OK, this expresses a lot of my thoughts on the use of gen AI for writing (or any creative endeavor) in general and fanfic in particular, but I have several additional thoughts after stewing on this all day yesterday.
So something that's been driving me crazy is how I haven't really seen anyone addressing the deliberate and ongoing lying by some of the authors exposed by the John Doe PDF. I'm talking about the prominent anti-AI statements on some of these authors' fics. Like, not just a one-sentence "I don't fuck with robots lol" disclaimer. I'm talking an entire-ass paragraph going on on and about how they don't support AI, they don't give permission for their fic to be used to train AI, AI bad, etc.
And then there are all the people who keep blaming their betas for using AI to edit their works. It's not just fic authors. Mia Ballard has claimed this is why Shy Girl reads the way it does. (To which I say: baby girl. Honeybee. You know you can reject suggested changes, right? Like, you're the final arbiter of your work. You don't gotta blindly approve every dogshit suggestion! You're in control! But Ballard didn't do this because...drumroll...she's clearly lying! If you check out Shy Girl, it's pretty clearly AI-generated, not AI-edited, with a few chunks of human-written prose here and there for spice.) I've seen at least one fic writer claim that their beta inserted the Claude-generated prose into their fic which—Jeezey Pete why are you giving your beta your AO3 login and allowing them to make changes to your actual posted work in AO3 itself. That is. What. How do you.
This type of behavior, to me, rises above the level of "They're making me use AI in school so why not use it for fic," or "I felt pressured to update my fics on a set schedule and used AI to help me stay on target," or even "I don't give a shit lol I like getting kudos and comments." Like, it's one thing to use AI and not tag it or own up to it in the notes. It's a whole other thing to use AI but then deliberately and pre-emptively construct something to help deflect suspicion, and then to keep doubling down on the lies.
Listen, we want to give people the benefit of the doubt. We want to believe people, especially when they're friends or someone we admire. It feels bad to harbor this ugly little seed of doubt; to feel like we need proof for every little claim. My default genuinely is to believe people when they say something to me. But we also need to think through things critically sometimes and go huhhh none of this adds up. To go: yo dawg I think you're lying.
Like, the lying doesn't need to stem from some nefarious or malicious scheme. I've lied! Most everybody has lied! We usually lie because we're scared or ashamed. We want to protect something, usually ourselves. And yes, people lie to get ahead, too. None of these reasons matter, because none of them change the fact that on top of using a lying plagiarism machine, some of the writers uncovered by the John Doe PDF have tried to cover up their AI use from the beginning, and continue to construct wild new lies about how the Claude formatting ended up in their fics. I can't see into these people's hearts and divine oh they're lying because they're scared. What I can see is that they. Are. Goddamn. Lying.
Y'all. I find that enraging.
I also don't know what to do about that rage. Mostly I want these people to have not fuckin' used AI in the first place. The stolen valor issue already hits me right in my overdeveloped sense of justice; the continued lying is salt in the wound.
And then, of course, people are already making excuses for the most popular writers exposed by the John Doe PDF. Oh, it was only a couple of instances of Claude. Oh, their other works don't contain the inserted code, which means they didn't use it to write my favorite fic of theirs. Oh, they probably only used Claude to edit their text which is a totally different thing from using Claude to generate the text.
Listen. You don't have to care that your fave uses AI. But stop moving the freaking goalposts when they're exposed as having used AI. They've used AI, and they lied about using AI; what makes you think they were being honest about not using AI for all the fics that contain exactly the same disclaimer as the fic that did use AI?
I have an admission. One of my favorite writers in willmack fandom has, IMO, very clearly used AI in at least two of their works. One early work appears to be mostly AI-generated and was so strange and unreadable and unlike anything else they'd put out that the difference leaped out at me. The other work, published later, comes across as human-written with occasional chunks of AI-generated prose. I read this later work first, and I loved it! I didn't clock it despite a few passages feeling off to me until I went through the author's backlist and found the early work that was mostly if not all AI-written. That's when I went, wait oh shit hold up, and then re-read everything they'd written with a much more critical eye.
The thing is, some of this author's fics are still my faves. I wrestled with myself, and came down on: I find the stories they're coming up with compelling, so I'm still subscribed to them, and I'm still reading what they put out.
I fucking hate AI and I especially hate how it's infiltrated fandom, but the truth is, we need to stop characterizing AI writing as "bad" or "writing I don't/can't enjoy because it's so obviously AI." AI writing is enjoyable. Manifestly it is! Just look at the kudos and comment counts on all these authors who've been proven to use Claude! And it's also clearly hard to detect, at least for most people, because a lottt of people have been banging the drum about several of the fics in the John Doe PDF as having been AI-generated, but the pushback has been intense because people refuse to believe their fave is a cheat.
(As for AI-generated prose = bad writing, I have...a lot of thoughts on what constitutes bad writing that I won't go into here because this is already way too long lol.)
Anyway. In conclusion, this willmack author's newest works hit my ear as being mostly or entirely human-written. But I can't rule out the possibility that what I'm seeing as human-written prose is actually them paying a monthly fee for the upgraded Claude model, which from what I've heard is much, much more sophisticated and much better at hiding the AI tells. Like, maybe they have the upgraded model for work, and they don't see why they shouldn't use it to make hockey boys kiss.
So here's another thing about AI, and this one goes to the point about the politics of AI that tobermoriansass made in the post linked to in the ask—"i do not believe that anti-ai politics is a real or meaningful politics," to quote directly.
I do! And it goes beyond like, oh it's based on stolen labor (that's just capitalism, baybee!) or oh, it's bad for the environment (the truth about that is complicated, TBH). It's not even, wow many of the things we do that are destructive to the climate and to society like oil drilling are also needed for society to function, but AI is NOT an essential technology. Like. Magically removing gen AI from this world would, afaict, bankrupt a bunch of venture capitalists and tech bros (yay!) and force companies to re-hire people to do the job they've offloaded on AI agents (double yay!).
No, my qualms are related to the longer timeline of AI development and its implications for human labor.
So I had lunch recently with a friend who works with gen AI. She's a huge proponent, and she wanted to know why I'm so reflexively and doggedly anti-AI. (Except, apparently, when I like a fic author's work, lolsob.) We talked for over two hours, and she gave me a lot of context and information about the capabilities of AI, especially the cutting-edge paid models that she uses for work, that for real made me feel sick.
Basically, she said that the current cutting-edge models are better at doing rote administrative work than a human, and at more complex intellectual tasks, operating at something like the 90th percentile. Models already exist that generate prose indistinguishable from a human's, you just need to pay for them. She estimates that in three to five years, AI models will exist that will make about 90% of knowledge workers' jobs obsolete.
She thinks this will usher in some kind of utopian post-scarcity future where we can offload all the tedious work on robots and focus on pursuing our passions. I think we're going to get something a lot closer to 14th century Russia.
The truth is going to land somewhere between her optimism and my deep-seated pessimism. And who knows, the AI bubble might pop before the models get that good. My friend is someone who's been a highly sought-after, highly-skilled programmer for decades. She's smart, she's knowledgeable, and she works with AI models all day every day. She's using it to build a really complex interactive story engine. I think she's shortsighted because she's accelerating the pace at which she's making her human labor obsolete. She thinks I'm burying my head in the sand for refusing to engage with AI.
So yeah, my anti-AI politics are fucking meaningful. If my friend is right, and if you care about not being consumed by the (in this case literal) machine, you should in fact care about AI usage. It's not just a benign thing fanfic writers are using to gain clout/kudos/connection to a community. It has the potential to do some terrifying things. It is in fact already doing some terrifying things. It just so happens that the usage in fandom is the one domain I have an even remote chance of influencing.
Do I have a solution for any of this mess? Lol nope. On the one hand, I do think the John Doe PDF is going to result in an uptick in harassment, and I genuinely believe that nobody deserves to be threatened or harassed over this. On the other hand, I deeply appreciate the hard proof it provided and. God. Maybe it's the old-school Chinese parenting I was raised with, but sometimes I'm just like, y'all, ain't nothing wrong with some good old-fashioned naming and shaming. Some people need to be shamed!
There are two wolves inside me, etc.
I don't think a perfect solution exists. Not talking about it and going "don't like, don't read," however, strikes me as possibly the worst. If I had to guess on an outcome to this, however, I think it's going to work out pretty much exactly the same as every big plagiarism scandal, which is to say: we're going to be super worked up over it now, and there's going to be a lot of finger-pointing and a lot of HOW DARE YE, SIRRAH!
Then time will pass. Some of the authors will stop writing. Others will continue writing ("""writing"""). We'll look back at this a year from now and go, huh, remember the giant kerfuffle over Claude usage in AO3 fics? Anyway,
