I for one struggle with the uncertainty, and AI has only added to the swirling grey area that is reality. We definitely need to get more comfortable with just not knowing. Focus on what’s good versus not good, not what has “too many em dashes.”
The audience was dead before 2024 when AI debuted, because every major company jumped into social media for free marketing and advertising to quit from spending money, and that created enshittification bots.
AI now is just throwing salt on the fucking wound. People are tired of it all.
[If I’m putting all my cards on the table, I would also need to admit that I fed this article into ChatGPT and Claude. I use LLMs to fact-check, find typos, critique arguments, and poke holes in my logic all the time.]
And your writing is surely better for the extra time and effort you expended improving your draft using AI tools.
Instead of getting indignant, I try to savor the irony when I spend several days creating a video with AI tools only for someone to spend 10 seconds denouncing the result as AI slop. Their comments never reference the content of the work and could easily be a copy & paste from the same comment they've posted hundreds of times before.
The funny part is that no one believes in sincerity anymore. “That’s art? It MUST be fake!” Bots, third worlders, agents, and gangstalkers threaten anyone who writes their mind on X or Instagram while dark algorithms shadow ban any commentary outside the narrative. The gaslighting is on you, and you are told you are “weird” and “obsessed.” Psychological warfare is real, and the war is against art. Distract, demoralize, and deradicalize. Continue consuming and upholding the American system.
The problem I find is that the gut check “Is this AI?” is always followed by a quick answer. So many people get their news by only reading the headline. They’re likely doing the same with creator content. They’re thinking, “Ugh, AI” and moving on without a second thought.
"No one really wants to spend time consuming something AI-generated unless they were the one who generated it."
...wait, you're telling me that people who were trained to only care about identity/authorship don't actually care about the information being shared unless the identity of the author is beneficial to their social status??!!!! That's so weird, I wouldn't have ever guessed that would happen... if I was an idiot. Of course people "hate" AI... that is, when they can even tell that it's AI — that's why they use a different AI [pangram] to sniff out the AI usage of others... they don't think AI bad when *they're* using it. It's just another form of gatekeeping.
As we slip deeper into Techno-Feudalism, more people are starting to realize that their "identity" was never for them, it was for the elites. Back in the first era of Feudalism, writers, artists, and musicians used to be employed directly by the aristocracy and/or church — that wasn't a way to make a living for normal people — normal people had to do physical labor. It wasn't until those things became products/services that could be sold to the masses that one could make a living doing them, but even then the main beneficiaries were part of the elite class... that's kinda always been the game. Now they don't need you to make those things, so you have to go back to being a serf doing physical labor along side the people you thought you were better than.
Ah, this is such a great piece! I found it linked by Collin Brooke in his wonderful essay (https://cgbrooke.substack.com/p/the-indurable-lightness-of-ai-slop?r=22pnoc&utm_campaign=post-expanded-share&utm_medium=web), in which he asks some nuanced questions about how readers interact with texts at a complicated and sometimes deeply satisfying level. And today? Substack just announced that from now on, readers can feed any text or passage that's longer than 100 words into an AI detection program called Pangram. This "solution" completely undermines what you're arguing in your piece. Readers become trapped in relating to a text in a doubtful and punitive manner, further damaging the audience/reader experience that used to be so valuable. This is maddening.
Well, that's annoying, but probably necessary. If we can't make people take our writing on faith, then a certificate of authenticity that gets people over the hump wouldn't hurt, even if it's BS.
I've begun sharing my Pangram AI test results on my writing just to shut people up, and it's proven useful. At least, for as long as Pangram errs on our writing being human.
What worries me is the financial incentive for Pangram to deliver the results that people want. If there's more money to be found in justifying people's paranoia, that may begin to influence the results. Again, all Pangram really does is take a guess about text. There's nothing concrete about the proof it provides. It'd be easy to shift the weights on the results to consider text more likely to be AI-generated.
Thanks for your reply! I heard on the news tonight that AI stocks climbed in value; it's clear that these business deals between profitable platforms like Substack and an AI company like Pangram support the current hyped business model in the US. Beyond this obvious situation, what you're saying is indeed worrisome. Yes, if we now expect that any post on Substack might be written by AI, Pangram has to affirm that belief to some extent in order to justify its functionality. All of this works on the premise that most users do not understand how LLMs and their derivatives, like AI detectors, actually work. The illusion is too powerful, engendering strong emotions and user expectations.
In the real world we have something similar: fraudsters made it hard to believe people that genuinely ask for money or advice. Once i reflexively told the guy on the street “I am not interested” without even hearing the question only to hear him yell back “I just asked what time it was”
So AI content can be good, but there is so much knowledge today in fraudsters that it sours everything.
It's the equivalent of Junk Food content. Cheap to produce, highly addictive, neither nutritious nor beneficial.
Is everything made by AI slop?
No; AI can be deployed extremely effectively to generate novel, informative content.
Can humans make slop?
Yes.
The real question: is this thing beautiful, true, and good?
Also: writing, and soon videos, are effectively infinite now. Post scarcity. So in a world of a billion essays you could read, which one should you choose?
Analysis paralysis. Get the algorithm to choose for you.
I've raised a related issue on my Substack: Once reliable websites and pages for historical reference are being flooded with AI slop, some of it remarkably close to actual photos and artifacts of a particular era but not real. For instance, what was once a relative straightforward search into what a 1920s home kitchen would look like and what type of appliances used in it is now an exercise in determining if a specific appliance was real or not.
Other people's AI is actually slowing down my research.
It’s fun that the piece opens by relying on a representation (Crowe rather than Nash)
A writer using dictation is already producing language through a machine-learning system before any accusation of “AI writing” begins. Spellcheck, autocomplete, grammar correction, search ranking, image enhancement, recommendation systems, plagiarism detection, content moderation, detector scores, and platform distribution are all computational interventions. Some are explicitly AI. Many are treated as ordinary tools because they were in use before the current hub-bub.
That distinction matters because “human-made” is not actually a pure category anymore. It is usually a negotiated stack of human thoughtwork plus machine assistance, with different tools being granted different levels of legitimacy for historical rather than principled reasons.
Framing the use of LLMs for fact-checking, critique, and copyediting, as a confession— that such use complicates purity… is never turned into what I was hoping for: purity was already fictional before generative AI. (Without that move, “mystery meat” stays a contamination metaphor. Nearly all contemporary digital culture is already *processed food.*)
I tend to think in popular culture, so Crowe it is.
I completely agree with your take on the ending. That was actually going to be part of my original conclusion, but there were a few reasons I backed away from it.
The biggest is that the mystery meat isn’t contaminated. It can range from 100% pure, lovingly handcrafted by an Amish guy to complete machine-generated slop. Its contents aren’t what make it mystery meat. It’s the mystery. You don’t know how it was made.
I also knew the audience for this piece would have wildly different views on AI. Leaning into “we’re all sinners when you think about it” or arguing that everything is AI-assisted now wouldn’t land with the purists, and I think it misses the more important point anyway.
There will never be a consensus on the acceptable extent of AI use. More importantly, we’ll never have a reliable way to know or enforce it. Ultimately, we don’t know how the mystery meat is made, and I think that uncertainty just has to become okay.
My confession at the end was really there to throw the reader back into that uncertainty and make them realize the article itself had been Exhibit A all along. It was always mystery meat.
As for finding the line on acceptable AI use, I actually plan for that to be a completely separate essay. Every writer has to decide where that line is for themselves. It’s discussion I deliberately avoided here because this is more focused on the consumer.
I for one struggle with the uncertainty, and AI has only added to the swirling grey area that is reality. We definitely need to get more comfortable with just not knowing. Focus on what’s good versus not good, not what has “too many em dashes.”
Thanks for the tasty mystery meat.
👏
The audience was dead before 2024 when AI debuted, because every major company jumped into social media for free marketing and advertising to quit from spending money, and that created enshittification bots.
AI now is just throwing salt on the fucking wound. People are tired of it all.
[If I’m putting all my cards on the table, I would also need to admit that I fed this article into ChatGPT and Claude. I use LLMs to fact-check, find typos, critique arguments, and poke holes in my logic all the time.]
And your writing is surely better for the extra time and effort you expended improving your draft using AI tools.
Instead of getting indignant, I try to savor the irony when I spend several days creating a video with AI tools only for someone to spend 10 seconds denouncing the result as AI slop. Their comments never reference the content of the work and could easily be a copy & paste from the same comment they've posted hundreds of times before.
The funny part is that no one believes in sincerity anymore. “That’s art? It MUST be fake!” Bots, third worlders, agents, and gangstalkers threaten anyone who writes their mind on X or Instagram while dark algorithms shadow ban any commentary outside the narrative. The gaslighting is on you, and you are told you are “weird” and “obsessed.” Psychological warfare is real, and the war is against art. Distract, demoralize, and deradicalize. Continue consuming and upholding the American system.
And as AI gets more and more prevalent, there will be a greater and greater onus on creators to prove their authenticity. Strange world.
The problem I find is that the gut check “Is this AI?” is always followed by a quick answer. So many people get their news by only reading the headline. They’re likely doing the same with creator content. They’re thinking, “Ugh, AI” and moving on without a second thought.
"No one really wants to spend time consuming something AI-generated unless they were the one who generated it."
...wait, you're telling me that people who were trained to only care about identity/authorship don't actually care about the information being shared unless the identity of the author is beneficial to their social status??!!!! That's so weird, I wouldn't have ever guessed that would happen... if I was an idiot. Of course people "hate" AI... that is, when they can even tell that it's AI — that's why they use a different AI [pangram] to sniff out the AI usage of others... they don't think AI bad when *they're* using it. It's just another form of gatekeeping.
As we slip deeper into Techno-Feudalism, more people are starting to realize that their "identity" was never for them, it was for the elites. Back in the first era of Feudalism, writers, artists, and musicians used to be employed directly by the aristocracy and/or church — that wasn't a way to make a living for normal people — normal people had to do physical labor. It wasn't until those things became products/services that could be sold to the masses that one could make a living doing them, but even then the main beneficiaries were part of the elite class... that's kinda always been the game. Now they don't need you to make those things, so you have to go back to being a serf doing physical labor along side the people you thought you were better than.
Ah, this is such a great piece! I found it linked by Collin Brooke in his wonderful essay (https://cgbrooke.substack.com/p/the-indurable-lightness-of-ai-slop?r=22pnoc&utm_campaign=post-expanded-share&utm_medium=web), in which he asks some nuanced questions about how readers interact with texts at a complicated and sometimes deeply satisfying level. And today? Substack just announced that from now on, readers can feed any text or passage that's longer than 100 words into an AI detection program called Pangram. This "solution" completely undermines what you're arguing in your piece. Readers become trapped in relating to a text in a doubtful and punitive manner, further damaging the audience/reader experience that used to be so valuable. This is maddening.
Well, that's annoying, but probably necessary. If we can't make people take our writing on faith, then a certificate of authenticity that gets people over the hump wouldn't hurt, even if it's BS.
I've begun sharing my Pangram AI test results on my writing just to shut people up, and it's proven useful. At least, for as long as Pangram errs on our writing being human.
What worries me is the financial incentive for Pangram to deliver the results that people want. If there's more money to be found in justifying people's paranoia, that may begin to influence the results. Again, all Pangram really does is take a guess about text. There's nothing concrete about the proof it provides. It'd be easy to shift the weights on the results to consider text more likely to be AI-generated.
Thanks for your reply! I heard on the news tonight that AI stocks climbed in value; it's clear that these business deals between profitable platforms like Substack and an AI company like Pangram support the current hyped business model in the US. Beyond this obvious situation, what you're saying is indeed worrisome. Yes, if we now expect that any post on Substack might be written by AI, Pangram has to affirm that belief to some extent in order to justify its functionality. All of this works on the premise that most users do not understand how LLMs and their derivatives, like AI detectors, actually work. The illusion is too powerful, engendering strong emotions and user expectations.
In the real world we have something similar: fraudsters made it hard to believe people that genuinely ask for money or advice. Once i reflexively told the guy on the street “I am not interested” without even hearing the question only to hear him yell back “I just asked what time it was”
So AI content can be good, but there is so much knowledge today in fraudsters that it sours everything.
Unfortunately, it really is some people’s fault
I've done that before. Let me guess: You live in LA?
Far away in Europe
The solution to this problem is to live a more interesting life.
https://jesscarterjd.substack.com/p/the-ballad-of-lucy-mcflufflebutt?r=7j1x4v&utm_medium=ios
What is AI Slop?
It's the equivalent of Junk Food content. Cheap to produce, highly addictive, neither nutritious nor beneficial.
Is everything made by AI slop?
No; AI can be deployed extremely effectively to generate novel, informative content.
Can humans make slop?
Yes.
The real question: is this thing beautiful, true, and good?
Also: writing, and soon videos, are effectively infinite now. Post scarcity. So in a world of a billion essays you could read, which one should you choose?
Analysis paralysis. Get the algorithm to choose for you.
I've raised a related issue on my Substack: Once reliable websites and pages for historical reference are being flooded with AI slop, some of it remarkably close to actual photos and artifacts of a particular era but not real. For instance, what was once a relative straightforward search into what a 1920s home kitchen would look like and what type of appliances used in it is now an exercise in determining if a specific appliance was real or not.
Other people's AI is actually slowing down my research.
It’s fun that the piece opens by relying on a representation (Crowe rather than Nash)
A writer using dictation is already producing language through a machine-learning system before any accusation of “AI writing” begins. Spellcheck, autocomplete, grammar correction, search ranking, image enhancement, recommendation systems, plagiarism detection, content moderation, detector scores, and platform distribution are all computational interventions. Some are explicitly AI. Many are treated as ordinary tools because they were in use before the current hub-bub.
That distinction matters because “human-made” is not actually a pure category anymore. It is usually a negotiated stack of human thoughtwork plus machine assistance, with different tools being granted different levels of legitimacy for historical rather than principled reasons.
Framing the use of LLMs for fact-checking, critique, and copyediting, as a confession— that such use complicates purity… is never turned into what I was hoping for: purity was already fictional before generative AI. (Without that move, “mystery meat” stays a contamination metaphor. Nearly all contemporary digital culture is already *processed food.*)
It’s a good read. Nice essay.
I tend to think in popular culture, so Crowe it is.
I completely agree with your take on the ending. That was actually going to be part of my original conclusion, but there were a few reasons I backed away from it.
The biggest is that the mystery meat isn’t contaminated. It can range from 100% pure, lovingly handcrafted by an Amish guy to complete machine-generated slop. Its contents aren’t what make it mystery meat. It’s the mystery. You don’t know how it was made.
I also knew the audience for this piece would have wildly different views on AI. Leaning into “we’re all sinners when you think about it” or arguing that everything is AI-assisted now wouldn’t land with the purists, and I think it misses the more important point anyway.
There will never be a consensus on the acceptable extent of AI use. More importantly, we’ll never have a reliable way to know or enforce it. Ultimately, we don’t know how the mystery meat is made, and I think that uncertainty just has to become okay.
My confession at the end was really there to throw the reader back into that uncertainty and make them realize the article itself had been Exhibit A all along. It was always mystery meat.
As for finding the line on acceptable AI use, I actually plan for that to be a completely separate essay. Every writer has to decide where that line is for themselves. It’s discussion I deliberately avoided here because this is more focused on the consumer.