#why are so many fanfics written by chatgpt these days

[texlgbt]

I'm gonna be honest, guys. If I see the words; 'a stark contrast to' written within another fanfic I might actually weep.

[swanimagines]

Sorry for the Ray fics being late, even though I kinda promised to write them as they come in. On the night between Thursday and Friday, I was accused of using generative AI with Into the Nothingness, because according to them it's not in character at all + apparently there's multiple sentences that are commonly used by ChatGPT + the title sounds like it's been generated with it.

Some of you may remember that I have been accused of using it before too, previously people referred to my posts from 2022 where I talked about playing around with ChatGPT and laughing at stuff it generated, even though I said there too that I'm not using it to write stories and my stories are still mine, ChatGPT does not write one single word of my stories. I quit playing with it soon after though, after I researched and found out about enviromental damage + how exactly it works, how it learns, where does it get the content it generates etc. I became extremely uncomfortable with the idea of using it even for just passing time and fun, and walked away, and now I'm very much anti-genAI and advocate for it to get banned from consumers at the very least.

But ever since that, I've had this nagging feeling that using it even just for those few months permanently altered my vocabulary and that I'm often using ChatGPT-y language when I write, and I can't read my fics without thinking "if I didn't know I wrote this, I would 100% believe this was written using ChatGPT". I'm scared to describe too much in fear of sounding like ChatGPT, but describing too little make stories sound wooden.

I started to slowly get rid of that feeling only a few months ago, when I finally started getting my confidence back. But that message I got about them suspecting/accusing me of using AI to write, I have to admit that it did discourage me from writing at all again, and I haven't written a word after I received that. I keep reading that Ray fic over and over again and see so many faults in it, and the fact that it does so poorly also contributes into me fearing people think it's been generated by ChatGPT, even though I know it's not because it's necessarily bad but because there's just a small handful of people who might be interested in Ray x readers, or Ray/Recursive Dreams fanfics in general.

I'm trying to gain my confidence back again, and fortunately I have fics queued up until 18th of July right now so I'm not in an awful hurry*. But please be patient with those Ray fics. I will also likely be postponing my next poll for upcoming fics, as it was due to come out today.

*This is exactly why my posting pace is so slow, I get asked every few weeks about why am I so slowwwww and how I should consider posting at least 4 times a week instead, etc etc. This is exactly why I want to keep up a slow but steady posting pace, because either it's this or then posts come extremely randomly, sometimes a day in between and sometimes 3 months in between, and you'll never know.

[cappulcino]

I put your writing in an ai detector and it said it was 87% ai written

I write fanfiction and I'm ADHD-autistic. What did you expect?

80% of the world-building isn't mine. 99% of the characters aren't mine. And I always make sure I don't stray too far from the source plot so I reference/quote a lot of things I didn't write.

If you take my first fanfic on here, it has so many references or direct quotes from the Bible, other myths, and works such as Paradise Lost, Penny Dreadful, or even Good Omens, and some parts have been rewritten word for word. Of course it's not going to be interpreted as original.

It's like my thesis director in college who wondered why his AI and plagiarism detector said we weren't writing our own essays when literally 75% of our writing was based on other people's works and direct quotes.

And I know I write like a robot sometimes. My sentences are always too long, the rhythm is often the same, sometimes I struggle writing certain emotions... I'm working on it. Add to that the fact that I use en-dashes and "OH mY GoD!!! OnLy AI UsEs ThOse!!!" Fuck no. Use dahes.

I'm trying to work on certain things, but it's also my style: pedantic, with strict grammar, precise ponctuation, and little room for improvisation. What should I do then? Re-craft my entire style?

(Also, let me remind you that, so far, most of what I've written called for such a style, according to me. You don't write about noble supernatural beings or knights in the medieval era with your usual, day-to-day speech –at least I don't.)

I'm wondering if this is the same anon again and again, just trying to make some sort of point after the whole ordeal the other day. Your obsession with me is starting to get weird, and I am starting to get tired of answering your idiotic comments. If at least you showed yourself... but you even hide in anon, like you're ashamed.

Nobody is forcing you to like my style –or even me. You've made very clear that AI was a perfect alternative to fandom writers, and I could not care less if you think ChatGPT or CharacterAI are capable enough. It says more about you than it does about anything else.

And you know what? After writing all this, I was curious to see what caused this extraordinarily high number of AI-generated content. So I ran 3 chapters of Seven Days Til Fall, Where the Wild Things Rest, The Shape of Us, and even the fic I'm writing right now through an AI detector, too. And guess what? The highest I got was a 10 % likelihood (yes, because that's all it measures: likelihood), and it was, unsurprisingly, on the longest text.

So next time you want to drag me through mud or something, find a better lie. I'm petty as fuck.

[asaka-lucy-dr-rc]

It’s cool to see someone in the Japanese fandom on Tumblr talking to English speakers, and your English is really good! I’ve been studying Japanese for about eight years now and even made friends over there, and my reason for learning originated with Danganronpa, yet the western and JP fandoms are so disconnected so it’s cool you’ve made an effort to reach out!

Yay, thank you very much for your kind words! 😊

To tell the truth, I still can't let go of translation and grammar correction tools when writing English, and I often consult ChatGPT to find appropriate expressions. Still, I always worry if my English is okay, so I'm glad to hear that my English is really good from your point of view!

It's amazing that you have been studying Japanese for such a long time! :D After I started Tumblr, I realized that I can study harder if it's for something I love. (For example, to read Danganronpa fanfics written in English! 😄) Also, I think it is really important to have friends who speak the language you are studying. I tend to put off studying until later, but if I get a message from a friend, I want to respond to it immediately. So I can use English much more continuously than if I decide to study for that many minutes a day.

I first played Danganronpa in the summer of 2012 (about 2 weeks before SDR2 was released), and at that time I really had no idea that this game would be so popular among overseas people, because there are obviously many parodies of Japanese games, manga and anime, etc., and I thought it would be impossible to translate them properly, and even if they did, I thought it would not be fun for non-Japanese. I am sure that there are people like you who have been supporting the Danganronpa series for a long time, and that is why the fandom is so lively today.

Again, thank you for your message! Feel free to talk to me anytime whether via ask or comment. (Actually, I need to improve my English skills for my job, so it's nice to have someone who can talk to me in English 😉)

[writers4all]

AI Bots - Are They Problematic and Should I Use Them?

With all of the uproar going on with Artificial Intelligence having the capabilities of scraping various fanfiction sites (AO3) as well as artwork sites (DeviantArt) and plagiarizing them, I wanted to give my two cents on whether or not it's okay to even use AI sites or AI programs or AI chatbots.

Some of the most iconic programs, sites, and chatbots include Cleverbot, Grammarly, Notion, and ChatGPT. Everyone has used Cleverbot at one point or another for laughs and seeing what nonsense it'll come up with. Grammarly is great when it comes to ... well ... grammar. Notion is unique in the sense that it's a writing document app with an AI built into it (and you can actually use Grammarly along with it). Which leaves us with none other than ... ... ... ... ... ChatGPT.

AIs like ChatGPT (and Dall-E) are nothing short of problematic because there are ways for people to scrape various websites. This is problematic because not only do they invade privacy but also tends to more or less take bits and pieces, if not all, of people's works (be it written, drawn, both, neither, etc.). Another problem with using AI bots is that there are a lot of people who use them to write their essays and creative assignments for school (especially college). This is incredibly problematic for a couple of reasons.

It can be taken as (and assumed that) the students were being lazy and aren't learning anything. It's like what this article on Scribber says, "Not doing the work yourself is academically dishonest, undermines your learning, and is unfair to other students."

It can, and perhaps will, cause problems later on down the road. The reason for this is because we all have a distinct voice and way we phrase things. And this is true when it comes to not only writing stories and drawing art but doing essays, art projects, and what have you.

AIs are slowly, but surely, making people lose their jobs. There are fast food restaurants and movie theaters that are already using AI machines and are somewhat in the process of hiring fewer people than before. We also have those who are in the art industry (mainly writers and drawers) who are in constant fear of losing their jobs because of AI. Doesn't help that we, the artists, aren't taken seriously as it is hence there being a Writers Strike.

Now me bringing up these three major points doesn't necessarily mean that I'm anti-AI. I'm anti-plagiarism and anti-scraping. I'm also perfectly fine with people who are disabled and are in need of using AI bots (or a program) to assist them. That's why we have programs such as Grammarly, Hemmingway App, Notion, Google Translate, and DeepL. They help us get a general idea of how to phrase, word, punctuate, and even translate things. They might not always be 100% accurate or reliable but they are there.

So, at the end of the day, it all really boils down to what you're using AI bots for. Using it because you want to improve on your grammar, spelling, and punctuation? That's okay. Using it to write your whole entire five page essay? That's not okay. Using it for research purposes like briefly explaining the similarities and differences between Thor and Raijin? That's okay. Using it to write a whole ass fanfic for you? That's not okay. Using it because you have a disability and need a voice (or whatever the case may be)? That's okay. Using it as your sole resource for research, translation, and whatnot followed by saying "AI bots are flawless, you're just jealous."? That's not okay.

So with all of that being said, AI works aren't exactly transformative, original, or creative. It's amusing as hell to see kind of shenaningans and malarkey will pop out when asking for a fanfic of a specific pairing, sure, but I wouldn't post any of it online because it's not my cup of tea. Something else to keep in mind is that, like us humans, AIs tend to make mistakes. And last, but not least, keep in mind that many of them genuinely don't care about our privacy. You can use the argument that most websites don't, and I wholeheartedly agree with you, but the point still stands.

Either way, I think it all ultimately depends on the situation for what is and isn't okay in terms of AI usage. The same is equally true on personal preferences and opinions.

[zoueriemandzijnopmars]

@vogelmeister said: i can assure you i am ready. but can you please roast my uni bot who picked up the word "transcript" and disregarded the rest of my email, because im upset

Yeah sure! Firstly though, to defend the bot, humans also sometimes do stuff like this. I was actually reminded of this meme I once made in frustration:

I don't remember what I was asking about but it was something spelled similar to convolution mentioned in an assignment and I couldn't figure out what they meant so I set an email asking about it and I got that reply.

That brings me to my point about AI bots though, because a human, you can email back, you can clarify and they will probably own up to their mistake, because hey, miscommunications happen. Chat bots in my experience however are less good at owning up to their mistake and it is not always easy to get them to reconsider.

My mom once had an issue with Bol.com and had to chat with their chatbot Billy to resolve it, now after a while it seemed like she was getting there, but then he asked for her phone number. She typed it like 20 times before giving up because somehow he didn't recognise phone numbers? But then why ask for it? It was incredibly frustrating and confusing and to this day it is still used as an example to make fun of chatbots around our house.

The problem with many old-school chatbots is that they are programmed to follow a specific script and are not always properly programmed to deal with divergences of that script. (Side note: I say chatbot, because that's what most people call it, but the correct term for these customer service bots is dialogue system (Jurafsky & Martin, 2023) (the link leads to my favourite book on speech and language processing, I think it is written really well and it is very approachable, and they quote Agatha Christie, Jane Austen and Shakespeare))

In the case of your uni's bot it is probably programmed to handle a couple of regularly occurring tasks, like people asking for a copy of their transcripts and will hence take any e-mails mentioning them and try to handle them as it is programmed.

While this is annoying, it is not necessarily a problem, as long as chatbots are used as I think they should, namely with an option to contact an actual human at all times. Like with Billy and the phone number it would have been really nice to get to talk to a human and pass on the phone number. Unfortunately companies want to save money by only using chatbots and it can be quite difficult to get to talk to an actual human.

A bigger problem occurs with the actual 'chatbots', while early chatbots like ELIZA (Weizenbaum, 1966) were mainly used just to illustrate the possibilities of computers (and although impressive, once you know what to look for it is pretty clear the answers are generated by a surprisingly simple algorithm). They didn't come up with original thoughts and often just repeated what the user said or used the same phrase designated to keep the user talking, over and over again (e.g. Can you tell more about X).

(One interesting early concern was already privacy. Many users of ELIZA discussed quite personal topics with her which could be logged. More recent chatbots have had problems with them learning from humans and therefore accidentally repeating private information that they got told.)

The newest hot AI chatbot, chatgpt (my beloved, as my sister claims) can do very impressive things, it speaks multiple languages, it can write your code for you, it can write songs, poems, stories, and it can even do math (sometimes). I have personally mainly just played around with it but I did use it to generate a fantasy name for an oc side character in a fanfic and to generate an anagram name for the fictional app I had to pitch for an assignment. In my opinion it is pretty great for these kinds of tasks, where it can give you some suggestions and you can pick what you like best and possibly even adapt it. (Disclaimer: I said it can write your code but it is mainly good at code snippets, whole programs are a lot harder, so again it is mainly good for inspiration.)

While this would be my preferred use, what has gotten a lot of traction is the fact that it can write whole papers for people, but also that it sometimes "hallucinates" references that do not actually exist. Many AI systems have been optimised to not be wrong. My hypothesis (I am not an expert in this field and did not look at the source code) is that chatgpt has learned that it is better to guess than to not answer, quite similar to students writing an exam in school, not answering is always wrong, better make something up and hope for some points. This is why you should never trust what an AI tells you up front, it can help you do things, but you should always check the output and use your own common sense.

So far so good, chatgpt is a tool that can be used as long as the user checks the output. Similar to the earlier discussed dialogue systems, it is good to be able to have a human in the loop. The real problem with chatgp in my opinion is not that it makes mistakes, but that it refuses to own up to them. Most people who follow me here would know that I am quite the fan of the Dutch duology The Letter for the King/Secrets of the Wild Wood. I hence couldn't resist the chance to talk to chatgpt about them. There were a lot of fascinating things in that conversation, but the thing relevant to this essay is that it claimed that Piak was a thief. For people not aware of the plot of the books, Piak is a 14 year old boy, he was orphaned and has lived with Menaurus (a Gandalf-like wise old man figure) ever since, during the first book he becomes Tiuri's (the main character) best friend. He is most definitely not a thief, but chatgpt didn't accept my counter arguments and stayed by its story.

Therefore although it looks like we got a long way from the first customer service bots. Chatgpt still suffers from the same underlying problem, being programmed to optimise always being right. This means it is still unable to admit to gaps in its knowledge, it doesn't recognise when it made mistakes, and it cannot change its approach accordingly.

I have so many nuanced takes on AI that this website really is not ready for