I’ve been building my own AI prototypes in order to grapple with the same kinds of questions.
Just reading through, I don’t think either of you are wrong, exactly, but I do think you’re circling a pretty closed-loop (lol tech joke) version of the AI-literature debate—one that’s still operating within the air-hockey vibes of resistance vs. acceptance.
What feels like the more urgent and useful question isn’t whether AI should be resisted or embraced, but how it fundamentally changes the cognitive landscape of readers, writers, and storytelling itself. I took a break from reading books as a test recently, and I had to actively retrain myself in order to pay attention and consume the way I usually did.
The real conversation isn’t just about AI as a tool, or even AI as a threat—it’s about what kinds of stories will feel necessary, inevitable, and alive in a world where AI is shaping the way people think, process, and engage with narrative.
AI at its core is about pattern recognition, synthesis, and predictive modelling.
It simply doesn’t create the way we do, and it is built off what is called “completion logic”.
Instead of tangentially spinning ideas as an individual does, through experience and influence, it remixes and recombines anything existing. Funnily enough, it’s kind of similar to Burroughs in that way. Lol.
People already have a sense of pattern recognition (apophenia is bloody real), but remove the way we currently think, to when we start thinking more like AI. When our cognitive habits shift toward instant synthesis, hyperlinked thought, and fragmented information retrieval, it will change the way we read. The literature that emerges from this era (and it will, literature is just communication when you boil it down, and we will unflinchingly still be programmed biologically to do this with one another) won’t just be responding to AI as a subject—it will be responding to the ways AI has altered our ability to perceive, structure, and understand stories.
I think to give you both some hope, the gaming world has already been reshaping how narrative functions, and it’s done so in ways that literary fiction hasn’t fully reckoned with yet.
Systemic narratives, where the “story” isn’t just written but generated through dynamic interactions between AI systems and player agency.
Procedural storytelling, where writers create frameworks for stories rather than single, authored texts—essentially curating experiences rather than dictating them.
Nonlinear storytelling, where players construct meaning through fragmented or emergent experiences.
If we look at where narrative is evolving in gaming, it becomes clear that AI’s influence on literature won’t just be about generating words from thought—it will be about fundamentally changing how stories unfold, how readers engage with them, and what it means to “experience” a narrative. When the way we experience things evolves with everything around us, as to will the ways we communicate.
There’s also the question of literary nostalgia and the canon, which neither of you quite dig into. The literary establishment treats old books as sacred, but at some point, today’s books will be the past, and people will frame them through the same kind of nostalgia. I think I’m more interested in what the dominant literary homogeneity of right now—the thing future readers will see as the defining limitation of things written in this time. Because this is what will be absolutely pushed against when AI-era literature fully matures.
If literature is always evolving in response to cultural and technological shifts, then what’s coming next isn’t just about “AI-generated slop” vs. “human-authored truth.” Maybe the most engaging writing of the future won’t be the kind that tries to out-human AI—but the kind that embraces and metabolises the way AI has rewired our brains.
It’s here, regardless of whether anyone accepts it or not, and why that may be. Ultimately you write to be read by others, so personal opinions on the matter are really just “intellectual clouding” of what could be a seriously hopeful conversation between the two of you who are quite obviously talented, engaged and critical thinkers.
I think we all need to get out of fear + acceptance of AI. It's irrelevant whether AI is good or bad. Not whether it replaces or complements human creativity. But how it’s already shaping the conditions of literature’s future—whether we like it or not.
I’m really glad you posted this though, as it made me think. As per, reading is supreme.
A thought here - based on both my own experiences/experiments with AI (for business, creativity and research purposes) and this interaction. The use of AI for many has taken on a moral tone - in many spheres. It's seen as a shortcut, as generating useless, soulless content, that its brain deadening. What this discourse fails to do is see AI as it really is - a tool. A tool unlike any other tool we've ever had at our disposal.
AI can be all of those pejorative things, but it's also an incredible assistant in creativity. Here's an example --
AI allows for the creation of characters -- who can reason and converse on their own -- through clever context and prompting. Is it soulless writing to take the time to create such an avatar, converse with it -- in the context of the story you're writing -- and use that prose in a novel? Might a character that *can reason on terms you've creatively set* surprise you with what it says? Is this art? I don't know! But its definitely something we need to reckon with and not ignore.
Writing is a craft - it is a technical skill. What you do with it is what makes it art or not. The boundary between craft and art has always been a fuzzy one (the Greek word techne encompassed both).
Like every craft, the tools have changed over the years and I see AI in that same mould.
We don't think anything less of a carpenter for using electric machines in their work but people will, generally, pay more for a piece of furniture made by a craftsperson, often seen as a unique work of art, than they will one made solely by machines.
this is the “talking past one another” mentioned in the post. one of them sees implementation as inevitable, whereas the other is yearning for a “handmade” piece of literature. it may be a new meme, but it is a relevant one. people will still value the “old ways” as they do today, and we may even see elevation of AI-free writing as we do for craftsmanship today.
All of this hemming and hawing about authenticity is entirely beside the point. The real point is whether AI can do what a human can do (most cogently in this particular discussion, in a piece of writing). It can’t. And it won’t ever, not as AI as it’s currently understood, not at least until it is an entirely other thing—that is, actually intelligent. Take one example: irony. Not irony as a funny one off—Bart Simpson delighting in a superhero knocking a villain to the sun and then quipping, “Hot enough for you?”—but irony as the state of expressing a knowing of something about the world and also not knowing that thing all at once, and the tensions created in that dichotomy. This is something that a good writer, not even a great one, can accomplish, but that AI can’t, and won’t ever be able to do. To pull off that kind of irony requires a consciousness that can both be in a moment of thought while also being able to stand apart from it, reflecting upon and back at itself. AI has no such consciousness or ability.
Confident claims about what AI will never be able to do strike me as a bit strange, especially given the rapid improvements we're all currently witnessing. Such claims were more understandable ten years ago (though they still fell victim to a logical fallacy) but now they just sound naive, like you're intentionally setting yourself up to look silly 2 years from now.
The logical fallacy is this: you're assuming that the one way a certain outcome has always been achieved is necessarily the ONLY way that outcome can be achieved. But that's not something we can know ahead of time. There are often multiple ways to do things.
You are 100% correct that AI lacks "a consciousness that can both be in a moment of thought while also being able to stand apart from it..." But this has little to do with whether AI can generate an arrangement of words that create the effect of "irony as the state of expressing a knowing of something about the world and also not knowing that thing all at once..." Obviously, if it achieves that effect, it will do so by different means than we have. Just as it accomplishes the writing tasks it does now (which not long ago had only ever been achieved by a human consciousness) by different means than we do.
The most responsible way for us to approach this discussion is to assume that AI can in principle produce anything we can, and eventually will (we could turn out to be wrong, but isn't it better to be prepared than sorry?). Then ask how we feel about that, and what we'd like to do about. I.e., some hemming and hawwing over authenticity is warranted.
I share Henry's expectation: "One day soon, you will read something that moves you, and you will not know, or be able to know, if a human wrote it."
So far, ai has done much less in much more time as any of its proponents or enemies has forecasted. Each time, I was right in my forecast that it wouldn’t do much at all (in terms of creative writing). Every new exclamation or consternation of increased capacity has been, to my eyes, a bit more whipped cream on a cardboard cake. I haven’t seen any true improvements over time. More rhetorical flourishes, a bit more of a “natural” voice, still no storytelling ability. Rather than a 7th grade book report, a bored poet hired to write cliff’s notes. One of these days I might be wrong, it’ll actually get better. That’s ok. I’ll still make my forecast. I don’t actually care if I’m wrong. I won’t use ai to write either way. Maybe I’ll get “left behind.” I doubt it. My house doesn’t have smart systems or whatever—no getting left behind. Left behind in what? There are 7 billion things to write about and 7 thousand ways to write them. Why does someone need to write with or about ai to “stay ahead”? And prepare for what? Whatever ai does or does not do is out of my control. I mean, I’m still talking about creative writing. If ai starts killing wolves or something, then I’d “do something” or something. Write stories? I don’t care. And anyways, the stories are always going to be bad. Have a good day.
"rather than a 7th grade book report, a bored poet hired to write Cliff's notes" 😂 well put, and yeah I can't deny that so far we've seen more hype than substance. We'll see what happens!
As a philologist in the old Russian style, I understand nothing about technology, but I accept the normality of technical progress in some very small technical degree. Lev Tolstoy could use AI in describing war scenes, but what do with wounded Andrei Bolkonsky, lying under the tree and thinking about death and life? Or Natasha's movement of the heart to give all their horses for the wounded? With AI, literature will lose a writer's individualism, style, talent, psychology, philosophy, and so on... I remember that in the Soviet Union, only one method was permitted for arts - Social Realism. The result was -colorless, impotent literature. (I don't talk about Pasternak. He was Russian, not Soviet, or about the Nobel prize with Sholochov; his novel was plagiarism. Will the same be with AI in literature, especially poetry?
I'm with Henry! AI has never been bitten by a rattle snake, frostbite, had to stop for anautomobile accident and witness mangled bodies, and all other human emotions. But generating graphics and such, that's OK by me. I just hope Walgreens pharmacy catches wind of AI so I can get my fkg scripts on time.
Eisenhower was the last genuine adult to preside over the USA, probably the last we will ever see. There is something fascinating about him, especially watching video clips of him.
On the issue of AI, it is an incoming tide. I cannot bring myself to have an opinion about it. So far I haven't used it for anything, but I eventually I will have to. We are all going to have to surf on that incoming tide, and we will probably find many surprises and very few accurate predictions as we go.
This was a very interesting discussion, and I enjoyed getting both perspectives. Your original post regarding The Metropolitan Review got me to check them out. Also, this post linking to the Eisenhower excerpt from your book was my first time reading that story. Thanks for the good rabbit hole!
I think it's pretty clear that Henry values the art of literature.
I think the key point he's making, in starker terms, is this: It is entirely possible to be moved by a piece of writing (or an artistic artifact more generally) without regard to its source or the details of its creation.
An emotional exchange (between a human creator and the audience moved by their creation) is absolutely magical. But it does not preclude the possibility of, nor threaten the validity of, an experience of being equally moved as a consumer of art without a human creator. In the latter case, the emotional exchange is absent (or the nature of the exchange is different) but a compelling and worthwhile experience may still be there (one half of the exchange is preserved).
In practice, I believe we all have both kinds of experience as consumers of art all the time. Many of us are only compelled to Google stalk the creators of the art that moves us immediately AFTER we have been moved. In the meantime, we are simply being moved. How many times have I read a novel with no knowledge of the author besides a name on the cover? Many, typically because it comes recommended from a source I trust. And what I'm trusting from that source is a high likelihood that I will have a compelling experience in reading that book, not necessarily that I will have an emotional exchange with that author, though that is how all great books have come about so far (until, perhaps, AI).
I think what Henry's responding to (and it frustrates me as well), is a tendency among anti-AI folks to claim that they only do the emotional exchange kind of consumption, or that only an emotional exchange can result in a compelling experience. Those who have incessantly repeated this since ChatGPT burst onto the scene, and never said it before that, seem disingenuous. It seems like they are parroting an anti-AI talking point, without fully interrogating their own experience and behavior.
The fundamental difference is that AI has no lived physical experience, it’s a repository of the art of people who have. It is derivative, it is mechanical, and it is theft. It degrades over time, because it is feeding on itself in a loop. I would argue that the biggest proponents of AI in art are people who lack an understanding of their own creativity, or the patience and discipline it takes to hone their creativity to create art. They just want to steal from others, push a few buttons, and call themselves artists.
This critique assumes a binary that I think is unhelpful: EITHER a human writes the entire piece OR an AI writes the entire piece ("at the push of a button")
Realistically, there's an infinite spectrum of hybrid authorship approaches where a skilled author could use AI for portions of the text or specific elements of writing (this is already happening). As far as how much of the final wordcount was produced by the human vs. by the AI, we could imagine a 90/10 split, 50/50, etc. It probably also matters to us WHICH words were produced by which entity in many cases. More broadly, I think further progress in this discussion demands that we start talking about concrete examples and debating them case by case.
In the meantime, what do you think of the following:
I'm a writer with a deep reverence for the craft and pride in my voice and the unique individual perspective I bring to my fiction. Part of self-awareness as an artist is knowing both your strengths and weaknesses (all authors have them). It's certainly wonderful to improve weaknesses over time, through, in your own eloquent phrasing, "the patience and discipline it takes to hone their creativity...". But another aspect of "understanding your own creativity" is realizing that playing to your strengths is ultimately more important than spending that same time trying to fix all your weaknesses. We love and remember our favorite artists for what they do uniquely well. We love some of them despite identifiable flaws, or relative gaps in their craft toolkit.
Do you think there's a valid use of AI for authors who want to mitigate a weakness, specifically to bring out and complement their individual special sauce, and as a result have more time to hone the uniquely compelling thing that readers are reading them for, and produce more total works over the course of a limited lifetime??
Of course I do! That's a very nuanced question. I think you're suggesting using AI as a partner, much like you'd use a human editor. There's a profound difference between that and tapping a prompt into a keyboard to generate some finished work of art that is entirely derived by the creativity of other people. To me, the real promise of AI in partnership with humans is exactly what you're describing, using it as a tool in your toolbox to improve your work.
"At the end of the thirteenth century Raymond Lully (Ramon Lull) attempted to solve all the mysteries by means of a frame with unequal, revolving, concentric disks, subdivided into sectors with Latin words. At the beginning of the nineteenth century John Stuart Mill expressed the fear that the number of musical combinations would some day be exhausted and that the future would hold no place for new Webers and Mozarts. At the end of the nineteenth century Kurd Lasswitz played with the overwhelming fantasy of a universal library that would record all the variations of the twenty-odd orthographic symbols, or rather everything that can be expressed, in all the languages of the world. Lull’s machine, Mill’s fear, and Lasswitz’s chaotic library may make us laugh, but they merely exaggerate a common propensity to consider metaphysics and the arts as a sort of combinatory game. Those who play that game forget that a book is more than a verbal structure, or a series of verbal structures; a book is the dialogue with the reader, and the peculiar accent he gives to its voice, and the changing and durable images it leaves in his memory. That dialogue is infinite. Now the words arnica silentia lunae mean “the intimate, silent, and shining moon,” and in the Aeneid they meant the interlunar period, the darkness that permitted the Greeks to enter the citadel of Troy. Literature is not exhaustible, for the sufficient and simple reason that a single book is not. A book is not an isolated entity: it is a relationship, an axis of innumerable relationships. One literature differs from another, either before or after it, not so much because of the text as for the manner in which it is read. If I were able to read any contemporary page— this one, for example—as it would be read in the year 2000, I would know what literature would be like in the year 2000. The concept of literature as a formal game leads, in the best of cases, to the good work of the period and the strophe, to a proper craftsman (Johnson, Renan, Flaubert), and in the worst of cases, to the vexations of a work formed of surprises dictated by vanity and chance (Gracian, Herrera Reissig).
If literature were nothing but verbal algebra, anyone could produce any book simply by practicing variations. The lapidary formula Everything flows abbreviates the philosophy of Heraclitus in two words. Raymond Lully would tell us that, after saying the first word, one needs only to substitute intransitive verbs in order to discover the second word and to obtain, by a methodical chance, that philosophy and many, many more. But we would reply that the formula obtained by elimination would lack value and even meaning. If it is to have any virtue we must conceive it as Heraclitus did, as an experience of Heraclitus, although “Heraclitus” is only the presumable subject of that experience. I said that a book is a dialogue, a form of narration. In the dialogue an interlocutor is not the sum total or the intermediate value of what he says: it is possible for him not to speak and yet to reveal intelligence, or to emit intelligent observations and still reveal stupidity. The same occurs with literature."
> You are not describing what writing is; you are describing how writing makes you feel.
What do you think art is? At its bottom, it's an emotional exchange. The artist puts an emotion into their work. The audience experiences the emotion through the work.
It's so strange seeing these debates. It's always these artless nerds who don't appreciate art in the first place talking about how AI will replace something they don't even value.
My conservative take is that the phenomenon of using generative AI to write literature is bad for literature as it further degrades our collective ability to write, read and understand it.
The question of whether AI will be able to produce great works of art (probably) is not relevant: First because art has no intrinsic value and second because, despite AI's ability to produce great works, it will do so much more rarely than humans otherwise would who had cultivated literary talent. But humans won't bother to cultivate that talent if there's an easier way. To quote Jimmy Eat World, "cheating gets it faster." Instead we shall see an even greater increase of bad and middling works.
Henry, you say that AI will increase access to producing literature but what if that itself is bad? Is making literature good even if it's not done through writing? Is that a good process of formation to go through? I think we ought to have less literature. I think the barrier to make literature ought to be higher than it is now. There is way too much bad writing on the Internet and in our bookstores.
Most of my friends don't read books at all, and I can't say that their lives are diminished by it. Reading and making books can be just as vicious an activity as anything else. Similarly one can have a full life without them. The presence of more books and more readers is not automatically good.
The arrival of AI-generated books won't be a positive thing for literature. We will have more bad books, fewer good ones, and a literary class less able to understand literature.
I feel no hesitancy in saying that AI is an existential threat to our ability to write. If authors let AI write for them, their skills will atrophy over time. The same can be said for any student who expects to be well-educated when it comes time to get a job. Even if every use of “iconic” (everywhere) and myriad cliches are expunged, what AI writes that the human does not is a fake, a fraud, and, like all frauds weakens the real and genuine, even if it is flawed. It’s not an answer to say “no one will know the difference.” It’s ignoring the entire point of AI, which is to “create” what humans are too lazy or oblivious to do themselves.
It is already a challenging space to make a living. The deterioration in taste of readers to a preference for generic genre fiction is already a problem. Good writers, including those that use AI will just get lost in a swamp of shitty content.
I’ve been building my own AI prototypes in order to grapple with the same kinds of questions.
Just reading through, I don’t think either of you are wrong, exactly, but I do think you’re circling a pretty closed-loop (lol tech joke) version of the AI-literature debate—one that’s still operating within the air-hockey vibes of resistance vs. acceptance.
What feels like the more urgent and useful question isn’t whether AI should be resisted or embraced, but how it fundamentally changes the cognitive landscape of readers, writers, and storytelling itself. I took a break from reading books as a test recently, and I had to actively retrain myself in order to pay attention and consume the way I usually did.
The real conversation isn’t just about AI as a tool, or even AI as a threat—it’s about what kinds of stories will feel necessary, inevitable, and alive in a world where AI is shaping the way people think, process, and engage with narrative.
AI at its core is about pattern recognition, synthesis, and predictive modelling.
It simply doesn’t create the way we do, and it is built off what is called “completion logic”.
Instead of tangentially spinning ideas as an individual does, through experience and influence, it remixes and recombines anything existing. Funnily enough, it’s kind of similar to Burroughs in that way. Lol.
People already have a sense of pattern recognition (apophenia is bloody real), but remove the way we currently think, to when we start thinking more like AI. When our cognitive habits shift toward instant synthesis, hyperlinked thought, and fragmented information retrieval, it will change the way we read. The literature that emerges from this era (and it will, literature is just communication when you boil it down, and we will unflinchingly still be programmed biologically to do this with one another) won’t just be responding to AI as a subject—it will be responding to the ways AI has altered our ability to perceive, structure, and understand stories.
I think to give you both some hope, the gaming world has already been reshaping how narrative functions, and it’s done so in ways that literary fiction hasn’t fully reckoned with yet.
Systemic narratives, where the “story” isn’t just written but generated through dynamic interactions between AI systems and player agency.
Procedural storytelling, where writers create frameworks for stories rather than single, authored texts—essentially curating experiences rather than dictating them.
Nonlinear storytelling, where players construct meaning through fragmented or emergent experiences.
If we look at where narrative is evolving in gaming, it becomes clear that AI’s influence on literature won’t just be about generating words from thought—it will be about fundamentally changing how stories unfold, how readers engage with them, and what it means to “experience” a narrative. When the way we experience things evolves with everything around us, as to will the ways we communicate.
There’s also the question of literary nostalgia and the canon, which neither of you quite dig into. The literary establishment treats old books as sacred, but at some point, today’s books will be the past, and people will frame them through the same kind of nostalgia. I think I’m more interested in what the dominant literary homogeneity of right now—the thing future readers will see as the defining limitation of things written in this time. Because this is what will be absolutely pushed against when AI-era literature fully matures.
If literature is always evolving in response to cultural and technological shifts, then what’s coming next isn’t just about “AI-generated slop” vs. “human-authored truth.” Maybe the most engaging writing of the future won’t be the kind that tries to out-human AI—but the kind that embraces and metabolises the way AI has rewired our brains.
It’s here, regardless of whether anyone accepts it or not, and why that may be. Ultimately you write to be read by others, so personal opinions on the matter are really just “intellectual clouding” of what could be a seriously hopeful conversation between the two of you who are quite obviously talented, engaged and critical thinkers.
I think we all need to get out of fear + acceptance of AI. It's irrelevant whether AI is good or bad. Not whether it replaces or complements human creativity. But how it’s already shaping the conditions of literature’s future—whether we like it or not.
I’m really glad you posted this though, as it made me think. As per, reading is supreme.
A thought here - based on both my own experiences/experiments with AI (for business, creativity and research purposes) and this interaction. The use of AI for many has taken on a moral tone - in many spheres. It's seen as a shortcut, as generating useless, soulless content, that its brain deadening. What this discourse fails to do is see AI as it really is - a tool. A tool unlike any other tool we've ever had at our disposal.
AI can be all of those pejorative things, but it's also an incredible assistant in creativity. Here's an example --
AI allows for the creation of characters -- who can reason and converse on their own -- through clever context and prompting. Is it soulless writing to take the time to create such an avatar, converse with it -- in the context of the story you're writing -- and use that prose in a novel? Might a character that *can reason on terms you've creatively set* surprise you with what it says? Is this art? I don't know! But its definitely something we need to reckon with and not ignore.
May I ask which AI does this ?
Anthropic's Claude - Sonnet 3.7 model. If you train the project models, you can get some amazing stuff.
This is a very interesting direction!
Writing is a craft - it is a technical skill. What you do with it is what makes it art or not. The boundary between craft and art has always been a fuzzy one (the Greek word techne encompassed both).
Like every craft, the tools have changed over the years and I see AI in that same mould.
We don't think anything less of a carpenter for using electric machines in their work but people will, generally, pay more for a piece of furniture made by a craftsperson, often seen as a unique work of art, than they will one made solely by machines.
this is the “talking past one another” mentioned in the post. one of them sees implementation as inevitable, whereas the other is yearning for a “handmade” piece of literature. it may be a new meme, but it is a relevant one. people will still value the “old ways” as they do today, and we may even see elevation of AI-free writing as we do for craftsmanship today.
All of this hemming and hawing about authenticity is entirely beside the point. The real point is whether AI can do what a human can do (most cogently in this particular discussion, in a piece of writing). It can’t. And it won’t ever, not as AI as it’s currently understood, not at least until it is an entirely other thing—that is, actually intelligent. Take one example: irony. Not irony as a funny one off—Bart Simpson delighting in a superhero knocking a villain to the sun and then quipping, “Hot enough for you?”—but irony as the state of expressing a knowing of something about the world and also not knowing that thing all at once, and the tensions created in that dichotomy. This is something that a good writer, not even a great one, can accomplish, but that AI can’t, and won’t ever be able to do. To pull off that kind of irony requires a consciousness that can both be in a moment of thought while also being able to stand apart from it, reflecting upon and back at itself. AI has no such consciousness or ability.
Confident claims about what AI will never be able to do strike me as a bit strange, especially given the rapid improvements we're all currently witnessing. Such claims were more understandable ten years ago (though they still fell victim to a logical fallacy) but now they just sound naive, like you're intentionally setting yourself up to look silly 2 years from now.
The logical fallacy is this: you're assuming that the one way a certain outcome has always been achieved is necessarily the ONLY way that outcome can be achieved. But that's not something we can know ahead of time. There are often multiple ways to do things.
You are 100% correct that AI lacks "a consciousness that can both be in a moment of thought while also being able to stand apart from it..." But this has little to do with whether AI can generate an arrangement of words that create the effect of "irony as the state of expressing a knowing of something about the world and also not knowing that thing all at once..." Obviously, if it achieves that effect, it will do so by different means than we have. Just as it accomplishes the writing tasks it does now (which not long ago had only ever been achieved by a human consciousness) by different means than we do.
The most responsible way for us to approach this discussion is to assume that AI can in principle produce anything we can, and eventually will (we could turn out to be wrong, but isn't it better to be prepared than sorry?). Then ask how we feel about that, and what we'd like to do about. I.e., some hemming and hawwing over authenticity is warranted.
I share Henry's expectation: "One day soon, you will read something that moves you, and you will not know, or be able to know, if a human wrote it."
So far, ai has done much less in much more time as any of its proponents or enemies has forecasted. Each time, I was right in my forecast that it wouldn’t do much at all (in terms of creative writing). Every new exclamation or consternation of increased capacity has been, to my eyes, a bit more whipped cream on a cardboard cake. I haven’t seen any true improvements over time. More rhetorical flourishes, a bit more of a “natural” voice, still no storytelling ability. Rather than a 7th grade book report, a bored poet hired to write cliff’s notes. One of these days I might be wrong, it’ll actually get better. That’s ok. I’ll still make my forecast. I don’t actually care if I’m wrong. I won’t use ai to write either way. Maybe I’ll get “left behind.” I doubt it. My house doesn’t have smart systems or whatever—no getting left behind. Left behind in what? There are 7 billion things to write about and 7 thousand ways to write them. Why does someone need to write with or about ai to “stay ahead”? And prepare for what? Whatever ai does or does not do is out of my control. I mean, I’m still talking about creative writing. If ai starts killing wolves or something, then I’d “do something” or something. Write stories? I don’t care. And anyways, the stories are always going to be bad. Have a good day.
"rather than a 7th grade book report, a bored poet hired to write Cliff's notes" 😂 well put, and yeah I can't deny that so far we've seen more hype than substance. We'll see what happens!
As a philologist in the old Russian style, I understand nothing about technology, but I accept the normality of technical progress in some very small technical degree. Lev Tolstoy could use AI in describing war scenes, but what do with wounded Andrei Bolkonsky, lying under the tree and thinking about death and life? Or Natasha's movement of the heart to give all their horses for the wounded? With AI, literature will lose a writer's individualism, style, talent, psychology, philosophy, and so on... I remember that in the Soviet Union, only one method was permitted for arts - Social Realism. The result was -colorless, impotent literature. (I don't talk about Pasternak. He was Russian, not Soviet, or about the Nobel prize with Sholochov; his novel was plagiarism. Will the same be with AI in literature, especially poetry?
Thank you for accepting my POV.
I'm with Henry! AI has never been bitten by a rattle snake, frostbite, had to stop for anautomobile accident and witness mangled bodies, and all other human emotions. But generating graphics and such, that's OK by me. I just hope Walgreens pharmacy catches wind of AI so I can get my fkg scripts on time.
Of some relevance here:
https://aeon.co/essays/why-poetry-is-a-variety-of-mathematical-experience
Eisenhower was the last genuine adult to preside over the USA, probably the last we will ever see. There is something fascinating about him, especially watching video clips of him.
On the issue of AI, it is an incoming tide. I cannot bring myself to have an opinion about it. So far I haven't used it for anything, but I eventually I will have to. We are all going to have to surf on that incoming tide, and we will probably find many surprises and very few accurate predictions as we go.
This humility seems warranted. More of this.
This was a very interesting discussion, and I enjoyed getting both perspectives. Your original post regarding The Metropolitan Review got me to check them out. Also, this post linking to the Eisenhower excerpt from your book was my first time reading that story. Thanks for the good rabbit hole!
Thanks!
I think it's pretty clear that Henry values the art of literature.
I think the key point he's making, in starker terms, is this: It is entirely possible to be moved by a piece of writing (or an artistic artifact more generally) without regard to its source or the details of its creation.
An emotional exchange (between a human creator and the audience moved by their creation) is absolutely magical. But it does not preclude the possibility of, nor threaten the validity of, an experience of being equally moved as a consumer of art without a human creator. In the latter case, the emotional exchange is absent (or the nature of the exchange is different) but a compelling and worthwhile experience may still be there (one half of the exchange is preserved).
In practice, I believe we all have both kinds of experience as consumers of art all the time. Many of us are only compelled to Google stalk the creators of the art that moves us immediately AFTER we have been moved. In the meantime, we are simply being moved. How many times have I read a novel with no knowledge of the author besides a name on the cover? Many, typically because it comes recommended from a source I trust. And what I'm trusting from that source is a high likelihood that I will have a compelling experience in reading that book, not necessarily that I will have an emotional exchange with that author, though that is how all great books have come about so far (until, perhaps, AI).
I think what Henry's responding to (and it frustrates me as well), is a tendency among anti-AI folks to claim that they only do the emotional exchange kind of consumption, or that only an emotional exchange can result in a compelling experience. Those who have incessantly repeated this since ChatGPT burst onto the scene, and never said it before that, seem disingenuous. It seems like they are parroting an anti-AI talking point, without fully interrogating their own experience and behavior.
This is in response to @dms's comment from Mar 4 above, but Substack put it in the wrong place
The fundamental difference is that AI has no lived physical experience, it’s a repository of the art of people who have. It is derivative, it is mechanical, and it is theft. It degrades over time, because it is feeding on itself in a loop. I would argue that the biggest proponents of AI in art are people who lack an understanding of their own creativity, or the patience and discipline it takes to hone their creativity to create art. They just want to steal from others, push a few buttons, and call themselves artists.
Meh.
This critique assumes a binary that I think is unhelpful: EITHER a human writes the entire piece OR an AI writes the entire piece ("at the push of a button")
Realistically, there's an infinite spectrum of hybrid authorship approaches where a skilled author could use AI for portions of the text or specific elements of writing (this is already happening). As far as how much of the final wordcount was produced by the human vs. by the AI, we could imagine a 90/10 split, 50/50, etc. It probably also matters to us WHICH words were produced by which entity in many cases. More broadly, I think further progress in this discussion demands that we start talking about concrete examples and debating them case by case.
In the meantime, what do you think of the following:
I'm a writer with a deep reverence for the craft and pride in my voice and the unique individual perspective I bring to my fiction. Part of self-awareness as an artist is knowing both your strengths and weaknesses (all authors have them). It's certainly wonderful to improve weaknesses over time, through, in your own eloquent phrasing, "the patience and discipline it takes to hone their creativity...". But another aspect of "understanding your own creativity" is realizing that playing to your strengths is ultimately more important than spending that same time trying to fix all your weaknesses. We love and remember our favorite artists for what they do uniquely well. We love some of them despite identifiable flaws, or relative gaps in their craft toolkit.
Do you think there's a valid use of AI for authors who want to mitigate a weakness, specifically to bring out and complement their individual special sauce, and as a result have more time to hone the uniquely compelling thing that readers are reading them for, and produce more total works over the course of a limited lifetime??
Of course I do! That's a very nuanced question. I think you're suggesting using AI as a partner, much like you'd use a human editor. There's a profound difference between that and tapping a prompt into a keyboard to generate some finished work of art that is entirely derived by the creativity of other people. To me, the real promise of AI in partnership with humans is exactly what you're describing, using it as a tool in your toolbox to improve your work.
I direct you to Borges:
"At the end of the thirteenth century Raymond Lully (Ramon Lull) attempted to solve all the mysteries by means of a frame with unequal, revolving, concentric disks, subdivided into sectors with Latin words. At the beginning of the nineteenth century John Stuart Mill expressed the fear that the number of musical combinations would some day be exhausted and that the future would hold no place for new Webers and Mozarts. At the end of the nineteenth century Kurd Lasswitz played with the overwhelming fantasy of a universal library that would record all the variations of the twenty-odd orthographic symbols, or rather everything that can be expressed, in all the languages of the world. Lull’s machine, Mill’s fear, and Lasswitz’s chaotic library may make us laugh, but they merely exaggerate a common propensity to consider metaphysics and the arts as a sort of combinatory game. Those who play that game forget that a book is more than a verbal structure, or a series of verbal structures; a book is the dialogue with the reader, and the peculiar accent he gives to its voice, and the changing and durable images it leaves in his memory. That dialogue is infinite. Now the words arnica silentia lunae mean “the intimate, silent, and shining moon,” and in the Aeneid they meant the interlunar period, the darkness that permitted the Greeks to enter the citadel of Troy. Literature is not exhaustible, for the sufficient and simple reason that a single book is not. A book is not an isolated entity: it is a relationship, an axis of innumerable relationships. One literature differs from another, either before or after it, not so much because of the text as for the manner in which it is read. If I were able to read any contemporary page— this one, for example—as it would be read in the year 2000, I would know what literature would be like in the year 2000. The concept of literature as a formal game leads, in the best of cases, to the good work of the period and the strophe, to a proper craftsman (Johnson, Renan, Flaubert), and in the worst of cases, to the vexations of a work formed of surprises dictated by vanity and chance (Gracian, Herrera Reissig).
If literature were nothing but verbal algebra, anyone could produce any book simply by practicing variations. The lapidary formula Everything flows abbreviates the philosophy of Heraclitus in two words. Raymond Lully would tell us that, after saying the first word, one needs only to substitute intransitive verbs in order to discover the second word and to obtain, by a methodical chance, that philosophy and many, many more. But we would reply that the formula obtained by elimination would lack value and even meaning. If it is to have any virtue we must conceive it as Heraclitus did, as an experience of Heraclitus, although “Heraclitus” is only the presumable subject of that experience. I said that a book is a dialogue, a form of narration. In the dialogue an interlocutor is not the sum total or the intermediate value of what he says: it is possible for him not to speak and yet to reveal intelligence, or to emit intelligent observations and still reveal stupidity. The same occurs with literature."
> You are not describing what writing is; you are describing how writing makes you feel.
What do you think art is? At its bottom, it's an emotional exchange. The artist puts an emotion into their work. The audience experiences the emotion through the work.
It's so strange seeing these debates. It's always these artless nerds who don't appreciate art in the first place talking about how AI will replace something they don't even value.
My conservative take is that the phenomenon of using generative AI to write literature is bad for literature as it further degrades our collective ability to write, read and understand it.
The question of whether AI will be able to produce great works of art (probably) is not relevant: First because art has no intrinsic value and second because, despite AI's ability to produce great works, it will do so much more rarely than humans otherwise would who had cultivated literary talent. But humans won't bother to cultivate that talent if there's an easier way. To quote Jimmy Eat World, "cheating gets it faster." Instead we shall see an even greater increase of bad and middling works.
Henry, you say that AI will increase access to producing literature but what if that itself is bad? Is making literature good even if it's not done through writing? Is that a good process of formation to go through? I think we ought to have less literature. I think the barrier to make literature ought to be higher than it is now. There is way too much bad writing on the Internet and in our bookstores.
Most of my friends don't read books at all, and I can't say that their lives are diminished by it. Reading and making books can be just as vicious an activity as anything else. Similarly one can have a full life without them. The presence of more books and more readers is not automatically good.
The arrival of AI-generated books won't be a positive thing for literature. We will have more bad books, fewer good ones, and a literary class less able to understand literature.
I feel no hesitancy in saying that AI is an existential threat to our ability to write. If authors let AI write for them, their skills will atrophy over time. The same can be said for any student who expects to be well-educated when it comes time to get a job. Even if every use of “iconic” (everywhere) and myriad cliches are expunged, what AI writes that the human does not is a fake, a fraud, and, like all frauds weakens the real and genuine, even if it is flawed. It’s not an answer to say “no one will know the difference.” It’s ignoring the entire point of AI, which is to “create” what humans are too lazy or oblivious to do themselves.
It is already a challenging space to make a living. The deterioration in taste of readers to a preference for generic genre fiction is already a problem. Good writers, including those that use AI will just get lost in a swamp of shitty content.