At what point does a story count as AI written?

[Ginger Herten]

With out giving too much away, the basic premise of one of the stories I’m working on is that the main character starts using AI, and then ends up relying on it. I have been using chatGPT to write the sections that are supposed to be the AI in the story, because I want them to sound AI like. Everything else has been written by me though, because I write these things as a creative outlet. I’m also editing and manipulating the chatGPT sections pretty heavily because I have a pretty clear idea of what I want it to say, and the prompts don’t always get me there.

So here’s my question at what point does it make sense to label a story as AI written?

It’s not like anything I write is ever purely AI free these days, I write in google docs, and it just automatically points out not only my numerous spelling mistakes but also what it veiws as grammar mistakes. Sometimes it’s right about the grammar and I change things, sometimes I have just worded things in a less typical way for stylistic reasons, like using the passive voice to create a feeling of lack of control or emotional detachment. But obviously I would never think of labeling those as AI written, they are just barely AI edited.

But this upcoming story is going to have a few significant and important parts that are AI written, but yet it’s not like I said to chatGPT “write me a story about a character who lets AI take over her work writing.”

[klaatu48]

This could get all sorts of philosophical…

First, I wouldn’t aim to set any hard and fast rules for what qualifies as AI just for this reason.

From what you’ve said, I don’t think it would qualify as AI-written because you’re incorporating AI into the story you’re writing — you’re not just telling it what story to write.

Ai writing, to me, tends to lack soul. The program doesn’t know what grammatical rules can be broken and when — it doesn’t know when emotions are conveyed better through run-on sentences than through simple sentences, and it doesn’t know how to explain the physical rush you get from a first kiss. It’s just… there.

Maybe the answer is that it depends on how much of themself the writer puts into it — does your heart start racing when you get into the flow of the story, or are you just entering prompts and seeing what comes up?

Maybe I’m a literary snob… Who knows.

But to answer your question, it doesn’t sound like an ai-written story to me.

[f35h]

For me, the ‘AI Written’ category should be used for stories which are completely AI generated, or only have minor modifications to the AI output.

What constitutes ‘minor modification’? I don’t think that’s easy to define, I guess we have to leave it up to the author. Certainly what you describe I wouldn’t call ‘AI Written’, you’re really just using for inspiration / assistance, not getting it to write the story.

[Ginger Herten]

I guess I am asking on a more philosophical level, as well as a practical one.

If the sole purpose of the category is to warn readers that the story might be flat and repetitive, though rather readable since it will be typo free and utilize reasonable paragraph breaks, then my story wouldn’t count. Since I have written most of it, most of it will be my usual warm messy style, probably including a few typos that I will have looked at a dozen times and my brain just refused to see anything other than what I intended to write. (There may even be a typo or two in the AI written parts because I edited them because I didn’t manipulate chatGPT into writing exactly what I wanted, more for the sake of privacy than AI limitations. I suspect anything I ask it will influence the advertising I see.) And of course, if the category is viewed that way by most, it would be misleading to readers for me to use it.

The AI in my story is pretty much a character, it is extremely obvious in my story what parts are AI written, because of that. Not mentioning it, feels like leaving out a major element.

Of course then again, our categories are already limited. We can’t have 100 different categories. If it were practical, we should have dozens categories that cover the gray area in between “forced” and “consensual.” Like even within the singular scenario of having to get a haircut for a job, the line between forced and consensual is blurred: does the character just enjoy the job; does the character need the job to maintain the comfortable lifestyle they are used to; is the job the only one they could find and they are about to be evicted along with their elderly father who needs expensive medication not covered by insurance? So as it is, we don’t draw hard lines, we just put both tags on the story, since either can fit.

[Ginger Herten]

So of course, posting that got me thinking, so I went back and checked my own stories, and it looks like I tend to lean slightly towards just labelling them “consensual“ in the need to for a job scenario. I seem to only label them as forced when the job makes it extra difficult to say “no” or the person doing the cutting added an extra level of coercion to the scenario.

So I guess I do tend already to use the forced tag some what conservatively. So applying the same standard to the AI tag, I would leave it off.

[klaatu48]

That’s a valid point about the grey areas. For me, in regard to the consensual/forced aspect, if it’s something the person wouldn’t do on their own but they are willing to to get the job, it’d be consensual. If it’s something they don’t want to do but they have to do to get a job and they have no other options, then it’d be forced. So, like you said, it’s definitely blurred and open to interpretation.

[Marquis.d.ayre]

I tried using AI (ChatGPT) but I found it too restricive for the kind of stories in this. Also, there’s a lot of work in getting the AI to write something to the point – at this moment it’s too overrated and for purpose of this site, too woke. Nowadays I tend to write, convert it to audio (I’m auditive) and make my corrections. I do use AI once in a while to make grammatical corrections (I’m not a native English speaker).
AI is to wooly

[Gabriel]

Hi, about a year ago I started using ChatGPT. Still use it for quick generated stories. But as I want it to write something in which I can feel a bit of tension… (in myself 🙂 ) it doesn’t want to help me. Often I turn up in RED or it starts to write and breaks off, RED “no adult material or explicit stuff” or something like that.
Also haircutting is not one of it its main topics, not able to write much of the process or the feelings of the character.

But OK, within it limits I now and then nevertheless get satisfied with the result (RED or not). But it is no quality to publish it, as it uses those standard sentences as if it is a fairy tale (lived long and happily every after). But okay, as I said for quickies, wonderful to see with what it comes.

[Marquis.d.ayre]

Yeah Gabriel, that’s what I was trying to say, you put it more eloquent. At least ChatGPT is too woke to be any good – and they you should try to make it pictures for those stories. . . .

[Author]

Posts