Click, Clack, Beep: Bots That Type
Promises and products of AI
The Space Between
Imagine an abacus. There are two beads on this particular abacus. With your mind-fingers, take hold of each of the beads. Now move them to the left. Now, the right. Back; forth; again. Make sure to maintain the exact same distance between the two beads throughout. That’s crucial.
And that, roughly, is where we are with AI right now. The promises of AI grow ever more messianic, even though we continually fall short of reaching them at last. The reality of AI, though, keeps getting nudged on up, enabling actual people to do actual things might actually be useful. And all the while, the distance between the two—the promise and the reality—remains equal, with seemingly nothing able to bridge the gap.
I am not an expert in AI. I am, however, personally acquainted with the promises of AI, and how fervently people can believe in them. Last fall, my wife lost her higher education job, where she worked in marketing. She was fired in no small part because her superiors believed that her job could better be done with AI programs. Which means that, since then, our household has been getting by on the (lesser) income I earn from working retail, along with whatever gigs I can land. Speaking of which . . .
So you’ll forgive me if I grow wrathful at the spectacle of writers on Hamish McKenzie’s internet saying that they use AI to write their work and they like it that way, thank you very much. Like, okay. You use LLM to write your fantasy fiction, fine. Just don’t expect me to read it.
Still, I must give that article the tiniest morsel of credit. That author is actually making use of LLMs. He is not simply touting their ability to sail you to Mars and cook dinner. He is doing something—or other, having something done for him. Something is getting done, somewhere in there.
Yet that very modesty got me thinking. Thinking about those beads. The distance between promise and reality.
For authors, the promise of AI—or the fear of it, which is simply a matter of perspective—is totalizing and all-encompassing. Input a prompt, press a button, and boom! Instant novel! Take me away, Calgon! The publishing industry as it currently exists will be razed to the ground by this time next month, thanks to the power of Cluade and Opus and whichever other models get reiterated. The whole problem with publishing, you see, is that they’re gatekeepers who won’t publish actual good books. Take that power away and place it in your hands.
Is that really the problem with the publishing, though? That they refuse publish good books in favor of the same ole same ole? It is tempting to think so—and I know I've given into that temptation myself—but it is, I’ve come to think, flat-out wrong. In multiple ways. Wrong as far as the solution it offers, yes. But in an even more fundamental way, wrong in the problem it identifies.
Problems, Problems
Aspiring authors look at the publishing industry and offer a simple explanation to its woes. “Publish good books! You know, the ones you refuse to publish! More specifically, the ones we write!” And like, way to believe in yourself. Keep at it for real. I loved Marty Supreme too.
But even though I am also an aspiring author myself, writing draft after from the depths of my basement compound, I have learned enough to know that the problem facing the publishing industry of today is not a dearth of good books. If anything, there is a surplus of good books. Good books that get published, as well as good books that don’t, due to a variety of factors from trend season to whether or not an editor got food poisoning at one of those famous lunches and thus wound up in a bad mood. There is not a lack of good books. Not even close. Every minute of every day, the inboxes of editors and agents ping with submissions from bold new voices and far-flung talents. The books are there.
I guarantee you, there is not a single editor sitting in her office and saying, “You know what? I need MORE submissions! MORE!”
The problem is not finding books. The problem is finding those readers for those books.
Identifying potential readers for books, and delivering information about those books to them. That is the real problem facing publishing, the $64,000 question. And when you put it that way, one can’t help but think: would AI be useful in that regard?
The ongoing story with AI is that its prophets herald the dawning of a new age, and then that fails to come to pass. During that same time frame, though, actual uses for AI get discovered, which causes work to get done. (And workers to get fired, lolsob.) What if that same dynamic might play out in publishing? What if the actual application is not a giant button that one presses to write an instant bestseller that mashes up A Song of Ice and Fire with the latest Ina Garten cookbook, but rather some set of tools that connects readers to books?
Or, perhaps more significantly, connects readers to authors?
Face to Face on the Page
AI can write a book. Sure, maybe. But can AI sign your book at a reading? Can AI Zoom to your book club meeting? Can AI call you a wan little husk?1 ghg


There are plenty of writers who will hate to hear this, and I include myself in that category, but still: the proliferation of AI means the connection that human authors form with human readers will become that much more meaningful.
That will look different for different writers. Not all of us can be Instagram girlies posting unboxing videos in our immaculate apartments; some of the less photogenic among us may still need to resort sloppy handheld videos on Substack Live. It is simply unrealistic to think that a writer can pull a full Pynchon and eschew promotion entirely.
But maybe that kind of offline connection can become a part of the larger whole. A piece of the project of the book in question, carried out away from the prying eyes of AI. A place that looks and more compelling.
Actually, you know what, it probably can.

This is hitting home today. A 20-something member of my team sent an email clearly written with AI, listing all the criteria her next assignment should meet and asking for a response documenting how it meets those criteria. I am tempted to respond with AI but where does that end? Instead, I will verbally suggest she have an actual conversation with her supervisor about expectations. But what would you do?