Dear Living Dark reader,
I bring you something different today. I have been doing a lot of thinking, experimenting, and soul-searching on the matter of artificial intelligence and whether it has any valid place in a writer’s life and work. This post represents my current thinking. As you’ll see at the end of it, I’m also keenly interested to hear your own thoughts.
1. The Art of Writing Without Writing
I’ll start by giving a direct answer to my self-posed question: Yes, a serious writer can use AI without losing his soul—but only by exercising great care, as it’s a deeply fraught and subtle issue. This affirmative response may not be what you were expecting, given the skeptical and negative attitude of my previous writings on artificial intelligence and human creativity. The explanation of how I got here makes up the rest of this essay.
Recently, I wrote the foreword to a new edition of the book The Art of Unwriting by Youri Hermes that was published last month. As indicated by its title, Youri’s book is about “unwriting,” a word and concept that, as I’m only just now noticing, very much wants to remind me of one of my favorite scenes in one of my favorite movies, Enter the Dragon.
“What’s your style?” the cocky New Zealand martial artist asks Bruce Lee in a challenging tone as they’re both riding the boat to the criminal mastermind Han’s island, there to compete in his famous martial arts tournament. “My style?” Bruce replies with supreme disinterest. He reflects a moment and says, “You might call it the art of fighting without fighting.” The New Zealander demands to see some of it. And of course Bruce gives him a pointed demonstration, neatly humiliating the other man with a simple trick that plays on his vanity.
Recasting the point in relation to Youri’s book: Unwriting is to writing what Bruce’s approach was to fighting. It’s the art of writing without writing. Or, as Youri puts it, it’s writing that seems to be its opposite, writing not as a positive creative act requiring effort and ego, but as an act of stepping aside to let something come through that wants to be written. It’s not laziness, nor is it simple non-writing, and it’s certainly not automatic output. Rather, it’s writing as receptivity, as an act of attention and service to what wants to be said.
This is very much in the mold of my own authorial experience and the philosophy it has given rise to. This philosophy or model frames creativity as an inner collaboration with one’s daemon muse or inner genius, in which your conscious role is not to generate creativity but to shepherd it, to midwife it, to serve as the conduit for what presents itself as ideas and fascinations that you didn’t choose, but that have chosen you: your unique gift, vision, onus, burden, and calling. This is the whole substance of my book A Course in Demonic Creativity and a central thematic pillar in my more recent Writing at the Wellspring.
It’s also something that has been dramatically complicated by the advent of artificial intelligence large-language models, which threaten to make mud of the whole thing. The trouble, as I see it, is that AI is the first technology that can convincingly mimic the muse. There’s no more need to wait for inspiration, or to actively court and cultivate it the way I recommend and explain it in the demonic creativity book. You don’t even have to do the more prosaic work of preparing your mind and spirit through things like research, outlining, and making planning notes. Instead, you can simply supply a prompt, sit back, and watch an LLM produce an entire essay, story, poem, play, or book at a moment’s notice. It’s the muse as machine, an authorial genie that generates writing on command. Magical push-button creativity.
Maybe you can see why this bothers me. I have spent years saying in various ways that the best writing feels as if it comes not from the conscious self, but from the wellspring, the daemon, the inner genius, the place beneath, before, or beyond the ego self that signs the contract and checks the email. Youri’s “unwriting” model says much the same thing with its focus on authorial stepping aside. So here’s the challenge posed by AI: If writing at its best has always felt as if it arrives from some mysterious elsewhere beyond the ego, why shouldn’t AI be welcomed as another form of that elsewhere? Why not call the machine a different door into the same mystery? What’s the difference between the inner genius and the large-language model glowing patiently in the browser tab? If you’re someone to whom the concepts of unwriting and the daemon muse naturally appeal—someone who, for instance, reads a publication like this one—have you considered whether the muse and the machine can validly be distinguished from each other, given the latter’s uncanny ability to simulate the former?
And yet, of course, the distinction needs to be made. We obviously can’t just cede writing to a machine. Because, in point of fact, AI can’t actually “write for you,” no matter how many AI evangelists, content hustlers, and desperate students want to think or act as if it can. Watching words appear on a screen is different from generating them yourself, or even from serving as their living conduit. But how, exactly? What is the nature of that difference, and why is it important?
Which all leads back to the question of whether serious writers can use AI without losing their soul. As I said at the outset, I want to say yes. But that yes depends on what we mean by writing, and the soul, and whether we can tell the difference between unwriting and what might be called outsourcing.
2. Not All Elsewheres Are Equal
The beginnings of an answer are found, I think, in the recognition that not all forms of “elsewhere” are equal, and not every act of stepping aside is an act of receptivity. Sometimes, stepping aside is an act of abdication. There’s a difference between getting yourself out of the way and removing yourself from the work.
When you unwrite and collaborate with your daemon, you’re still involved, still engaged, still an active and necessary part of the process. The words, though they form visually before your eyes, flow from an inner synergy within your psyche and form a feedback loop with it. This is how writing is a way of discovering what we actually think, and what we actually mean.
When I taught writing for years to high school and college students, many of them told me, with absolute sincerity, “I know what I want to say, but I don’t know how to say it.” And almost always, the real situation was subtler and more interesting than that, because in fact the student didn’t know, or at least not yet, what he or she wanted to say. The idea hadn’t yet become clear, precisely because it hadn’t gone through the ordeal of articulating it.
Writing isn’t a tool for expressing something that’s fully formed inside us. Instead, it’s one of the ways the inside becomes formed at all. It doesn’t just communicate thought. It’s one of the ways thought happens. You start with something that feels real and pressing but indistinct, and you follow that thread through trial and error, trying to express it adequately to yourself. Eventually, if grace descends or the daemon cooperates, you start to see what you meant. Or rather, you start to mean something more exact than what you meant at the outset.
This is categorically different from what happens when you enter a prompt on your device and receive back a flow of textual output on the screen. The “elsewhere” of your phone’s or laptop’s processor, or of the distant data center where the prompt was algorithmically parsed, is not the elsewhere of your own deep psyche and what lies beyond it, prior to it, the “beyond within.”
3. Writing Was Always a Technology
And yet, having said all that, it’s still the case that AI really can produce writing-like text. It feels like the most significant technological advance in writing since the invention of writing itself—which, as we will do well to remember, was called into question by Plato in the Phaedrus in his myth of Thamus, the king who rejected writing when the god Theuth brought it as a gift to humankind. Thamus made this decision on the grounds that writing would not be a boon as promised, but a bane, because it would allow people to offload their own rightful task of thinking and knowing onto an external crutch, thus encouraging the buildup of a false facsimile of real knowledge and wisdom. As Neil Postman pointed out when referencing this myth, while it’s true that all technologies involve some kind of trade-off, most of us would probably disagree with Thamus and hold instead that writing has constituted a net benefit for the human race. Today, with the advent of LLMs, we have arrived at another inflection point in the long history of our ecological interactions with our own communications technologies, and we’re called upon to make another judgment, with serious consequences.
In other words, and to make the point clear, writing is already a technology. The question before us now is what this new technology of artificial intelligence does to the inward act that writing enables.
4. Orphaned Text and Living Vision
So here’s one reason, and maybe the core one, why the arrival of AI-generated writing is so disorienting. If “writing” means nothing more than producing grammatically coherent sentences in an organized sequence—or producing poetic lines or anything else with a recognizable form—then the machine can already do it, and with pretty impressive fluency. But that’s a definition of writing that confuses the outer form with the inward truth. The more important question isn’t whether AI can produce text that resembles writing, because it obviously can. The question is whether text produced in that way accomplishes the thing that writing, at its deepest, accomplishes for both the human writer and the human reader. Does it focus consciousness? Does it make one person’s inwardness available to another? Does it transmit not just information or style, but the living charge of a particular consciousness encountering itself, the world, and the mystery of both?
I think most of us know by now, from wearisome contact with the world of AI-generated slop, and perhaps from personal experimentation with various LLMs, that text written by AI doesn’t do these things in the way human writing does. Such text often feels hollow. It feels orphaned. An AI-generated passage may be interesting, and it may be practically useful for some purpose, and it may even move us in some ways. But there’s an irreducible and inescapable distinction between an effect produced in a reader by machine-written text and one produced by writing that’s a transmission from another living soul. AI text is like cloud patterns or tea leaves. We read meaning into it, not out of it.
(There is, I should add, another and stranger way to think about this, one that I may return to in a separate essay: the possibility of approaching AI not as an author but as a kind of oracular pattern-recognition engine, something closer to bibliomancy, the I Ching, or Tarot than to ordinary writing. In that mode, the meaning would still arise through the human act of reception and interpretation. The machine would not be “speaking from a soul,” but the human encounter with its output might become meaningful in the way cloudscapes, cards, coins, dreams, and books falling from shelves can become meaningful. But that’s a different question from the one I’m pursuing here.)
This is what I take Thomas Ligotti to mean, or at least part of what I take him to mean, when he says in “The Consolations of Horror” that horror fiction can console because someone else has shared some of our own feelings and made of them a work of art. The consolation is that someone else has been there. Someone else has seen the same thing and sent back a signal. Can a machine do that?
And yet—and I really want to pause to emphasize this, both because I think it’s important and because it contrasts with so much of what I have tended to think over the past few years, and also what most of my writer friends and colleagues are emphatically saying these days—this doesn’t mean AI can never be useful to a writer. It means only that we have to be very clear about what kind of usefulness we’re talking about.
5. The Strange Bonus
Another way to get at this is to notice that real writing has a certain feeling on the writer’s end. I think every writer knows, or at least hopes to know, the state in which something really catches. William Stafford, in his brief essay “A Way of Writing”—which I cherish—spoke of the “strange bonus” that sometimes arrives when the elements of a piece begin to cohere and lead by themselves to new connections. I recognize that so very deeply. For me, it comes as a sense of inner alignment, a sprouting of new networks of connections as I think and feel my way through a new act of writing in real time. It’s a sense of fluency both internal and external, something humming through my eyes, mind, and fingers, and even my chest and wider body, as everything seems to constellate in a kind of creative synchronicity, as if I’m caught in a gravitational field that’s bringing everything within me and around me toward a convergence on the developing page and project at hand.
This all feels wonderful. There’s a huge rush to it. And yes, of course, this means it’s also crushingly awful when it doesn’t work, when for some reason there’s an inner short circuit or a dry well. It’s important to listen to the silences, to heed them. I damaged myself after the publication of Divinations of the Deep, my first book, in 2002, by pressing on with new writing over the next two or three years because I felt I ought to, even though everything in me was curdling and telling me it was time to stop and let the well recharge. It took me a little while to find my way back to the energetic rush of it, to remember how to invite and receive that strange bonus. The experience made me appreciate that blessing all the more. It also helps me to recognize how AI can become dangerous because it offers a way to bypass creative silence instead of honoring it.
This whole phenomenon also gives us, or at least gives me personally, one practical test for AI: Does a given use of the machine help return me to that strange bonus, that living current, that gravitational field of a creative synchronicity? Or does it replace all this with an instantly produced, finished-looking artifact? Does it help me enter the blessed ordeal and gift of writing, or does it give me the ersatz pleasure of having “generated output”? Most pointedly, does it bring me closer to the inward source? Or does it interpose itself between me and that source while pretending to speak in its voice? Instead of helping connect me to my inner genius, does it lure me into outsourcing my writing to something else?
6. Assistance or Avoidance
If AI helps me return to the work, and deepen the question, and focus me on the sentence I need to write, and find the flow of that strange bonus, then perhaps it genuinely serves the writing. If, on the other hand, it gives me a finished piece of text in place of the inward/outward act of writing that would have helped me bring forth and form what is within me, then it hasn’t helped me write. It has helped me avoid writing while giving the appearance of having written.
That, I think, is the first condition for saying yes. A serious writer can use AI without losing his soul only if he remembers that writing is not mere output, and it’s certainly not external output, words produced remotely without the person’s actual, intimate, firsthand involvement.
What I’m arriving at is the position or perspective that AI may serve the writer in several ways if it’s carefully controlled and approached with a certain ironic distance, a conscious refusal to let its imitation of intelligence convince me, even unconsciously, that there’s really someone there, someone with the interiority of a writer. There are various valid roles for it: mirror, questioner, foil, interlocutor, organizer, friction-remover—though not all friction by any means, since friction is often what I need the most. AI can help reveal things that are latent in reams of my notes. It can ask a question that reignites my fire for a piece I’m flailing with. It can help test a structure, or expose a weak transition, or maybe even point out that what I thought I was writing isn’t quite the piece that’s trying to be born.
I have found this technology to be frankly fascinating, even as I’m deeply wary of it, and have been increasingly experimenting with it over the past three years in private as I’ve tried to figure out what, exactly, it is relative to me, and what it can and cannot and may and may not do for me—just as I did when, after a childhood of being obsessed with writing stories, first by hand, and then on my parents’ old manual typewriter, and then on an electric typewriter, I found my way into the new world of word processing and learned about the trap doors of spelling and grammar checkers and on-demand thesauruses. I have found AI to be genuinely helpful with certain kinds of planning, questioning, organizing, testing, and editorial reflection. I have also repeatedly come up against the danger that it might rob me of the very difficulty that makes my writing my own. As a preliminary matter, I’ve found that it “wants” to divert my attention from my inner creative relationship, luring me with a kind of siren song to just go ahead and use it as the writing genie while my daemon muse hovers behind my conscious self, silent and abandoned.
Despite this apparently inbuilt temptation, I’m beginning to think that generative AI may eventually become one more technology in the long history of writing technologies, something that definitely changes things in substantial ways, but that doesn’t have to be framed and feared as a herald of creative Armageddon. Maybe it can, in its own way, support our creativity instead of replacing it.
7. My Provisional Position
So this is where I find myself, at least for now. I don’t think serious writers need to reject AI outright. I also don’t think we can safely treat it as just another tool, as if nothing fundamental has changed. Something fundamental has changed. The machine can now imitate not only the generation of writing but some of the feeling of inspiration itself, at least if we fail to consider it carefully. That makes it powerful and dangerous.
There are, of course, other serious objections to AI that I’m not addressing here: environmental costs, energy and water use, the proliferation of data centers, copyright and labor concerns, institutional disruption, and more. I don’t dismiss any of these as trivial. Some of them may prove decisive for many people, and maybe they should. The copyright matter intersects with me personally, as my own books show up among the sources that have been used to train the prominent LLMs. But the question I’m trying to think through here is narrower and more inward than whether the whole AI enterprise is ethically, politically, or ecologically justified. I’m asking whether a serious writer, confronted with the fact of this technology, can use it practically and/or creatively without damaging the spiritual core of the work.
My thinking has also been sharpened by my day job in higher education, where AI has become impossible to ignore. As a vice chancellor of academic affairs, I’ve helped lead my college’s efforts to think through responsible AI use among faculty, staff, and students, including policy work and a broader institutional project focused on AI, teaching, learning, and student success. I’m also in regular conversation with faculty who are trying to decide what kinds of AI use should be permitted, encouraged, discouraged, or forbidden in their courses. All of that has contributed to making the issue feel immediate and practical to me.
So the “yes” with which I began is actually brimming with qualifications. As I said, this is a complex question that requires careful consideration. I’m talking here about the writer’s soul, by which I mean the living inward relationship among attention, conscience, vision, and the spiritual source of one’s creative calling. Stated as a principle, my current answer is this:
A serious writer can use AI without losing his soul, but only if it assists the work instead of becoming the source of the work.
Plain and obvious as it sounds, everything I’ve been saying here goes into that.
I hasten to add that this position is provisional and subject to change. At the end of Richard Bach’s book Illusions: The Adventures of a Reluctant Messiah, the protagonist, a fictionalized version of the author, stumbles upon a line in the “messiah’s handbook” that he had been eagerly consulting for weeks, and it turns his world upside down: “Everything in this book may be wrong.” That’s how I feel about what I’ve said above, and about everything I may go on to say about the subject in the future. Things are moving and changing fast, and what counts as a valid position is almost impossible to judge. But, for better or worse, my current outlook is what I’ve been laying out here.
8. Finding the Line
At the moment, behind the scenes—or now partly onstage, in the form of this essay—I’m beginning to think through these questions in a more sustained and practical way as part of a larger project I’m developing called The Self-Determined Writer. It will focus on humane authorship, creative freedom, ethical technology, and how to preserve voice and judgment in an age of automation. But for now, I’m less interested in announcing anything than in asking the AI question honestly.
I don’t just want to raise the question and then pontificate about it by myself. I want to hear from you, too. For those of you who write, teach, read deeply, make art, or just care about culture and creativity and therefore feel the force of the AI question, where do you draw the line? What uses of AI feel like valid assistance from a sophisticated tool, and what uses feel like abdication and outsourcing? Does AI enrich your relationship with your inner genius, your daemon muse? Does it empower you in your writing or other creative endeavors? Or does it hijack your creative spirit by offering a plausible but artificial substitute?
And maybe most simply: After using AI, do you feel more alive to your work, or less?
Warm regards,
So well written. Thank you.
If he feels fine about using AI, I wish the serious writer would proudly label it AI so that those of us who do not want to read AI-generated material can skip it without wasting our finite human energies on it.