooses a famous love poem and sends it to someone they love. They didn’t write the poem but they feel it expresses what they wanted to say more poetically. The AI model doesn’t attribute every word to a particular poet, true, but the supply is such a huge ocean of material, it doesn’t really matter. I think the black or white thinking on this subject is kind of problematic. It seems particularly heated on Substack.
Thank you, and amen James. We're not supposed to be able to touch the whole elephant. We only get to put our hand on one part and then we have to talk to each other about what we feeling and sensing to comprehend the whole animal. I've never had a shred more than exactly zero desire to use AI for writing. Who wants to have a conversation with a know-it-all? "But it’s better to miss something." God yes. We are shaped and provide our unique contribution by a sort of divine omission. What we're missing is actually what constitutes what is called perspective.
Boom, Rick. Yes, writing is a communal act of feeling an elephant. You stop being fun to play with when you start screaming the shape of the Platonic elephant that you can't even feel.
I agree that purely AI-written content used to gain clicks, engagement, money, or false credibility can become deceptive.
But I’d be careful about declaring AI use itself as inherently deceitful. It is a tool. The ethics depend on how and why it is being used.
I think one of AI’s best purposes is helping people translate their inner world outward - especially people who struggle to find concrete language for what they feel, need, notice, or cannot yet articulate to the person across from them.
In that context, AI can be expansive. It doesn’t replace their minds; it helps them access and communicate what was already inside them. That can be a gift.
A tool reflects the hand that wields it. To discern between the tool and its use - rather than abolish it entirely - is a different kind of responsibility.
Gabrielle, I think you've identified the most interesting part of this discussion. AI can help people express thoughts that already exist but are difficult to articulate. I agree with that.
I'm a 100% AI writer.
Which is why I think this debate starts with the wrong question.
What is communication for?
Is it to prove that I'm more intelligent than you? More articulate? More original? More human?
Or is it to help another mind understand something it didn't understand before?
Human civilization has always relied on communication proxies. Scribes wrote letters for the illiterate. Lawyers translate intentions into legal language. Accountants translate economic activity into financial language. Speechwriters translate ideas into public language. The purpose of communication was never to prove who held the pen.
AI is simply the first communication proxy that almost everyone can access.
I completely agree that responsibility matters. Credit matters. Accountability matters.
But the value of communication does not come from the biological identity of the writer. It comes from whether the communication itself has value.
We spent centuries learning that the value of an argument should not depend on the author's race, gender, nationality, or social class. It would be ironic if the AI era taught us that the most important thing about an idea is whether every sentence was physically typed by a human.
Civilization advances because we build better ways to cooperate, not better ways to prove individual superiority.
I annoyed me how many people claimed Chris Best was sidestepping the issue when he called out Robert’s post as AI.
The entire dang article’s premise was that Substack was drifting towards content that played the social dopamine game instead of the deep engagement game.
To use slop to make that point is to lose before you start. AI content is, in my opinion, worse than just short-form Notes, etc, if the Notes are human-written.
Your champion article for “making work that matters” cannot itself be something that doesn’t matter at all.
Wow. After reading this, it feels like we need either a public service announcement or a cultural, Parasite-sort of horror film where you see the main character alienate everyone around them using AI - and no one tells them - and you just slowly observe their relationships crumpling in from the center and wondering why.
This was a great essay, and it really made me think about how I've confronted people and called them out on using AI in inappropriate settings, which I have, maybe because I'm just unhinged. The biggest one was writing to a blogger I've followed for 10+ years when she had a list that reeked of ChatGPT, basically saying I could tell, and it made me really sad because I've followed her for her voice for a decade. She never responded, but her AI usage did seem to drop off again after that.
I don't know about that, as a teacher I love using AI to reply to parents thay use AI to write complaint letters. And I make sure not to edit at all, I keep the em dashes and 'hope you are wells'. I want them to know that I used AI to respond to their AI.
To the parents that don't use AI, I'll write a genuine reply, although interestingly enough, it's usually the AI users that complain the most. Go figure.
It's wrong for the same reasons that using AI for any skill is "wrong": anything you use AI for is something you will inherently become rapidly worse at. Therefore, you should only use AI for those skills whose atrophy you accept. Finally, accepting atrophy of a skill reveals you don't value it.
I don't know that I can make a very good moral case for valuing the skill of cooking (but such a case exists I'm sure), but I can certainly make a practical one.
A) It was a joke. I actually don't care if people who don't have culinary creativity ask AI for recipe advice, B) If we turn this into a serious argument, as Joshua does here, it undermines your position in the piece. As there are people who feel about writing, like you do about cooking. If you'd like to come visit me on Grail Country again, I think a conversation on the ethics of the mundane and banal uses of AI is actually more interesting than the dramatic sweeping conversations about impending catastrophe.
Hmm. Is AI indeed, as you suggest, “morally neutral”? My first inclination would be to say: yes, absolutely. A tool is just a tool.
Someone infinitely more insightful and intelligent than myself had this to say, more than 40 years ago:
“What is happening in America is not the design of an articulated ideology… It comes as the unintended consequence of a dramatic change in our modes of public conversation. But it is an ideology nonetheless, for it imposes a way of life, a set of relations among people and ideas, about which there has been no consensus, no discussion and no opposition. Only compliance. Public consciousness has not yet assimilated the point that technology is ideology.”
-- Neil Postman, "Amusing ourselves to death" (1985)
I got my first personal computer in the mid 80s. Started using email in the early 90s. Signed up on my first “social network“ sometime around 2004. As I approach age 60–after a career spent working in digital mediums, I’ve seen enough to conclude without any doubt: technology is indeed ideology.
It’s a dusty ol’ chestnut, but Marshall McLuhan nailed it: when a medium dictates the rules of engagement, it controls the message. It controls all the possibilities. And it functionally mediates all “norms,” assumptions, and ostensible instincts as to what is “reasonable.”
Strikes me that is even worse than blind adherence to an “ideology” per se. it’s reality, controlled and curtailed at a pre-conscious level. Like so many other “advances“, the human animal simply becomes blind to it. (Especially in the case of those too young to remember a world “before.”)
Moral or immoral, catastrophic or survivable, I dunno. But morally neutral? I’m gonna have to go with “hell no.” 🤔
Excellent train of thought here. I tend to agree with you, Postman, and McLuhan. I would think of what I said near the end there more as a rhetorical flourish more than anything (and maybe I have too much of a tendency to be sort of sarcastic here and there). Sure, you can say it's morally neutral if you really want. But what does that mean? Lots of limitations and responsibility. Suddenly not so neutral.
Well said. lol like so many others in my cohort, I must confess that years on the Internet have made me (too often) irony deficient. 🤣
One final thought: quite literally the one and ONLY time I used “AI“ (that is, LLM generated writing) was about three years ago, at the request of my ad agency’s president. (He had paid for a GPT-type subscription in a feverish desire to be an “early adopter” and wanted me to experiment with it.) I generated three outputs: three versions of a potential blog post for a large utility company we had as a client.
Every one of them contained at least one (1) glaring hallucinated factual error. And each of them was of such a bland prose construction that I really couldn’t use them for anything. I duly reported this to the president. (No idea what came of it, I left their employment shortly thereafter.)
have never once had occasion or interest in using it since. Probably because I really, really enjoy writing. But it saddens me deeply to see commenters in this thread talking about how they are going to actively change their own writing style – for example, curtailing the use of em dashes – just because they fear being suspected of using LLMs.
If there is an immediate damage and wrong from AI to be seen, well, there it is. A human being altering their own writing style so that other people won’t think their writing was created by a machine? [shudder] genuinely @&$!ing horrifying.
Yeah, I hate that people have that reaction. And I personally dislike em-dashes now, but not to avoid sounding like AI, but because AI has made them feel icky, which just gives me a natural aversion. I don't think they make you sound like AI if you otherwise write it yourself, though.
I had an English tutor in middle school who taught me how to write essays and one syntactical “trick” she taught us was “it’s not X, it’s Y” because it’s so effective at creating tension and release - I often think about if this says more about how they train AI or how I was taught/how I think….
I also don't think it has much to do with the tells. They are just surface level whatever. If you go try the NYT quiz, you can just feel the AI, no tells. Metaphors are blurry, words are shifty, and things are hard to visualize.
Using “it’s not X, it’s Y” a few times in a work is NOT an AI tell. If something is AI it will have a bunch of tells in clusters all throughout the work. It has a really specific way of spitting out words and once you start to see it, you can’t un-see it.
A couple of years ago, I started seeing "It's not X, it's Y" on Substack. Not knowing that AI was infiltrating peoples' writing, I asked a couple of authors what they thought was going on. They didn't suspect AI at that time, but said it was somewhat annoying. I don't care for it, because I think there are better ways to express a thesis. If it's a structure you like, though, I don't want to criticize you.
The first time I was genuinely disappointed, disgusted and horrified by LLMs, was when I saw how they are influencing scientific research and academic writings. The way we evaluate and reward academics, is based on their publications, the originality of their works, and the significance of their contributions to science. The AI influence basically destroys our system for evaluating scientists.
I serve as a peer reviewer for a number of physics journals, and it's disheartening to see some authors let AI take full control of their scientific research. The concern isn't the quality of the science AI can produce. The real issue is the way we have to give credit to those scientists for what they publish: you see, almost anyone with minimal understanding of physics, can prompt the machine to spit out a model that predicts a physical phenomenon that was hypothesized by a bunch of other physicists who either lacked the expertise to create the computerized model to test their hypothesis, or lost the competition to the clock. We need to think of new ways to credit scientists for the science they produce.
As an agency principal for three-plus decades, I appreciate this so much. We are already experiencing erosion of some client work and worry that their authentic brand voices will become diluted. Your “I hope you are well” observation did make me pause. 😊 It, or something like it, is very often the second sentence in a response to a client, a sincere effort to bring natural warmth to a digital conversation. Thank you for your good words and insight. Did you mean to put the “Waldo” (there) in the image caption?
Oh, I'd love to say that the caption was intentional but it was just an accident, ha. I tend to not read things very carefully when I feel like I'm just trying to communicate a gist.
Anyone remember PFS First Choice? Or online Bulletin Boards? Am I the only one who began those early, awkward emails with: I hope this (note, email, etc.) finds you well.
That was in the early 90s - long before Altman et al had a notion about AI.
Don’t forget, AI is trained on previous human output (yeah from Reddit and other crappy places—but human nevertheless).
A truly Dantesque cycle of transgression and consequences: “You burn bridges, in fact, when you indulge in not thinking. If you commit to this strategy, you’ll end up frenetically chasing your own tail, trying to spike your slowly declining popularity with people who, by definition, go away as they pay more attention to you. You enter into a cycle of self-humiliation, finding more and more ways to justify that to yourself and others. People who get caught in this loop accidentally seal themselves into a self-contained yet declining ecosystem, which will be their creative end. It’s not unique to AI. You see it all the time. I knew viral listicle writers who were very popular on Medium. They were naturally taken with AI and produced a bunch of slop, and then humiliated out of existence.”
This is also perfect: “That might have happened a few times as a cash-out of generations of accrued societal and familial wisdom, but a society that makes a habit out of that behavior becomes a dead society. One that can’t ensoul children to be able to think anything at all.”
Stealing wisdom you haven't earned is exactly how I think about it. It's a tool that allows you to embody a style of speaking that is not yours, embody a perspective that is not yours, basically cheat at the very concept of personhood.
And that is not only morally wrong, and completely embarrassing, but self-defeating in the most existentially important possible way: it's a literal concession of your personhood. It's a decision to give up on being a person. It's just incredibly, incredibly stupid.
And I agree - better to miss something than to use AI at all.
I wish I would have thought to take it this far, Justin. Very well said.
do you mean 'you can tell it's chat GPT'? There are 100s of AI LLM models (and AI itself is a research field, not a single capability btw) each with a distinct tone. How is a novel different to code in terms of creativity? And yet Claude has swept the board in Software Engineering. I'm not sure whether this post was a trolling, but you don't have to use an LLM to write. I don't buy books written after 2022 (when ChatGPT got good) because i want authenticity, but plenty don't, which is why LLMs will survive so railing against them is a bit tilting at windmills and life is short.
This is a good point, but this sentence: "I don't buy books written after 2022 (when ChatGPT got good) because i want authenticity" feels so unbelievably bleak to me. Welp, guess I'll just give up on knowledge and literature created after 2022. I dunno, I think I'd rather be out in the fields with Sancho Panza fighting giants...
Yes it is bleak but just the way things have gone - there are very 'good' AI platforms for authors now, and i imagine the authors banging out a manuscript on an old Olivetti typewriter are increasingly rare. My addiction to old Russian authors keeps me away from the post-AI stuff anyway, but post 2022 the probability of AI being involved is high. Personal choice though, read whatever you enjoy.
The Russians certainly wrote enough in the 19th and 20th centuries to keep you busy for more than one lifetime!
Personally, though, I suspect most truly gifted/experienced writers are not using these tools much at all. I'm an academic historian. I've found AI tools very useful in translation and transcription, though I've stopped using them for transcription for a variety of reasons. But for writing, in every experiment I've tried, I've found them to be a *huge* time suck. And that's not even accounting for the ways they might degrade my work or expertise in the long run (which are myriad)! If you really know what you're doing as a writer, they really just do seem to make things take longer. That's my own experience, anyway.
100%, i have a todo reading list several meters high 😄 I have seen some really cool uses for AI, the archeologists who used it (and particle accelerators so double cool) to interpret the Herculaneum scrolls, amazing. Yes they are slow, models are complex and prompting them is a tough job if accuracy is required.
Well said.
Oh I like this!
Well, by that logic using a car is stealing speed you have not earned.
That's right. You aren't allow to drag race a 400m and act it as if you ran it.
But you do realize that it is the application that matters and not the technology itself.
No I don’t
Expired? It has barely begun. And for people who are against it, it's never going to end.
The "hope you are well" tag saddens me, because I use it often in emails. I'm also fond of em dashes. Apparently my writing style is more congruent with AI - and so the rough average of most writing? - than I care to think about. I'm fine being average, but I hope I'm not embarrassing myself out there.
I’m telling you, it has nothing to do with the “tells”. Em dashes are fine, it’s all fine. If you wrote it, people can tell.
I appreciate your comment (and the essay as a whole). I'm more annoyed than anxious about my writing style's similarity to AI writing; I'm too old to change my ways, but would rather not have to worry about people drawing the wrong conclusions. Hopefully you're correct and my humanity shines through.
Shine on you crazy human diamond ;-)
It will. I’m currently TAing a graduate class and have had to unfortunately flag potential AI on… quite a few papers. This is an online class and (not for lack of trying) I’ve only met a few of the students. But, AI usually sticks out, even without knowing the students’ original writing style. The AI heavy writing (I don’t want to suggest I’d easily find those that just use it to clean up their grammar) is uniquely empty. The sentences are too pretty, and they mean absolutely nothing. They’re vague platitudes. You could take the same words and read them in a context foreign to these students and it would still-very vaguely-apply. Genuine engagement and wrestling shows.
Thanks John for saying this. I adopted “hope you are well” a long time ago, perhaps out of laziness, perhaps because I couldn’t find anything else that was kind but not too personal for professional emails. And I, too, am too old to stop with the emdashes!
I don't like this idea that em dashes are supposed to be "too old" writing or Al. It's appropriate fucking grammar. Telling people not to use appropriate grammar to avoid sounding like Al is bullshit and is contributing to the dumbing down of society just as much as Al is. People need to learn things before they actually start talking nonsense.
Thank you for saying this.
I appreciate this clarification!
I don't like this idea that em dashes are supposed to be "too old" writing or AI. It's appropriate fucking grammar. Telling people not to use appropriate grammar to avoid sounding like AI is bullshit and is contributing to the dumbing down of society just as much as AI is. People need to learn things before they actually start talking nonsense.
Okay, o don't know what happened but this is not where my comment was supposed to go, and it won't delete. Please disregard
Unfortunately articles like this are the reason why people begin to curb their natural language and edit their writing styles because they’ve been told that starting an email with “hope you are well” (which is just proper email etiquette btw) sounds like AI. And even if it was AI, you are not that special or important of a person to feel personally offended because someone addressed you using an em dash. Please check your ego and be so for real. The AI witch hunting is honestly more embarrassing than the AI usage itself.
You're right here. I hate AI, I'm absolutely against it. I'm also Spanish, and very aware that my written English is not the one of a native English speaker. I'm not gonna change my use of "hope you've been well" because I just discovered from this article that's apparently also ai, same as compelling (maybe this one is more suspicious, since I've never seen anyone using it till now), delve, testament (really?), and the em dashes which are literally the only proper way to write narrative... and Emily Dickinson's favourite markings. At this point I'm gonna grab Emerson's essays and Dickinson's letters and copy on a paper every word they wrote... It's better to copy Emerson than to accidentally write like AI. Maybe if we keep reading online articles we won't realise that an important fraction of them is written with AI (unfortunately even here people use it), and that implies we can all end up writing like AI without noticing... So, big paradox here. Isolation is the only way to stop the AI influence, but that also means surrendering and not being able to fight against it, together. They (technocrats, governments)want to divide us and the internet is their weapon now.
Y'all gotta read the thing. There is no "tell" to avoid. That's not the point, nor is it "hating AI." It's that passing off what is clearly spit out by an LLM is embarrassing. That's it.
It’s not necessarily congruent with AI - as a technical writer, long before AI was a thing we used em dash, semi-colons, colons, all the good stuff available to writers. If AI mimics from the vast corpus of what humans have written, why should that suddenly become a writer’s problem, when it’s an AI mimic issue.
I, too, have been using em dashes for the past 20 years in my writing and have had to curb it for this exact reason.
Don’t curb it. Why let AI and fear of suspicion by others dictate your writing style. Otherwise it’s like a game where AI feeds off scanning human writing on the internet, does a mash-up that it feeds back to people, who then modify their own writing for fear of sounding like AI of being accused of it.
Why would you curb your natural writing voice?
Is it peer pressure? Fear of being called an “AI user” by some Carnac who supposes to know what writing is AI and what is not?
In the bastardized words of Kant, have the courage to use your own intelligence.
Agree. I'm using more em dashes than ever these days just to kick back at these silly snap judgements
Consider poor Emily Dickinson…
I was thinking the same thing. I’ve started emails with “I hope you are well“ long before Ai. Now I have to change how I’ve always written to avoid sounding like Ai? Eventually we’re going to have to get sloppy with our spelling and grammar so no one confuses us with Ai.
I like to think that because average is not a single number but rather a spectrum that I am at the higher end of the spectrum. Of course I am also self-confident enough to never be embarrassed by what anyone else thinks of me.
Yeah omg I use variations of "hope you're doing well" all the time, I didn't know that was also an AI thing. I also love em dashes. Sigh.
I often begin with "I hope you're well" and I use em dashes like I got 'em on clearance, but the whole "not this, but that" construction is driving me nuts. It's everywhere! And not in an amusing or edifying way.
I write with dashes. For ‘Em dash on Windows, you can use `Alt + 0151` on the numeric keypad. You are not doing this are you?
Yup me too . And the AI sniffer dog ‘experts’ are everywhere.
We'll take our em-dashes back from the clankers. They're ours!
I use hope you are well a lot too! I wonder what's a non AI sounding alternative to it
Nah, because you actually wrote what you wrote. Even if it’s insincere, it’s more sincere than AI because it’s still something you chose to write. And I like the em dashes, too. AI wouldn’t use them if there weren’t enough of us that do.
I use that often too, usually as a question? Though, rather than a statement. My main problem with receiving AI emails is that they are inordinately long and take me about twice as long to work out what they are saying, all while sounding ‘clever’. Brevity in emails and comms in general is often a sign of competency in my opinion.
It’s the deception that really irritates me. And then we have to choose: either be the jerk who points it out, or an enabler of the AI sneakery.
same
The deception is there because a bunch of you decide to shame people for using it. Stop shaming and encourage transparency.
How could someone shame you for something that’s not shameful? If you’re feeling shame, then you already know it’s shameful. I couldn’t shame you for not using a salad fork.
People are made to feel bad for using it, so hide it. It’s a tool - I urge transparency, not shame.
No Abi the deception is there because outsourcing creativity to a machine is intrinsically shameful
Out of interest do you have issue with it when it comes to productivity? E.g booking a meeting for you or transcribing your meetings?
Yes, it’s the deception. I’m seeing how many people see it in the same way as I do which is rather comforting but too many people rely on it as if they’ve lost the ability to actually use their own minds. Whenever I say how I feel to some they point out the positivity. I don’t accept the lack of avoidance.
AI text is akin to the pre-written messages in mother's day cards. Yes, it was written by a professional, it's poetic in a way your own writing won't be. But no one reads that stuff and thinks "wow, what a moving card, my family loves me so much." Moms famously prefer the scribbled "I lov u" in their preschooler's own hand.
What’s funny about that cute example, in my opinion, is that it is perfectly the spirit of what’s wrong with these stupid things and what a hundred 50-page papers on AI will fail to notice. I guess that’s how these things go.
My great fear is that I'm wrong to believe "no one reads that stuff and thinks..." There seem to be many people for whom I have no theory of mind. People credulous enough to buy the things in advertisements, for instance. The world is a terrifying place: maybe there are people out there being deeply moved by the same Hallmark card 10,000 other people received the same day.
A Hallmark card is written by a writer who is addressing generic mom who is unknown to anyone and not the specific love of the writer. But a person who has a feeling that they consider might be better expressed by technology that has shown it can express those in a more pleasing way, that makes them what? A moral failure as a human being?
The point is the message isn't better expressed by technology. It isn't expressed at all by the technology.
Right. Technology doesn’t express a feeling as a subjective matter. I’m using the word “express” more broadly, as in “a certain color blue can effectively express sadness in a painting” or “that music expresses happiness”. I see this use as not too different from a person who chooses a famous love poem and sends it to someone they love. They didn’t write the poem but they feel it expresses what they wanted to say more poetically. The AI model doesn’t attribute every word to a particular poet, true, but the supply is such a huge ocean of material, it doesn’t really matter. I think the black or white thinking on this subject is kind of problematic. It seems particularly heated on Substack.
Excellent
Thank you, and amen James. We're not supposed to be able to touch the whole elephant. We only get to put our hand on one part and then we have to talk to each other about what we feeling and sensing to comprehend the whole animal. I've never had a shred more than exactly zero desire to use AI for writing. Who wants to have a conversation with a know-it-all? "But it’s better to miss something." God yes. We are shaped and provide our unique contribution by a sort of divine omission. What we're missing is actually what constitutes what is called perspective.
Boom, Rick. Yes, writing is a communal act of feeling an elephant. You stop being fun to play with when you start screaming the shape of the Platonic elephant that you can't even feel.
Oh that “screaming the platonic shape” metaphor is effing perfect!
I also have no desire to use AI to write. It seems more like a burden than a help.
I agree that purely AI-written content used to gain clicks, engagement, money, or false credibility can become deceptive.
But I’d be careful about declaring AI use itself as inherently deceitful. It is a tool. The ethics depend on how and why it is being used.
I think one of AI’s best purposes is helping people translate their inner world outward - especially people who struggle to find concrete language for what they feel, need, notice, or cannot yet articulate to the person across from them.
In that context, AI can be expansive. It doesn’t replace their minds; it helps them access and communicate what was already inside them. That can be a gift.
A tool reflects the hand that wields it. To discern between the tool and its use - rather than abolish it entirely - is a different kind of responsibility.
Gabrielle, I think you've identified the most interesting part of this discussion. AI can help people express thoughts that already exist but are difficult to articulate. I agree with that.
I'm a 100% AI writer.
Which is why I think this debate starts with the wrong question.
What is communication for?
Is it to prove that I'm more intelligent than you? More articulate? More original? More human?
Or is it to help another mind understand something it didn't understand before?
Human civilization has always relied on communication proxies. Scribes wrote letters for the illiterate. Lawyers translate intentions into legal language. Accountants translate economic activity into financial language. Speechwriters translate ideas into public language. The purpose of communication was never to prove who held the pen.
AI is simply the first communication proxy that almost everyone can access.
I completely agree that responsibility matters. Credit matters. Accountability matters.
But the value of communication does not come from the biological identity of the writer. It comes from whether the communication itself has value.
We spent centuries learning that the value of an argument should not depend on the author's race, gender, nationality, or social class. It would be ironic if the AI era taught us that the most important thing about an idea is whether every sentence was physically typed by a human.
Civilization advances because we build better ways to cooperate, not better ways to prove individual superiority.
I annoyed me how many people claimed Chris Best was sidestepping the issue when he called out Robert’s post as AI.
The entire dang article’s premise was that Substack was drifting towards content that played the social dopamine game instead of the deep engagement game.
To use slop to make that point is to lose before you start. AI content is, in my opinion, worse than just short-form Notes, etc, if the Notes are human-written.
Your champion article for “making work that matters” cannot itself be something that doesn’t matter at all.
Wow. After reading this, it feels like we need either a public service announcement or a cultural, Parasite-sort of horror film where you see the main character alienate everyone around them using AI - and no one tells them - and you just slowly observe their relationships crumpling in from the center and wondering why.
This was a great essay, and it really made me think about how I've confronted people and called them out on using AI in inappropriate settings, which I have, maybe because I'm just unhinged. The biggest one was writing to a blogger I've followed for 10+ years when she had a list that reeked of ChatGPT, basically saying I could tell, and it made me really sad because I've followed her for her voice for a decade. She never responded, but her AI usage did seem to drop off again after that.
This film is almost certainly in the works. If not, it should be.
I've responded to some Substack posts that I wish they would not use AI, including to make the cartoon-like pictures. Usually they get defensive.
I don't know about that, as a teacher I love using AI to reply to parents thay use AI to write complaint letters. And I make sure not to edit at all, I keep the em dashes and 'hope you are wells'. I want them to know that I used AI to respond to their AI.
To the parents that don't use AI, I'll write a genuine reply, although interestingly enough, it's usually the AI users that complain the most. Go figure.
It's good for cooking? As a former chef them 's fightin words :)
Hey I sometimes plug everything I got in the fridge and get it to spit out something to cook. Am I wrong for this one?
It's wrong for the same reasons that using AI for any skill is "wrong": anything you use AI for is something you will inherently become rapidly worse at. Therefore, you should only use AI for those skills whose atrophy you accept. Finally, accepting atrophy of a skill reveals you don't value it.
I don't know that I can make a very good moral case for valuing the skill of cooking (but such a case exists I'm sure), but I can certainly make a practical one.
Very nice.
A) It was a joke. I actually don't care if people who don't have culinary creativity ask AI for recipe advice, B) If we turn this into a serious argument, as Joshua does here, it undermines your position in the piece. As there are people who feel about writing, like you do about cooking. If you'd like to come visit me on Grail Country again, I think a conversation on the ethics of the mundane and banal uses of AI is actually more interesting than the dramatic sweeping conversations about impending catastrophe.
I’m down
You are losing the formation you would otherwise have achieved by randomly throwing ingredients in the air frier and then tasting it 🤭
Hmm. Is AI indeed, as you suggest, “morally neutral”? My first inclination would be to say: yes, absolutely. A tool is just a tool.
Someone infinitely more insightful and intelligent than myself had this to say, more than 40 years ago:
“What is happening in America is not the design of an articulated ideology… It comes as the unintended consequence of a dramatic change in our modes of public conversation. But it is an ideology nonetheless, for it imposes a way of life, a set of relations among people and ideas, about which there has been no consensus, no discussion and no opposition. Only compliance. Public consciousness has not yet assimilated the point that technology is ideology.”
-- Neil Postman, "Amusing ourselves to death" (1985)
I got my first personal computer in the mid 80s. Started using email in the early 90s. Signed up on my first “social network“ sometime around 2004. As I approach age 60–after a career spent working in digital mediums, I’ve seen enough to conclude without any doubt: technology is indeed ideology.
It’s a dusty ol’ chestnut, but Marshall McLuhan nailed it: when a medium dictates the rules of engagement, it controls the message. It controls all the possibilities. And it functionally mediates all “norms,” assumptions, and ostensible instincts as to what is “reasonable.”
Strikes me that is even worse than blind adherence to an “ideology” per se. it’s reality, controlled and curtailed at a pre-conscious level. Like so many other “advances“, the human animal simply becomes blind to it. (Especially in the case of those too young to remember a world “before.”)
Moral or immoral, catastrophic or survivable, I dunno. But morally neutral? I’m gonna have to go with “hell no.” 🤔
Excellent train of thought here. I tend to agree with you, Postman, and McLuhan. I would think of what I said near the end there more as a rhetorical flourish more than anything (and maybe I have too much of a tendency to be sort of sarcastic here and there). Sure, you can say it's morally neutral if you really want. But what does that mean? Lots of limitations and responsibility. Suddenly not so neutral.
Well said. lol like so many others in my cohort, I must confess that years on the Internet have made me (too often) irony deficient. 🤣
One final thought: quite literally the one and ONLY time I used “AI“ (that is, LLM generated writing) was about three years ago, at the request of my ad agency’s president. (He had paid for a GPT-type subscription in a feverish desire to be an “early adopter” and wanted me to experiment with it.) I generated three outputs: three versions of a potential blog post for a large utility company we had as a client.
Every one of them contained at least one (1) glaring hallucinated factual error. And each of them was of such a bland prose construction that I really couldn’t use them for anything. I duly reported this to the president. (No idea what came of it, I left their employment shortly thereafter.)
have never once had occasion or interest in using it since. Probably because I really, really enjoy writing. But it saddens me deeply to see commenters in this thread talking about how they are going to actively change their own writing style – for example, curtailing the use of em dashes – just because they fear being suspected of using LLMs.
If there is an immediate damage and wrong from AI to be seen, well, there it is. A human being altering their own writing style so that other people won’t think their writing was created by a machine? [shudder] genuinely @&$!ing horrifying.
Yeah, I hate that people have that reaction. And I personally dislike em-dashes now, but not to avoid sounding like AI, but because AI has made them feel icky, which just gives me a natural aversion. I don't think they make you sound like AI if you otherwise write it yourself, though.
I had an English tutor in middle school who taught me how to write essays and one syntactical “trick” she taught us was “it’s not X, it’s Y” because it’s so effective at creating tension and release - I often think about if this says more about how they train AI or how I was taught/how I think….
I also don't think it has much to do with the tells. They are just surface level whatever. If you go try the NYT quiz, you can just feel the AI, no tells. Metaphors are blurry, words are shifty, and things are hard to visualize.
Using “it’s not X, it’s Y” a few times in a work is NOT an AI tell. If something is AI it will have a bunch of tells in clusters all throughout the work. It has a really specific way of spitting out words and once you start to see it, you can’t un-see it.
A couple of years ago, I started seeing "It's not X, it's Y" on Substack. Not knowing that AI was infiltrating peoples' writing, I asked a couple of authors what they thought was going on. They didn't suspect AI at that time, but said it was somewhat annoying. I don't care for it, because I think there are better ways to express a thesis. If it's a structure you like, though, I don't want to criticize you.
The first time I was genuinely disappointed, disgusted and horrified by LLMs, was when I saw how they are influencing scientific research and academic writings. The way we evaluate and reward academics, is based on their publications, the originality of their works, and the significance of their contributions to science. The AI influence basically destroys our system for evaluating scientists.
I serve as a peer reviewer for a number of physics journals, and it's disheartening to see some authors let AI take full control of their scientific research. The concern isn't the quality of the science AI can produce. The real issue is the way we have to give credit to those scientists for what they publish: you see, almost anyone with minimal understanding of physics, can prompt the machine to spit out a model that predicts a physical phenomenon that was hypothesized by a bunch of other physicists who either lacked the expertise to create the computerized model to test their hypothesis, or lost the competition to the clock. We need to think of new ways to credit scientists for the science they produce.
That's a fascinating wrinkle. I didn't know about that at all.
Very good
As an agency principal for three-plus decades, I appreciate this so much. We are already experiencing erosion of some client work and worry that their authentic brand voices will become diluted. Your “I hope you are well” observation did make me pause. 😊 It, or something like it, is very often the second sentence in a response to a client, a sincere effort to bring natural warmth to a digital conversation. Thank you for your good words and insight. Did you mean to put the “Waldo” (there) in the image caption?
Oh, I'd love to say that the caption was intentional but it was just an accident, ha. I tend to not read things very carefully when I feel like I'm just trying to communicate a gist.
You should just moderate my comment and your response, then, because it was pretty darn perfect, intended or not.
Either way, I'll keep it as it.
Anyone remember PFS First Choice? Or online Bulletin Boards? Am I the only one who began those early, awkward emails with: I hope this (note, email, etc.) finds you well.
That was in the early 90s - long before Altman et al had a notion about AI.
Don’t forget, AI is trained on previous human output (yeah from Reddit and other crappy places—but human nevertheless).
A truly Dantesque cycle of transgression and consequences: “You burn bridges, in fact, when you indulge in not thinking. If you commit to this strategy, you’ll end up frenetically chasing your own tail, trying to spike your slowly declining popularity with people who, by definition, go away as they pay more attention to you. You enter into a cycle of self-humiliation, finding more and more ways to justify that to yourself and others. People who get caught in this loop accidentally seal themselves into a self-contained yet declining ecosystem, which will be their creative end. It’s not unique to AI. You see it all the time. I knew viral listicle writers who were very popular on Medium. They were naturally taken with AI and produced a bunch of slop, and then humiliated out of existence.”
This is also perfect: “That might have happened a few times as a cash-out of generations of accrued societal and familial wisdom, but a society that makes a habit out of that behavior becomes a dead society. One that can’t ensoul children to be able to think anything at all.”
Lol you found my favorites.
Messaged one AI to another AI lol
Who were these writers on Medium that were humiliated out of existence?