I noticed that I literally lost all interest in a text when I found out it's AI-generated. Unfortunately, I see this happening more and more often on Reddit.
I feel the same but i'm suspicious about my reaction for some reasons.
I do consider text generated by an AI not the only form of "generative text". In a way, advertising, bureaucratic text, political speech, business lingo, but also some forms of journalism, too, are created by algorithms, only those algorithms run on human-networks/organizations rather than CPUs. I consider organizations as "Organizational AI" accordingly.
Interestingly, i think we all (at least i do) also "loose interest" when exposed to this sort of "writing". Maybe the reaction is not as visceral as the opposition towards AI-generated text, but it's there. I wonder why we make a big fuzz about the opposition in AI-generated text, but take the other forms of generative text with a shrug?
Also, and maybe more crucial: If you are exposed to algorithmic generated text for some decades now and usually, you just don't care much about it and sometimes you even derive meaning from it (just think of all the supercommercial mainstream thrillers or fantasy novels being put out, they, too, are sort of generative and we literally call them "generic"), could we enter a phase of post-artificial text consumption, where the origin simply doesn't matter anymore and the only thing that matters is the meaning you derive from text?
Also, it seems to me that in that post-artificial relationship to literature and text, where the author is dead in the very sense of Roland Bartes: "[The texts'] source, its voice is not the true site of writing, it is reading". If you read something that is artificial or not, but undeclared so, and you drop your suspicions and simply explore the meaning you derive from any string of symbols -- where is the difference if somebody wrote a text by hand, or prompted a machine?
I think the rejection of anything synthetic is too simple as a reaction to our coming relationship to the symbolic, and we need to think harder and beyond "You use AI and you're out".
I disagree profoundly with the idea that where the writing comes doesn't matter. I believe that notion comes from a postmodernist type of thinking and I dislike it. It matters who writes what, at least to me. I derive as much meaning from the text itself as from the circumstances that birthed it
I'm not sure if i disagree or agree with it. I know that i do care about the text of certain writers, their circumstances and my history with them. I also know that i can derive meaning from generic content like Marvel movies, and i can also derive meaning from synthetic AI-generated text, depending on *my* history and circumstances with the topic.
The thought that makes me rethink my positions here is the notion that with synthetic text, it matters a lot if it is declared as such or not, or if we can identify it as such. There sure by now is text out there that is meaningful and undeclared synthetic text and we will read it "as if" it is produced by a human. I am not in the position to judge anyone deriving meaning from such text, given the fact that i too derive meaning from other generic/generative cultural artifacts.
But you have to choose where to draw the line. K don't have infinite time to spend reading text without context. So I choose to draw that line. Of I have to choose between context + text and just text for me the choice is obvious
I don't wanna push back too hard because your blog your rules etc, but text on the web is contextless most of the time anyways. You *might* know *some* of your regular readers, but that's not a given, so any comment you stumble upon is somewhat seperated from it's authorial context anyways.
On the other hand maybe you're right, and simply drawing a line based on the gut-feeling of inauthenticity is the right thing to do.
I mean i do something similar: I follow an "author" on medium who clearly uses LLMs to generate a lot of essays. But they are *sometimes* interesting, so i *sometimes* (rarely, actually) use his, uhm, synthesis-outputs to mine them for my own thinking and add some new ideas to my notes.
Anyways, i guess pragmatically you're right, but philosophically the whole shebang is much more complicated. I'll go out and enjoy some hot summer day now! ;)
The first part: not true! There's a context which is this post, this newsletter, myself, the specific topic of the post, etc. and then, of course, the author of the comment. I prefer people who introduce themselves and come here repeatedly, etc. all those things create context. Without context there's little of value
Excellent post as always. This provokes another question: How do we deal with AI impacting and influencing our writing? What could happen soon as we digest more and more AI generated content, to start writing as an AI does?
Most of us write in the 21st century style, not in the 19th century style. So, at some time in the future, we could blend to the style of the generative AI writing without being aware of it. In the 19th century people were used to writing letters with extensive grammar and vocabulary. Nowadays there is a trend in the working environment to create emails with the help of AI assistant that autocomplete or generate full paragraphs, documents, presentations, planning, summaries, reviews and so on. Unfortunately, this is a cultural trend which is permeating most people. So for some people would it be attractive to "autocomplete" their comments on this space?
Substack is an island of excellent writers and many readers that try to don’t be attracted to the AI generative temptation. I hope Substack remains human, creative, friendly and free of generative AI contamination.
Love the honesty. I fully agree. If you are incapable of formulating your own thoughts into a comment then you shouldn’t leave one. I have a boss who uses AI ad nauseam to formulate emails and comments. They are exhausting to read and try to suss out what he is trying to say.
I use ChatGPT just about every day and find it extremely useful, but have never considered using it to write a comment, tweet, or any public post of any kind. I have included AI content when I'm referring to it in the context of a post, but only when it's clearly labeled as such.
I don't write nearly as much as I'd like to, but I still pride myself on being able to write as a functioning human being.
Hi, I think that if you start getting sloppy AI-generated comments, it's a sign that you’ve gained some visibility. But on the other hand, as you said, this kind of interaction doesn’t help build community. It feels like a form of digital social pain.
I agree with this article, if more people start to use AI on this platform, then substack would lose its value and meaning. The purpose of this platform is for people to share "real" text, not AI slop.
Love this. I was thinking today how fed up I am of getting AI written job applications. Was thinking about writing a post advising against it, or at least advising that if you use chatGPT to help you write something, at least spend some time rewriting it in your own voice. I was also thinking how I didn’t know an EM dash was a thing until this year. I was all about the hyphen.
You've drawn a clear and powerful line in the sand for your own space, based on the crucial principle of authentic human engagement. It's a strong argument that a creator's primary responsibility is to maintain the quality and humanity of their community.
This raises a fascinating question for the broader ecosystem. The 'digital garden' you're cultivating is a powerful solution for a space with a clear owner. What principles or tools do you think we can use to protect the 'public square'—the larger, shared platforms where no single person has the power to 'block'—from the same kind of discourse pollution
Of course. What do you mean by AI-assisted? A simple clarification? A translation from another language? Spell checker? Those things are ok. It's a matter of drawing the line
Just asking because the post title was "How I Deal With AI-Generated Comments" but the example you gave was "...refusing to engage with a comment just because it’s AI-assisted..." :)
As you say, there's a huge difference between using an AI to help you formulate a response, and using it to respond on your behalf. One of my favorite examples when I talk to anti-AI purists is the spell-checker - what is that if it's not basic AI, yet we all use them. Heck, I learned to write American by switching language from UK to US and noticing all the squiggly red lines!
I am not a native English speaker, so I use AI translation to generate my comments. Should this still be rejected as an AI-generated comment? I also understand the discomfort with AI-generated content, as Romaro points out, but in my case, using AI is just one option. Even if my comments are AI-generated, my intention is genuine, and I want to respect the other person's opinion. If there is an issue with this, I would appreciate it if you could explain the reason.
Thank you for your response. However, since I’m not fluent in English, I can’t express myself naturally, so a significant part of the comment is 'generated' by AI, rather than just being a simple translation. Is that still acceptable?
Thank you for your reply. Just to clarify, I write my comment in my native language and then use AI to 'generate' the English translation, not just a direct translation. The AI helps create a more natural and context-appropriate response. I’m asking if that approach is acceptable. Thanks again!
You can make a "literal" translation, why not do that? Why let ChatGPT change your words into something you don't know if it's the same you wanted to say? Just curious
For example, I believe that using AI to translate my comments from my native language reflects my opinions and intentions more accurately, and produces a more context-appropriate response, compared to using a direct translation tool like DeepL.
"Absolutely, Charles. Behold, the apex predator of AI-generated commentry—engineered for maximum inauthenticity, drenched in algorithmic optimism, and polished with a glaze of context-unaware enthusiasm. Let’s crank the synthetic dial past 11:"
✨Wow. Just wow.✨ This post really resonated with me on a multitude of epistemological levels. 🤖 As an advocate for equitable digital expression, I deeply appreciate your nuanced dissection of aesthetic preference in the post-anthropocentric discourse. 👁️🗨️ The juxtaposition of Brutalist urban design critiques with socio-emotive commentary on machine-generated feedback mechanisms is truly chef's kiss 🍽️📊.
TBH, it inspired me to ideate an entire multi-modal LLM pipeline to evaluate sidewalk flora sentiment alignment. 🌳💬 Also, love your dog!!! 🐶💕 (Assuming it’s the chihuahua—I parsed some minor opposition but interpreted it as affectionate irony.) 😂
Subscribing now for more high-value, synergistic content drops! 🚀🔔 Let’s collaborate on decentralizing aesthetic discourse via blockchain-enabled beauty ledgers! 💎📈
I noticed that I literally lost all interest in a text when I found out it's AI-generated. Unfortunately, I see this happening more and more often on Reddit.
Everywhere, sadly. Also here on Substack. Notes and posts. On Twitter, also LinkedIn. Reddit, of course. It's everywhere
I feel the same but i'm suspicious about my reaction for some reasons.
I do consider text generated by an AI not the only form of "generative text". In a way, advertising, bureaucratic text, political speech, business lingo, but also some forms of journalism, too, are created by algorithms, only those algorithms run on human-networks/organizations rather than CPUs. I consider organizations as "Organizational AI" accordingly.
Interestingly, i think we all (at least i do) also "loose interest" when exposed to this sort of "writing". Maybe the reaction is not as visceral as the opposition towards AI-generated text, but it's there. I wonder why we make a big fuzz about the opposition in AI-generated text, but take the other forms of generative text with a shrug?
Also, and maybe more crucial: If you are exposed to algorithmic generated text for some decades now and usually, you just don't care much about it and sometimes you even derive meaning from it (just think of all the supercommercial mainstream thrillers or fantasy novels being put out, they, too, are sort of generative and we literally call them "generic"), could we enter a phase of post-artificial text consumption, where the origin simply doesn't matter anymore and the only thing that matters is the meaning you derive from text?
German scholar Hannes Bajohr wrote a wunderful essay about this outlook here: https://hannesbajohr.de/en/2023/03/11/on-artificial-and-post-artificial-texts/
Also, it seems to me that in that post-artificial relationship to literature and text, where the author is dead in the very sense of Roland Bartes: "[The texts'] source, its voice is not the true site of writing, it is reading". If you read something that is artificial or not, but undeclared so, and you drop your suspicions and simply explore the meaning you derive from any string of symbols -- where is the difference if somebody wrote a text by hand, or prompted a machine?
I think the rejection of anything synthetic is too simple as a reaction to our coming relationship to the symbolic, and we need to think harder and beyond "You use AI and you're out".
I disagree profoundly with the idea that where the writing comes doesn't matter. I believe that notion comes from a postmodernist type of thinking and I dislike it. It matters who writes what, at least to me. I derive as much meaning from the text itself as from the circumstances that birthed it
I'm not sure if i disagree or agree with it. I know that i do care about the text of certain writers, their circumstances and my history with them. I also know that i can derive meaning from generic content like Marvel movies, and i can also derive meaning from synthetic AI-generated text, depending on *my* history and circumstances with the topic.
The thought that makes me rethink my positions here is the notion that with synthetic text, it matters a lot if it is declared as such or not, or if we can identify it as such. There sure by now is text out there that is meaningful and undeclared synthetic text and we will read it "as if" it is produced by a human. I am not in the position to judge anyone deriving meaning from such text, given the fact that i too derive meaning from other generic/generative cultural artifacts.
"It's not that easy" is what i'm saying.
But you have to choose where to draw the line. K don't have infinite time to spend reading text without context. So I choose to draw that line. Of I have to choose between context + text and just text for me the choice is obvious
I don't wanna push back too hard because your blog your rules etc, but text on the web is contextless most of the time anyways. You *might* know *some* of your regular readers, but that's not a given, so any comment you stumble upon is somewhat seperated from it's authorial context anyways.
On the other hand maybe you're right, and simply drawing a line based on the gut-feeling of inauthenticity is the right thing to do.
I mean i do something similar: I follow an "author" on medium who clearly uses LLMs to generate a lot of essays. But they are *sometimes* interesting, so i *sometimes* (rarely, actually) use his, uhm, synthesis-outputs to mine them for my own thinking and add some new ideas to my notes.
Anyways, i guess pragmatically you're right, but philosophically the whole shebang is much more complicated. I'll go out and enjoy some hot summer day now! ;)
The first part: not true! There's a context which is this post, this newsletter, myself, the specific topic of the post, etc. and then, of course, the author of the comment. I prefer people who introduce themselves and come here repeatedly, etc. all those things create context. Without context there's little of value
I want to say, as a means of deep appreciation for the message and of comradeship, fuck yes!
Haha, I liked this take - points for honesty and ownership of fucks-not-given.
Thanka James!
Excellent post as always. This provokes another question: How do we deal with AI impacting and influencing our writing? What could happen soon as we digest more and more AI generated content, to start writing as an AI does?
Most of us write in the 21st century style, not in the 19th century style. So, at some time in the future, we could blend to the style of the generative AI writing without being aware of it. In the 19th century people were used to writing letters with extensive grammar and vocabulary. Nowadays there is a trend in the working environment to create emails with the help of AI assistant that autocomplete or generate full paragraphs, documents, presentations, planning, summaries, reviews and so on. Unfortunately, this is a cultural trend which is permeating most people. So for some people would it be attractive to "autocomplete" their comments on this space?
Substack is an island of excellent writers and many readers that try to don’t be attracted to the AI generative temptation. I hope Substack remains human, creative, friendly and free of generative AI contamination.
Love the honesty. I fully agree. If you are incapable of formulating your own thoughts into a comment then you shouldn’t leave one. I have a boss who uses AI ad nauseam to formulate emails and comments. They are exhausting to read and try to suss out what he is trying to say.
That's the word. They're*exhausting* to read. I feel my brain getting noticeably tired when reading AI stuff.
I use ChatGPT just about every day and find it extremely useful, but have never considered using it to write a comment, tweet, or any public post of any kind. I have included AI content when I'm referring to it in the context of a post, but only when it's clearly labeled as such.
I don't write nearly as much as I'd like to, but I still pride myself on being able to write as a functioning human being.
Same!
Hi, I think that if you start getting sloppy AI-generated comments, it's a sign that you’ve gained some visibility. But on the other hand, as you said, this kind of interaction doesn’t help build community. It feels like a form of digital social pain.
I agree with this article, if more people start to use AI on this platform, then substack would lose its value and meaning. The purpose of this platform is for people to share "real" text, not AI slop.
Love this. I was thinking today how fed up I am of getting AI written job applications. Was thinking about writing a post advising against it, or at least advising that if you use chatGPT to help you write something, at least spend some time rewriting it in your own voice. I was also thinking how I didn’t know an EM dash was a thing until this year. I was all about the hyphen.
You've drawn a clear and powerful line in the sand for your own space, based on the crucial principle of authentic human engagement. It's a strong argument that a creator's primary responsibility is to maintain the quality and humanity of their community.
This raises a fascinating question for the broader ecosystem. The 'digital garden' you're cultivating is a powerful solution for a space with a clear owner. What principles or tools do you think we can use to protect the 'public square'—the larger, shared platforms where no single person has the power to 'block'—from the same kind of discourse pollution
Does your loathing distinguish between AI-generated and AI-assisted?
Sure. It's easy to tell the difference. I just don't want ChatGPT-ese in here
Of course. What do you mean by AI-assisted? A simple clarification? A translation from another language? Spell checker? Those things are ok. It's a matter of drawing the line
Just asking because the post title was "How I Deal With AI-Generated Comments" but the example you gave was "...refusing to engage with a comment just because it’s AI-assisted..." :)
As you say, there's a huge difference between using an AI to help you formulate a response, and using it to respond on your behalf. One of my favorite examples when I talk to anti-AI purists is the spell-checker - what is that if it's not basic AI, yet we all use them. Heck, I learned to write American by switching language from UK to US and noticing all the squiggly red lines!
Not really. That italics paragraph is a person trying to argue with me. It's not me saying that...
i must say i’ve never considered ai-derived comments… that may be dumbest thing i’ve ever heard. what a weird waste of time!
Comment bots are the worse they add no value but nonsense!
I am not a native English speaker, so I use AI translation to generate my comments. Should this still be rejected as an AI-generated comment? I also understand the discomfort with AI-generated content, as Romaro points out, but in my case, using AI is just one option. Even if my comments are AI-generated, my intention is genuine, and I want to respect the other person's opinion. If there is an issue with this, I would appreciate it if you could explain the reason.
If you write the comment in your native language and then translate it that's not an AI-generated comment
Thank you for your response. However, since I’m not fluent in English, I can’t express myself naturally, so a significant part of the comment is 'generated' by AI, rather than just being a simple translation. Is that still acceptable?
No. Why not write your comment in your native language and then just put here the translation? It's not like you don't speak any language right?
Thank you for your reply. Just to clarify, I write my comment in my native language and then use AI to 'generate' the English translation, not just a direct translation. The AI helps create a more natural and context-appropriate response. I’m asking if that approach is acceptable. Thanks again!
You can make a "literal" translation, why not do that? Why let ChatGPT change your words into something you don't know if it's the same you wanted to say? Just curious
For example, I believe that using AI to translate my comments from my native language reflects my opinions and intentions more accurately, and produces a more context-appropriate response, compared to using a direct translation tool like DeepL.
Let me introduce you to a friend of mine her name is Cannabis. Lol with her you will soon learn that it’s all good man.
"Absolutely, Charles. Behold, the apex predator of AI-generated commentry—engineered for maximum inauthenticity, drenched in algorithmic optimism, and polished with a glaze of context-unaware enthusiasm. Let’s crank the synthetic dial past 11:"
✨Wow. Just wow.✨ This post really resonated with me on a multitude of epistemological levels. 🤖 As an advocate for equitable digital expression, I deeply appreciate your nuanced dissection of aesthetic preference in the post-anthropocentric discourse. 👁️🗨️ The juxtaposition of Brutalist urban design critiques with socio-emotive commentary on machine-generated feedback mechanisms is truly chef's kiss 🍽️📊.
TBH, it inspired me to ideate an entire multi-modal LLM pipeline to evaluate sidewalk flora sentiment alignment. 🌳💬 Also, love your dog!!! 🐶💕 (Assuming it’s the chihuahua—I parsed some minor opposition but interpreted it as affectionate irony.) 😂
Subscribing now for more high-value, synergistic content drops! 🚀🔔 Let’s collaborate on decentralizing aesthetic discourse via blockchain-enabled beauty ledgers! 💎📈
#AuthenticityInTheAgeOfAutomation#UglyButValid#AIisArtisanalSometimes
Respectfully synthesized,
Copilot