A few days ago, I asked myself this question: “What happens when AI can say what you want to say better than you?” That question only scratches the surface of the malaise. For more than a year now, I have struggled with what to do, as someone who likes to write, when writing becomes ‘cheap’. Or should I say, to be more precise, that it is actually generating written artifacts that has become easier. There is a feeling that, now that one can write enormous amounts of blogs, essays, and reports, each one becomes less valuable. Text is no longer rare and precious like in ancient times. Just as synthetic diamonds are killing the natural diamond market, there is a sense that the synthetic writings we currently have wide access to are degrading the value of traditional human writing.
> Text is no longer rare and precious like in ancient times.
Text was never rare. What made it special was the people, their experiences, imperfections and the ways they expressed them. It may be corny to say but words do have power - to light up your imagination, let you visit worlds through the eye of the narrator, and spark that tiny flame which could lead to great discoveries.
AI-generated text has none of that effect on me. My brain just shuts down as soon as it detects it. My eyes glaze over and my fickle attention flickers away.
So, if anything, AI is making some human writing way more precious than before.
> AI-generated text has none of that effect on me. My brain just shuts down as soon as it detects it. My eyes glaze over and my fickle attention flickers away.
Does this apply to all AI-generated text? Don't you use ChatGPT or Claude, and thus read AI-generated text?
Long-form text had already declined as a medium before ChatGPT was launched in late 2022. Throughout the 2010s, people's default device switched to the smartphone with its small screen and inefficient keyboard, and social media had already first gone to 180 characters and then gradually to video. Quality blogs got hidden from Google search results in the middle of that decade, and many bloggers gave up or died out -- many were middle aged and it has been that long -- without being replaced (Medium or Substack don't count for me, such platforms are full of pathologies). By the 2020s, writing e.g. a Reddit comment of a mere two solid paragraphs could already make you look like a weirdo who couldn't read the room.
So human-generated long-form text was increasingly irrelevant to the world we live in already before LLMs. If you are, like me, a person who likes the stuff, one can only recognize that tidal shift in society, mourn it, and try to console oneself with the literary canon and surviving material from the early internet, while accepting this is a lonely endeavour.
As the article says, the actual semantic information to be communicated must still come from the prompt. While an LLM might improve the organization and prose wrapped around it, it doesn't add true information itself. I think historically we have conflated these things but these new pressures are forcing them apart. Writing style is going to become more information dense.
> Fully AI-generated content will always be ontologically different from human creative outputs.
I get the vague sense that the author and a contingent of others see the ontological difference as primarily metaphysical. Specifically I see a version of essentialism of interiority that captures what the reader believes about the author(or lack of) and then projects upon the text.
Like all essentialism, it happens usually without the reader knowing that they're doing it - that the observations are mere facts of nature as if the authorship of judgement were erased.
We see this both ways, when text is revealed to be AI written folks will revise their initial praise. Likewise, when text is revealed to be written by a human, previous critique is swept under the rug and interiority that wasn't initial found is somehow revealed. All while the text has remained the same.
Dear readers, this is not an honest way to read. Text doesn't have metaphysical baggage, it is simply words on the page. Interiority is always an internal sense and an individual deduction. It is no more a property of the text than your favorite color actually possessing the property of being the best color in the world.
I'm not saying we have to like AI generated text, but let's leave the metaphysical baggage at home and read things honestly for once. We owe it to ourselves.
You know what's funny is that, in the movie 'Her', Theodore works full time as a personal letter writer (writes letters for other humans). It's funny because, at the time the movie was made the idea of AI had nothing to do with LLMs so humans still had a need to write things themselves despite carrying powerful AI devices in their pockets.
Just an interesting idea I had about how things are envisioned and how they ultimately play out in the real world.
> the idea of AI had nothing to do with LLMs so humans still had a need to write things themselves
Well, not really, they were paying a company to do it. Sometimes throughout their entire relationship. It's possible that for this kind of letters customer valued that a human actually writes this for them. It could have been like paying for a musician rather than listen to the recording of perfect quality.
That's a good point, it may have been implied sub-textually and I missed it. Maybe the movie was actually more predictive than I thought, maybe the writers foresaw how, in a world saturated by AI, humans would ultimately reject AI slop letters, leading to an industry of well-paid human letter writers. Could be
A few days ago, I asked myself this question: “What happens when AI can say what you want to say better than you?” That question only scratches the surface of the malaise. For more than a year now, I have struggled with what to do, as someone who likes to write, when writing becomes ‘cheap’. Or should I say, to be more precise, that it is actually generating written artifacts that has become easier. There is a feeling that, now that one can write enormous amounts of blogs, essays, and reports, each one becomes less valuable. Text is no longer rare and precious like in ancient times. Just as synthetic diamonds are killing the natural diamond market, there is a sense that the synthetic writings we currently have wide access to are degrading the value of traditional human writing.
> Text is no longer rare and precious like in ancient times.
Text was never rare. What made it special was the people, their experiences, imperfections and the ways they expressed them. It may be corny to say but words do have power - to light up your imagination, let you visit worlds through the eye of the narrator, and spark that tiny flame which could lead to great discoveries.
AI-generated text has none of that effect on me. My brain just shuts down as soon as it detects it. My eyes glaze over and my fickle attention flickers away.
So, if anything, AI is making some human writing way more precious than before.
> AI-generated text has none of that effect on me. My brain just shuts down as soon as it detects it. My eyes glaze over and my fickle attention flickers away.
Does this apply to all AI-generated text? Don't you use ChatGPT or Claude, and thus read AI-generated text?
Long-form text had already declined as a medium before ChatGPT was launched in late 2022. Throughout the 2010s, people's default device switched to the smartphone with its small screen and inefficient keyboard, and social media had already first gone to 180 characters and then gradually to video. Quality blogs got hidden from Google search results in the middle of that decade, and many bloggers gave up or died out -- many were middle aged and it has been that long -- without being replaced (Medium or Substack don't count for me, such platforms are full of pathologies). By the 2020s, writing e.g. a Reddit comment of a mere two solid paragraphs could already make you look like a weirdo who couldn't read the room.
So human-generated long-form text was increasingly irrelevant to the world we live in already before LLMs. If you are, like me, a person who likes the stuff, one can only recognize that tidal shift in society, mourn it, and try to console oneself with the literary canon and surviving material from the early internet, while accepting this is a lonely endeavour.
There's also the factor of diminishing returns on mass produced text that shares a common voice and idea space.
This is not unique to text.
Why develop a game if AI generated games makes them less valuable?
Why take a photo or draw something if AI generated imagery makes images less valuable?
Ultimately, just do the thing if it pleases you. Don’t worry about the value.
Some people will always find value in things anyway if they know they are human created.
As the article says, the actual semantic information to be communicated must still come from the prompt. While an LLM might improve the organization and prose wrapped around it, it doesn't add true information itself. I think historically we have conflated these things but these new pressures are forcing them apart. Writing style is going to become more information dense.
> Fully AI-generated content will always be ontologically different from human creative outputs.
I get the vague sense that the author and a contingent of others see the ontological difference as primarily metaphysical. Specifically I see a version of essentialism of interiority that captures what the reader believes about the author(or lack of) and then projects upon the text.
Like all essentialism, it happens usually without the reader knowing that they're doing it - that the observations are mere facts of nature as if the authorship of judgement were erased.
We see this both ways, when text is revealed to be AI written folks will revise their initial praise. Likewise, when text is revealed to be written by a human, previous critique is swept under the rug and interiority that wasn't initial found is somehow revealed. All while the text has remained the same.
Dear readers, this is not an honest way to read. Text doesn't have metaphysical baggage, it is simply words on the page. Interiority is always an internal sense and an individual deduction. It is no more a property of the text than your favorite color actually possessing the property of being the best color in the world.
I'm not saying we have to like AI generated text, but let's leave the metaphysical baggage at home and read things honestly for once. We owe it to ourselves.
You know what's funny is that, in the movie 'Her', Theodore works full time as a personal letter writer (writes letters for other humans). It's funny because, at the time the movie was made the idea of AI had nothing to do with LLMs so humans still had a need to write things themselves despite carrying powerful AI devices in their pockets.
Just an interesting idea I had about how things are envisioned and how they ultimately play out in the real world.
> the idea of AI had nothing to do with LLMs so humans still had a need to write things themselves
Well, not really, they were paying a company to do it. Sometimes throughout their entire relationship. It's possible that for this kind of letters customer valued that a human actually writes this for them. It could have been like paying for a musician rather than listen to the recording of perfect quality.
This movie was prescient in many ways.
That's a good point, it may have been implied sub-textually and I missed it. Maybe the movie was actually more predictive than I thought, maybe the writers foresaw how, in a world saturated by AI, humans would ultimately reject AI slop letters, leading to an industry of well-paid human letter writers. Could be