So! How about those LLMs, Liz?
My mother was a lawyer.
This, I am realizing, has influenced my own moral compass quite a bit (although to be fair, it’s probably more who my mother was than the law itself). When Pinterest emerged, my mother took one look at the TOS and said “Absolutely stay the hell away from that.” The ease of copyright violation–even though that wasn’t the only potential use of the platform–rang all her alarm bells. Napster for pretty pictures.
Of course Pinterest has endured, and companies and artists and all sorts of people have embraced it as a way to share artwork and products. It’s secondary advertising, the way a lot of social media is. And there’s still massive copyright infringement there–just not of the sort anybody wants to prosecute, because money.
I don’t have a Pinterest account. I’m aware most Pinterest users use it for lovely things–mood boards, event planning, artistic references. In one of those inexplicable lines I draw, I don’t judge people who use Pinterest. I just never stopped taking my mom’s advice.
And you know, kids, I kinda think Pinterest is part of what’s made us absorb the massive copyright violation of LLMs. Different scale, different purpose, yeah–but we all embraced Pinterest because “oooh, pretty,” and now genAI is straight-up stealing people’s livelihoods and the courts are largely shrugging.
If you don’t defend your copyright, you will lose it, and we have. All of us.
But I’m not here to rant about the copyright bullshit. That ship has sailed. Do I hate that? Oh, I do. Ask anyone who knows me in real life. I’ve got a pretty good sense of humor, but there are a few topics that send me right to DEFCON 1. Which is a problem, because it makes productive conversation difficult, and there are actually people out there who are persuadable about these things.
I’ll confess, right here, up front: if you’re using genAI for something that is not required by your livelihood, I am going to judge you. Yes, I know there are a handful of genuinely decent applications for the stuff; if you’re using it for that, I will probably not slam the door in your face, but I’ll be keeping an eye on you. The rest of you? I have Thoughts about the environment-destroying plagiarism bot. It’s not going to be Skynet, it’s not going to cure cancer1, it’s not going to become sentient on any level. It’s not on any large scale going to make a damn thing better in this world, and on a small scale there are better solutions than something that eats both water and brains. The bubble cannot burst fast enough, and I hope the financial fallout is bad enough that a few of those dumbass CEOs get burned, even though millions of innocent people with mutual funds are going to suffer the most.
Where the hell was I going with this? Oh, yeah: LLMs and writing.
This morning I caught a post on Bluesky about a writer who was apparently wrote an article extolling the virtues of LLMs. (I did not read it.) The social media backlash has been pretty swift, but as I’ve learned in uncountable election cycles, what seems like consensus on social media is often very different than what the general public thinks. This writer is hardly the first to come out of the LLM closet–but today, probably because I’m already cranky as hell because I’m having surgery tomorrow, I’ve started thinking about why I do not, and will not, ever2 use LLMs for writing.
If I shove all judgement out of my head (yes, impossible, but I do try), it comes down to this: I cannot imagine any way, on any level, that an LLM could help me.
One thing that fascinates me about writing–and one reason I’d be a terrible, terrible teacher of it–is that each of us has a different process. Story generation is an alchemic thing, and the nature of the writer’s brain is the biggest differentiator in how the story is born and grows and is shaped.
(It’s also the reason one of the hardest skills for writers to learn is how to finish a thing, but I digress.)
I’ve been writing stories all my life, although I didn’t start writing them down until I was old enough to hold a pencil. (My dad saved a picture book I wrote when I was about 8. No, I won’t be sharing it. Yes, I think it’s adorable that he saved it.) Compositing fiction in my head has been a part of how I interact with the world for as long as I can remember. It started out as a coping mechanism3, and it still is.
For me, the alchemy usually starts with a mood, which coalesces into a scene. From there characters appear, and then events; if the scene is still interesting to me, my brain keeps spinning on it, and eventually something resembling a plot is born. At that point, it’s all refinement, which can take years.
Do I get why people want a shortcut, a way to take their amorphous thoughts and make them consumer-ready in less time than their mercurial muse does? Of course I do.
But for me, inserting an LLM into this process would break the process. It’d destroy the story. The story would no longer be contained in my head, and I would no longer have the ability to turn it into something that said what I wanted it to say.
It would be somebody else’s work. And because I have the attention span of a gnat, I’d lose interest, set it aside, and probably salt the earth it came from.
Much has been written about those very weird tech CEOs who say things like how nice it would be to be able to outsource their imaginations to a machine, along with most of the human interactions the rest of us find critical to a happy life. They seem addicted to the idea of some external entity offering them up Fulfillment McNuggets without them having to actually feel anything, or participate in society in even the most tangential way. It’s a deeply bizarre vision of existence, and they don’t even seem to recognize how creepy and pitiable they sound.
These are the same people who try to tell me how nice it will be for me when I don’t have to come up with ideas anymore, and what the fuck, guys, get therapy.
Like most writers, I am not short of ideas. Do I find, sometimes, I’ve written myself into an insurmountable corner? For sure.
But that’s the fun part.
All those unfinished stories, all those blind plot alleys, all those characters who just didn’t work in the end–without those foundations, I’d never be able to build the thing that works. It’s all part and parcel of that same thing, that seed that was born with my moody little scene. I don’t know what it’s going to grow into! Not everything I feed it is going to be visible in the end! But because I let it grow and change and shift and succeed and fail at its leisure in my head, I’m able to mold it into what I want it to be.
Even if you removed every moral, legal, and existential objection to generative AI, I would not be using it for my storytelling. I don’t avoid it because I’m some self-righteous jackass4. I avoid it because it’s the wrong tool for the job.
I do get that it’s a temptation for some writers. If you publish, or want to be published, it’s impossible to miss the push toward speed. And most writers have day jobs, which can make it hard (even for those who write faster than I do, which would be nearly all of them) to produce at the level you’d like. I do get the “it’s just massaging the ideas I’ve given it with the prompt” spin5.
But for me, with my process, with my unique, fumbling, Liz-shaped brain, I cannot imagine a single situation where it would do anything but throw me off and slow me down.
I honestly find the current moment a very strange one. I was always a big tech cheerleader. I was the early adopter. I couldn’t wait for the latest OS release, so I could upgrade my machine and play with all the new features. This has changed relatively recently, and I think that’s because tech companies are no longer solving consumer problems. At this point tech companies are focusing on ways to squeeze more money from customers, and to eliminate their human employees entirely6. I had occasion a few days back to load up some of the mobile games I used to play 10+ years ago, and damn, not only are they still awesome, I could still play them–full-featured–when we lost our internet for half an hour this morning. Tech used to be fun.
GenAI is human-hostile, and it is not solving any kind of large-scale problem. I suppose I should be glad Anthropic etc. ripped me off, because I leapt right over grief and into rage.
I might be wrong. The damn things might indeed get so entrenched into our lives that avoiding their use becomes impossible. But hey, I’ve got an old laptop, and some old word processors. I can type in fucking vi if I have to. I can even (gasp!) go back to pen and paper. And I will, before I ever put my name on words that are not 100% mine.
The main reason I have hope, actually, is because for most uses7 genAI tools are just not applicable. They don’t do most of what they’re sold to do. Pay attention to the tense when you hear these genAI evangelists talk–if it’s not something it’s demonstrably doing today, it’s at least even money they’re talking about something that’s architecturally impossible for it ever to do. And the financials behind these companies are absolutely hair-raising. Everybody’s living off loans, and the giant revenue streams are not appearing. Every time you hear some outlandish story in the press–AI will fix climate change, AI is angry with us and is going to destroy us, etc.–realize that there’s a closing window for these companies to get people to pay them enough to cover the tool’s usage, and they’re desperate for you to believe it can do far more than it does.
Most of AI hype is grift. Possibly hopeful grift–some of these Big Brains really seem to believe their own nonsense–but still. Most of the future-speak is from people well aware they’re not even close to getting there, and in fact are likely sitting on a tech dead-end.
Linguistically, of course, LLMs are fascinating, although not for the language they can produce. They’re fascinating because they expose our own mushy, fallible human brains, and how powerful language is as a tool. If something talks to us and sounds plausibly human, we seem to be wired to erroneously attribute all kinds of human characteristics to it. And of course we do–talking or not, we anthropomorphize nearly everything, from toys to computers to cars. And LLMs allow all of these objects to sound like real people. It’s a pretty neat trick, natural language generation–but the real crux of it is our human brains want so much to believe there’s a mind behind that language we’re willing to toss all common sense out the window.
For me LLMs are the antithesis of actual writing. They imitate. I might be able to phrase a prompt in a way that makes the tools generate something that’s imitative of what my imagination does. Why would I do that? What does it get me? For some audiences, it can be a neat party trick, and if they’d left it at that we’d probably all be cheerfully buying talking iDogs and moving on with our lives. Instead they’ve decided the answer is to put us all out of work, and have machines create at them so they don’t have to think, or move, or love ever again.
Yeah, I’m back to “why does anyone want this?”
I am a slow writer. I will probably always be a slow writer. Part of the joy, for me, is the percolation of the whole thing, the simmering of ideas, watching them morph and change and deepen, stirring in little bits now and then until it’s exactly what I want. I don’t need another cook in the mix. Dinner may be a ways off, kids, but I promise you: it’s 100% homemade.
Footnotes
GenAI is in the same kingdom as machine learning, in the same way that myxozoans and humans are both animals. Machine learning has been around for many, many decades. It predates me, and I am famously an Old. Machine learning has indeed accomplished some amazing things in its lifetime. GenAI is a dinky little toy that sometimes gets plopped on top of a machine learning core (or, more often, your basic human-written software tools). The term “AI” is not regulated and not defined in any way that’s useful to consumers.
↩︎
↩︎
↩︎
↩︎
↩︎
↩︎
↩︎









