I’ve been pondering this off and on over the past couple of weeks, since the AI checker has become the greatest trauma the Notes on Substack have ever experienced. 😂 On the one hand, I giggle over AI flagging classic authors as AI-written, since AI was trained on public domain literature to begin with. Classic authors sound like AI because AI was designed to mimic classic authors: cute.
On another hand, being the eternal optimist over here, I have to wonder if the whole mess only gives us authors a unique opportunity to develop our unique voices in such a way that we’re not sounding like AI because we’re not copying the styles of the past but are, in fact, challenging ourselves to *gasp* be original. What if we decide to forget about breaking the mold and go out and cast our own damn mold? But that’s just me, so. 🤣
It’s the new modernism! 🙃 And yeah, that’s the argument that history bares out—photography made realism cheap and easy so painters really began to explore expressionist modes. The question is what happens when competent prose becomes cheap and easily reproduced.
Yesterday, I was listening to a podcast episode done by two gay erotic authors on Substack on how to identify A.I. written ‘slop’. They pointed out that A.I. detectors are currently flawed, partly because they were trained using classic literature.
Also, I nearly failed a term paper in high school on Steinbeck. I had stupidly listened to my conservative parents who essentially told me Steinbeck was a socialist.
However, I am proud to say I met Vonnegut at a book signing when I was in college in the 90s! 🤓
I would love to read a Steinbeck Substack if he was reincarnated today.
Reading your argument, have you tried the same with Stephen King? While he tweaks it, I feel he is very much a formula writer. He has a methodology that is pretty consistent, he even writes about it himself. I wonder how he would score.
That would be a great one—Stephen King spent fifty years helping define what that statistical average is. If millions of writers learned clarity, pacing, and conversational prose by reading King, then today’s language models have inevitably absorbed those same patterns through the writing culture they were trained on. The detector then mistakes influence for imitation.
Apparently it's agreed that AI writes like neurodivergent authors, and we are more than likely to be the first flagged as 💯 AI. Bearing in mind AI was trained on human author's work, does that mean most of those authors were neurodivergent, or does it mean neurodivergent authors are equal to, or superior to, AI authorship. Why else would we need a machine to expose work that's not AI or similar to AI. And then dumb it down even more. That thought actually makes me feel a little superior for a change. 😅
Thoughts from the first glass of the second bottle of wine:
With the advancement of technology, especially AI, I had hoped to see the birth of super intelligence, instead I feel we are rewarding the dumbing down of intelligence. You know, keep the masses stupid, turn them into sheep etc etc.
To me, this is obvious with AI detectors and AI humanisers. Why would I want to dumb down a perfectly good argument or piece of fiction just to please a machine when the AI written words pack a better punch than mine?
Ok enough wine and enough nonsensical rambling, off to sober up.😂
Thoughts from my first joint 🌱: As someone with ADHD, AI doesn’t replace my thinking. It helps reorganize it. My brain is great at generating rapid-fire associations, connections, and ideas. AI helps reflect them back, scaffold them into an argument, and show me the shape of what I was already trying to say. That’s not outsourcing thought. It’s translating one cognitive style into another.
If irregularity is what reads as human, writers will start manufacturing irregularity on purpose, and the detector ends up authoring the prose style of the next decade. Same loop as your algospeak piece, one layer down. Last time the filter was shaping vocabulary. This time it shapes rhythm.
Hahah you’re seeing my microobsessions. And you’re totally right—now that competent, structured prose is “easy” and producible at scale, what will we value in writing moving forward? The invention of photography made realism “easy” and painting pivoted towards expressionist modes.
Excellent and most informative, Andy. As AI continues to pervade our (world) culture, and even as SAI (Super AI) verges on the event horizon, the debate kerfuffle over 'legitimacy', 'authenticity', and 'human originality' continues full bent. I personally find much of the controversy rather superfluous and irrelevant (to my own circumstances), since I believe that the true core-matrix of the issue is not "Neener-neener-neener! I'm gonna tell mommy on you for cheating!", but '...has the amalgam of human/machine interface produced something (or anything) of substantial value, added insight, new perspective or even perhaps expanded knowledge?' That's my criteria, at any rate, but so many people have an insatiable appetite for gossip, conspiracies, and 'tattling' that we're likely to find ourselves inundated with 'blogger-grade' opinionating and second-guessing for a good, long time.
I personally feel that Homo sapiens is already headed for the 'endangered species (if not 'extinct list') as a failed experiment in bio-chemical evolutionary development anyway, so who really cares if SAI finally decides that humanity is utterly expendable and initiates the BIG final-end-game scenario? Besides, chances are favorable that by the time that occurs, I'll be simply a posthumous genealogical statistic that the Mormon Church is trying to convert to an LDS-baptised (and saved...hallelujah!) soul.
Thanks for this stimulating and fascinatingly provocative commentary, Andy! -K2
I keep hearing about this button but I have yet to see it. I have hunted and hunted for it all over a bunch of different posts and I just can't find it. One Substack author and I even had a bunch of back and forth and I looked everywhere she suggested, but came up empty handed (or empty eyed). This post is no exception. There is simply no AI detect button on my screen.
I’ve been pondering this off and on over the past couple of weeks, since the AI checker has become the greatest trauma the Notes on Substack have ever experienced. 😂 On the one hand, I giggle over AI flagging classic authors as AI-written, since AI was trained on public domain literature to begin with. Classic authors sound like AI because AI was designed to mimic classic authors: cute.
On another hand, being the eternal optimist over here, I have to wonder if the whole mess only gives us authors a unique opportunity to develop our unique voices in such a way that we’re not sounding like AI because we’re not copying the styles of the past but are, in fact, challenging ourselves to *gasp* be original. What if we decide to forget about breaking the mold and go out and cast our own damn mold? But that’s just me, so. 🤣
It’s the new modernism! 🙃 And yeah, that’s the argument that history bares out—photography made realism cheap and easy so painters really began to explore expressionist modes. The question is what happens when competent prose becomes cheap and easily reproduced.
Yesterday, I was listening to a podcast episode done by two gay erotic authors on Substack on how to identify A.I. written ‘slop’. They pointed out that A.I. detectors are currently flawed, partly because they were trained using classic literature.
Also, I nearly failed a term paper in high school on Steinbeck. I had stupidly listened to my conservative parents who essentially told me Steinbeck was a socialist.
However, I am proud to say I met Vonnegut at a book signing when I was in college in the 90s! 🤓
Jealous! (About the Vonnegut)
I would love to read a Steinbeck Substack if he was reincarnated today.
Reading your argument, have you tried the same with Stephen King? While he tweaks it, I feel he is very much a formula writer. He has a methodology that is pretty consistent, he even writes about it himself. I wonder how he would score.
That would be a great one—Stephen King spent fifty years helping define what that statistical average is. If millions of writers learned clarity, pacing, and conversational prose by reading King, then today’s language models have inevitably absorbed those same patterns through the writing culture they were trained on. The detector then mistakes influence for imitation.
I tried to get an AI to write a short story in the style of King. I'm happy to say it was terrible.
Yrs, so yes. Can't wait until you have a second joint 😉
🤣🤣🤣
Thoughts from the bottom of a bottle of wine:
Apparently it's agreed that AI writes like neurodivergent authors, and we are more than likely to be the first flagged as 💯 AI. Bearing in mind AI was trained on human author's work, does that mean most of those authors were neurodivergent, or does it mean neurodivergent authors are equal to, or superior to, AI authorship. Why else would we need a machine to expose work that's not AI or similar to AI. And then dumb it down even more. That thought actually makes me feel a little superior for a change. 😅
Thoughts from the first glass of the second bottle of wine:
With the advancement of technology, especially AI, I had hoped to see the birth of super intelligence, instead I feel we are rewarding the dumbing down of intelligence. You know, keep the masses stupid, turn them into sheep etc etc.
To me, this is obvious with AI detectors and AI humanisers. Why would I want to dumb down a perfectly good argument or piece of fiction just to please a machine when the AI written words pack a better punch than mine?
Ok enough wine and enough nonsensical rambling, off to sober up.😂
Thoughts from my first joint 🌱: As someone with ADHD, AI doesn’t replace my thinking. It helps reorganize it. My brain is great at generating rapid-fire associations, connections, and ideas. AI helps reflect them back, scaffold them into an argument, and show me the shape of what I was already trying to say. That’s not outsourcing thought. It’s translating one cognitive style into another.
If irregularity is what reads as human, writers will start manufacturing irregularity on purpose, and the detector ends up authoring the prose style of the next decade. Same loop as your algospeak piece, one layer down. Last time the filter was shaping vocabulary. This time it shapes rhythm.
Hahah you’re seeing my microobsessions. And you’re totally right—now that competent, structured prose is “easy” and producible at scale, what will we value in writing moving forward? The invention of photography made realism “easy” and painting pivoted towards expressionist modes.
Really enjoyed this. (I say enjoyed but the subject matter of course is enraging)
Excellent and most informative, Andy. As AI continues to pervade our (world) culture, and even as SAI (Super AI) verges on the event horizon, the debate kerfuffle over 'legitimacy', 'authenticity', and 'human originality' continues full bent. I personally find much of the controversy rather superfluous and irrelevant (to my own circumstances), since I believe that the true core-matrix of the issue is not "Neener-neener-neener! I'm gonna tell mommy on you for cheating!", but '...has the amalgam of human/machine interface produced something (or anything) of substantial value, added insight, new perspective or even perhaps expanded knowledge?' That's my criteria, at any rate, but so many people have an insatiable appetite for gossip, conspiracies, and 'tattling' that we're likely to find ourselves inundated with 'blogger-grade' opinionating and second-guessing for a good, long time.
I personally feel that Homo sapiens is already headed for the 'endangered species (if not 'extinct list') as a failed experiment in bio-chemical evolutionary development anyway, so who really cares if SAI finally decides that humanity is utterly expendable and initiates the BIG final-end-game scenario? Besides, chances are favorable that by the time that occurs, I'll be simply a posthumous genealogical statistic that the Mormon Church is trying to convert to an LDS-baptised (and saved...hallelujah!) soul.
Thanks for this stimulating and fascinatingly provocative commentary, Andy! -K2
Less ornate, but my go-to is always the Shirley Bassey line “Why ask your head, when it’s your hips that are swinging?” Thanks for the comment! 💕👏
Shirley Bassey was a favorite of mine, way back when (1964). She had it nailed... -K2
Favorite of mine too 💕
Fascinating post!
I keep hearing about this button but I have yet to see it. I have hunted and hunted for it all over a bunch of different posts and I just can't find it. One Substack author and I even had a bunch of back and forth and I looked everywhere she suggested, but came up empty handed (or empty eyed). This post is no exception. There is simply no AI detect button on my screen.
Thanks Tim! You’re all the better for it, I’d say. We already have a bad writing detector in our skull :-)
So true!
No one is going to survive Substack!
Shoulda been called Domstack 🙃