AI in Fanfiction

This is a feature excerpt from Unwinnable Monthly #186. If you like what you see, grab the magazine for less than ten dollars, or subscribe and get all future magazines for half price.

———

Can AI write fanfiction?

No.

But that isn’t a satisfying answer, and this is an essay. An opinion article. It’s something. “Can AI write fanfiction?” feels like a question where the answer is, “yes.” I mean, it can write an essay or a sonnet or a recipe for something that may be napalm or could be mayonnaise. The only way to tell is to put it in your mouth.

But AI isn’t writing. AI is cranking up a plagiarism machine and having it plop text, parsed through an algorithm, into your waiting hands. It’s Silicon Valley’s hatred and misunderstanding of the artist, manifest into a program to take the joy out of creation and replace it with a handful of key presses.

It’s not writing. It’s one of those secret third things, but the secret is that it sucks. And that’s not a secret at all.

I hate that in talking about AI you have to quantify a bunch of terms. It’s like any sort of debate semantics with the worst dudes in undergrad. When I say AI in this essay, I am talking about LLM’s – large language models that have blown up in the last year or so. ChatGPT is probably the most popular of these, but all of the worst tech people have their own flavor. The real kicker is that the Common Crawl (the big data scrape that is the foundation of the house that is OpenAI, the poison tree from which all fruit is found) scraped Archive of Our Own along with the rest of the internet.

The Common Crawl is theft. And one of the things it stole was my work.

ChatGPT is built on AO3. On millions of words, by writers from around the world. It has the DNA of every publicly accessible Stucky work, millions of words of Sebastian Vettel’s gay romance, thousands of the worst My Hero Academia fics every written, works by authors dead where the only sign of their existence is a digital tombstone in the form of a Sarek/Amanda Grayson fic they wrote in the ’80s moved to AO3 by a diligent archivist. Millions of words, gobbled up unthinking by a machine, to be shit out later by someone too lazy to write an essay. Removed of context and intent, mashed and prodded until they were unrecognizable.

The first line of defense on this subject probably should have been AO3. While not perfect, the Archive has a long history of supporting fan artists and their works. They couldn’t have protected their users from the Common Crawl – no one knew it was really happening, and there are legal cases now trying to figure out just what was stolen, the extent of the material both copyrighted and not – but surely once the information came out they would take a hard-line stance. Their user’s work was stolen. It’s an easy stance to take.

They didn’t.

“Our goals as an organization include maximum inclusivity of fanworks. This means not only the best fanworks, or the most popular fanworks, but all the fanworks that we can preserve. If fans are using AI to generate fanworks, then our current position is that this is also a type of work that is within our mandate to preserve.” (Emphasis theirs). While their FAQ gives advice that if you don’t want companies to steal your work through data crawl you should set your privacy to Archive-only, they don’t currently have an anti-AI policy. In fact, you can find many Reddit threads on the subject of “how do I report clearly AI work?” You can’t.

There are active users on AO3 now who are using ChatGPT or Grok or Phleghm or whatever other service people use to “create fanwork.” They are regurgitating the reconstituted words of people who actually wrote their work. It’s plagiarism with extra steps. The Archive has no issue with it.

Betsy Rosenblatt, the legal chair for OTW (the organization that runs AO3) is incredibly pro-AI. When asked about the scraping that had, at that point already happened, Betsy had a lot to say. All of it has that kind of skin-crawling optimism of someone who doesn’t understand what it feels like to realize that your work has been harvested so that OpenAI can burn down a rainforest and so that Duolingo can fire their translators. Someone who believes that their job, their super special job, will always be safe.

I was going to cut her comments down for brevity, but I think she deserves to be shown in full context:

“One of the things that excites me – which is probably a bit off to the side of what most people are talking about with AI and copyright – is that AIs are reading fan fiction now. For a long time, machine learning relied almost exclusively on data sources that were known to be in the copyright public domain, such as works published prior to 1927 and public records. The result of that was that machines were often learning archaic ideas – learning to associate certain professions with certain races and genders, for example. Now, machine learning is turning to broader sources from across the internet, including fan works. That means that machines will learn how to describe and express a much more contemporary, broad, inclusive, and diverse set of ideas.

I’m also intrigued by some of the expressive possibilities that AI may create. Will DALL·E or ChatGPT become characters in fan fiction? Surely they will. I want to read the fan-created stories where DALL·E and ChatGPT fall in love with each other (or don’t), get into arguments (or don’t), buy a house together (or don’t), team up to solve (or perpetrate!) crimes….”

She goes on to say that she hopes that no one pursues litigation to stop this from happening. To push back on data scrapers and crawls and a few companies run by incredibly wealthy men scraping the work of the people she is supposed to be helping to legally protect, for their profit machines.

For diversity, she says.

Which brings us back to the work itself. I have never willingly read an AI “fanwork” because I do not read things that no one has written. If no one has written it, then there’s no reason for me to engage with it. It doesn’t matter to me if it’s “good.” That’s honestly secondary. An AI could write a fic that could make me cry or scream or re-evaluate my relationship with my mother, and I will never know. Because what is making me react like that is the reprocessed ghost of someone else’s artistic voice, broken down into a million bits.

———

Amanda Hudgins is an occasional writer, former rugby player and wearer of incredibly tall shoes.