Hi, @idreamtofmanderleyagain, I’m your anon (linking here for context). I approached you that way initially bc I didn't think that we had interacted enough times to have me approaching you about this not seem confrontational, but since you’ve been kind enough to put so much serious thought into your answer, I think it would only be fair to respond at all with my face attached. I tried to reblog this in direct response to your own post, but between the two of us, we seem to have hit tumblr’s post length limit. go figure.
Ahead of my response, I have to explain part of my perspective outside just being a fic writer: I’m a grad student in a literature department, still work there currently, and have taught undergrad english during my program, so this issue has already had a massive impact on my field. My own experiences have given me reasons to be wary of LLMs in particular, so that does influence my views to some degree. But your original post was mostly about fandom, so I won’t expand on those unless you’d want me to, for whatever reason.
keeping it to just fandom stuff, even with your research, I hate to say none of my previously voiced concerns are assuaged.
AI users have, knowingly or not, justified the initial theft by corporations to train their products that are potentially going to be made into for-profit models as something unfortunate but ultimately worth overlooking in the long run, and already, I know we’re going to disagree at a fundamental level. You said it yourself above, where you acknowledge this is a factor, but this doesn't give you enough pause to question your own use of these models.
You could stop reading now to save your own time and I would understand, but I still wish to respond to these other points.
While I appreciate your apparent genuine effort to interrogate these systems to the degree of their training and use of this data... you'll have to forgive me if I remain skeptical. I'm not surprised it told you what it told you, and as you yourself pointed out, it could easily be spitting out boiler plate answers programmed by the people in whose interest it would be answering. There is no promise of authenticity, as you noted, and I wouldn't be surprised if the program that generates prose based on user prompts is presenting to you what it thinks you'd want to read. Thanks nonetheless.
The arguments presented here seem to invoke the disabled community multiple times in ways that I’ve seen the larger community disagree with, and while we’re not? a monolith? I think this is something I want to address on from my corner of it. Let’s start with the Amazon/Bezos comparison. You could argue I should only shop locally, sure. But I’m sure you’re aware lots of people (myself included) use Amazon bc for things we absolutely need of limited access to certain supplies in our communities, etc, and that’s not even uncommon among non-disabled people. But the fact remains that LLMs are not like Amazon in the sense they’re a super conglomeration that has made itself almost unavoidable, especially to people who live in more isolated areas. I don’t think your exercises in a character chat bar are exactly comparable to that, primarily because using AI in your creative work is not a necessity; people are only doing so out of curiosity.
That, I think, is primary issue actual artists and writers find uncomfortable: people who don’t have to use this technology that was, again, built to some degree on the exploitation of people who have been developing their skills for whole careers, are taking umbrage against the fact that those same creatives are upset with those that don't think this exploitation is worth curtailing or interrogating their personal use of AI. This is an entirely optional part of your writing. Metaphorically, I would argue that this is not Amazon, this is closer to eating Chik-fil-a - you know something sucks for a particular community, but you still have the option to patronize it yourself if you really wanted to.
If that sounds unfair to you, sure, you can assert your right to use the LLMs. But the rest of us who didn’t consent to having our work scraped, especially in the era of a writer’s strike where AI is specifically an issue, maybe won’t be as receptive to the fact that this is something you’re willing to disregard to some degree for the sake of whatever benefits you see to AI in your work. You might not frame it this way, but this is still what it comes down to. Your experience with this new technology is ultimately outweighing voiced concerns that people, as you pointed out yourself, are going to see their livelihoods impacted. I don’t mean to be blunt, but that is genuinely what I believe is being weighed in this issue.
I also disagree as someone who is neuroatypical that the idea of “hard work” the artists and writers are invoking in their arguments against AI is ableist? Because I know exactly what they’re referring to; I have spent literal years working on my craft. I’ve dedicated a great deal of my education to it. I know you didn’t just start writing yesterday, either, and that this is a process that you find fulfilling in some not-minor sense, hence perhaps your initial desire for AI experimentation. While you seem to read the creatives against AI as invoking an ableist standard of “hard work” as in working towards literal health problems (the recent ComicsBrokeMe discussion on various platforms is timely), no one who’s trying to make a serious living out of their craft in the current economic climate is going to say that what they do isn’t hard work, or that they haven’t put hard work in to work at the level they do. Saying that the idea of hard work is in itself ableist downplays the effort many creatives, disabled and not, actually conscientiously undertake in their creative endeavors. I know we both know disabled creatives can do hard work. AI users trying to frame this idea as ableist are well-meaning, but this argument kind of shoots itself in the foot if you believe that talent in the humanities is learned, and not somehow inborn. People who try to frame AI as something to level the playing field are more or less saying on some level that the efforts to learn art or writing either don’t work, or aren’t worth the time needed to practice.
As said previously, disabled people are not a monolith, and disabled people using AI doesn’t mean other people, also disabled or not, have to find it acceptable. As you’re well aware, there are lots of disabled artists and writers who have established distinguished careers for themselves well before LLM and visual models were ever even a concept. There are many ways we’ve found to participate in these disciplines, which makes me deeply skeptical of anyone trying to claim this is the only way they’re able to create something, especially if they use their disability as a shield to excuse it. Again, this is not a necessity in any artistic endeavor, as there are plenty of us who are disabled who manage to create regularly.
As to AO3, I’ll be honest, I am one of the people calling for AI exclusion, or at least open labeling of such products. The process of generating a fic or even just a summary with an LLM is completely different from actually writing a fic yourself and the level of intention involved, as I’m sure you’re aware. The AI isn’t making creative decisions as much as it is filling out a written parameter based on previous examples it’s been given, no? It's not even something I'd really consider "brainstorming" -- It’s just a series of words the algo has arranged in what it believes to be the “correct” order. The operator can refine their prompt, but we both know that doesn’t engage the same process of planning and executing a story yourself does - even, I’d argue, down to a drabble.
You can argue this differentiation would create a culture of secrecy, but that secrecy only flourishes if people using AI purposefully try to conceal doing so. If AI is something you're going to continue to peruse in your own work, I don't see why other users wouldn't adopt with open identification, both so y'all can find each other and so those of us who would rather not engage with AI can filter appropriately. I don't think generations should be engaged with in same way, because they are in themselves the product of a different kind of effort, which you can argue is fannish/fen, but I will gently disagree.
And, if the results of the generated fics are just as vague, confused, and overly general as you describe in your own personal experience, I see that as a further argument as to why we shouldn’t want them there. If the AIs have stopped training, as you’ve read, they’re certainly not going to get any better. And if the AI is still learning, then that only underscores the original problem in the first place. Even if it’s not scraping AO3 or other sources anymore, people like yourself who are feeding their own fics in will help teach it, which you may not see as a problem. AO3 currently agrees with you, so if you want to post the products of your research, you’re certainly able to. If AO3 doesn’t want to ban AI fics, I don’t think the rest of us are in the wrong to either ask for them to be taggable, or to be filed as a completely separate option from fan fic or fan art. You can call them Fan Generations, even, idk but I do think there should be some delineation between the process that went into these and user-written fiction.
AO3 does have a public source code, IIRC, so if people like yourself wanted to gather and set up a dedicated archive of machine-generated work, I imagine you’d find some takers for volunteering and fundraising - not as many at first as the original, sure, but if this is something that a side of fandom wants it then has potential to grow.
On top of that, people continuing to feed/refine the models will only remove that vagueness/confusion you cite to justify why it’s harmless. We’ve already seen this be true of imaging AI being distinctive from human-created art with the numerous human fingers gradually whittling down, right? If that’s the case, then I don’t see why people who generate their fics wouldn’t want to have them tagged as AI, if it really makes no difference to them.
All of the above, plus citing some artists being incorrect about how AI works or being too mean as an argument towards why people should be able to use AI without pushback from the creatives they stand to impact, has made me think that your problem is less with community fracturing/safety and more an issue with the tone other people take when discussing this. This is where we come back to our initial disagreement over your examples w/r/t discussions of trans issues and vegan issues - because I do feel the people angry about the way they’re being treated unfairly/exploited are justified in expressing anger at the people who do so, but also those willing to ignore this treatment.
You can argue that this is people turning their anger with the AI threatening their livelihoods at the people using the AI, but then we come back to something similar to the “guns don’t/people do” argument that we’ve both doubtless heard a lot in the past few decades. Sure, you could say AI itself isn’t threatening our fields, people are - and maybe not even always the people directly using the AI, but the suits that will hire them for cheap, or the person who doesn’t see the value in human intention and artistry in the pop culture they want to consume. But if people continue to work with these models and they continue to learn, it only increases an ongoing social devaluation of our disciplines, and therefore decreases the value of our labor/effort by the people who don’t see the difference.
Again, as you seem fine to further experiment with AI yourself and feed it your own work, you might think that this is just a necessary trade to be made in the name of optimized individual experience with these systems. I don’t know.
So, ultimately - I centrally disagree with the initial framing of this as fans attacking other fans, or even an issue of safety or witch hunts. This is people who do creative work, professionally or for fun, having a problem with it being infringed upon by corporate entities in the name of training a machine, and then having the use of that exploitation justified by people who are seeking to ultimately take advantage of it. And not even due to the inescapable nature of doing so, but out of what ultimately boils down to their personal desire to use AI in their own work.
This lands especially heavy in a time where lots of creative people are already trying to
convince the powers that be that their cultivated knowledge and skills are worth respect and proper restitution,
convince the people that consume the work that they aren’t entitled to a creative’s labor for free just because it’s Art,
deal with this weird resentment from people who want to use AI even though many have explained how they’re being hurt by them, and seeing arguments that we’re somehow censoring the creativity of these people by doing so.
It’s a shitty hydra, and we don’t appreciate the people who are contributing to any of its heads.
If you want to call the initial theft a “technicality” and claim it shouldn’t be controversial to continue your use of AI, my only suggestion is to not ask creatives impacted by this to forfeit their right to be angry at this use. That doesn’t excuse targeted abuse, of course; nothing does, end of. But I also don’t think that artists, writers, and other people who have been exploited being generally frustrated and even angry at AI, and the people who perpetuate that exploitation by creating the demand for AI is on the same level as someone being shitty to you as a person or anyone else. There’s not a witch hunt. It’s just going to have to be recognized as a pushback to this ongoing adoption of AI that its makers, funders, and users are pulling for.
You frame this as innocent people getting hurt, but innocent people already have been hurt by these systems, and have detailed how, and that doesn’t seem to warrant as much concern by the proponents now calling for amnesty when that they’ve got something new to tinker with.
You’ve already said you won’t be making any further posts about it, so don’t feel obligated to respond to this. I want to respect your space, and I’m also not really looking to argue - I didn’t intend to write this much about it, as it’s something I find disheartening and frustrating all around.
I don't think I'm actually going to convince you of anything, having read your own justifications, etc. I just wanted to acknowledge the work you put into this post with my own thoughts. Thanks for the willingness for dialogue, you have been someone whose takes/analysis I enjoyed previously.
