• A notice about Rule 3 regarding sites hosting pirated/unauthorized content has been made. Please see here for details.
  • Staff is working to deal with the problem of synonymous tags. See here for more information and to suggest tag mergers.
  • QQ will have forum downtime on October 21, starting at 10:00 AM Eastern time, for software updates.
  • Due to issues with external spam filters, QQ is currently unable to send any mail to Microsoft E-mail addresses. This includes any account at live.com, hotmail.com or msn.com. Signing up to the forum with one of these addresses will result in your verification E-mail never arriving. For best results, please use a different E-mail provider for your QQ address.
  • For prospective new members, a word of warning: don't use common names like Dennis, Simon, or Kenny if you decide to create an account. Spammers have used them all before you and gotten those names flagged in the anti-spam databases. Your account registration will be rejected because of it.
  • Since it has happened MULTIPLE times now, I want to be very clear about this. You do not get to abandon an account and create a new one. You do not get to pass an account to someone else and create a new one. If you do so anyway, you will be banned for creating sockpuppets.
  • Due to the actions of particularly persistent spammers and trolls, we will be banning disposable email addresses from today onward.

What does "AI Assisted" even mean?

Narcissist

Narcissist

Not too sore, are you?
Joined
Apr 27, 2014
Messages
364
Likes received
4,889
So, what's the deal with the AI Assisted tag? Are we talking, "It wrote the entire rough draft that I rewrote in my own words." Or is it more, "I used AI to point out grammar mistakes and repetitive word choices."

As someone who's writing two fanfics, I plan to tag them "Humane Certified" to demonstrate that I have not used AI to write my fics. But the AI Assisted tag bugs me. Grammar Checkers have been around for a hot minute, as have spell checks. However many of them have gone over to utilizing AI to determine the context of what you're writing, correct your homophones, and yell at you for not using the oxford comma.

So, where do you draw the line? At what point does a fic stop being au naturel grass fed free ranged human writing?
 
Personally, I would say more the former than the latter. One of the main problems with AI is that it isn't human. It's a Large Language Model using data nodes and parameters. Groomed to speak with confidence, even if the information it reiterates is filled with errors and/or hallucinations.

It gets things wrong a lot. Suffers from memory issues, and writes in a format that eventually becomes easy to discern due to the fact that it's the same data nodes and parameters being utilized over and over by anyone using the technology.

That's why it's so common to get a "feel" for AI even if some are incapable of pointing at any specific passages why something "feels" AI. Your brain is trained to look at patterns, which is cranked to the extreme when you read several stories by different authors but written by the same ghostwriter.
 
Narcissist said:
So, where do you draw the line? At what point does a fic stop being au naturel grass fed free ranged human writing?
Click to expand...
Click to shrink...
Anything beyond spell checkers, imo.

There's no clear-cut divide, more like spectrum and arbitrary points on this spectrum that people consider to be their line, when the thing stops being human writing.

AI assisted is vibe based tag and it depends on the author what it means because of how vague this is.

Even using AI for ideas and writing completely "free ranged human writing" can be easily called an assistance from AI.
 
Narcissist said:
Grammar Checkers have been around for a hot minute, as have spell checks. However many of them have gone over to utilizing AI to determine the context of what you're writing, correct your homophones, and yell at you for not using the oxford comma.
Click to expand...
Click to shrink...
BTW, if you're on Firefox and you use/used LanguageTool, you can utilize their previous versions to avoid their bait and switch to AI, just make sure you never update it.
 
The tag means "my work is so blatantly AI-generated that I don't believe I could get away with lying about it." Anybody using an LLM at the level of mere "assistance" will just keep quiet about it entirely.

As for how the tag should be used, I'd say no more than a few words in a row should ever be an LLM's direct output, or it's no longer just an assistant. If you're using AI as an assistant, it should be answering questions, collecting data, reminding you of missed details, you know, like an assistant. If it's writing, it's at minimum a co-author.
 
kelpsie said:
The tag means "my work is so blatantly AI-generated that I don't believe I could get away with lying about it." Anybody using an LLM at the level of mere "assistance" will just keep quiet about it entirely.
Click to expand...
Click to shrink...
Yeah. Thing is, "AI" writing isn't popular at all; as far as I can tell there's two basic positions on it, "don't care" and "hate it". Which in turn means there's no positive reason to admit using it (outside of, perhaps, being too honest for your own good), since it will be either ignored or drive potential readers away. From a self-interested point of view it only makes sense as a means of evading accusations of lying.
 
I don't think an official definition of what constitutes AI work exists. It will probably be based on current Copyright/Plagiarism rules. If X% of your work is created by a computer then you did not write it kind of stuff.

If we are looking for a definition i personally think the following.

1. 50% or more of the words are AI created = AI generated.
2. 20% to 49% = AI assisted.
3. Anything below 20% falls into the realm of editing which i don't think anyone has a problem with.

Having said that, it's also not something that normal people are going to be able to tell at a glance. AI use is very much a Yes or No question right now with no way to verify the inbetween.

Edit: Any rules or definitions we come up with for AI are just going to be used for training. The next version won't do those things and we will have to come up with new rules.

The entire thing falls under the do not give an order you can't enforce rule.
 
Last edited:
kelpsie said:
The tag means "my work is so blatantly AI-generated that I don't believe I could get away with lying about it." Anybody using an LLM at the level of mere "assistance" will just keep quiet about it entirely.
Click to expand...
Click to shrink...

Exactly this. I've yet to see a story tagged as 'ai-assisted' that did not very blatantly just have the AI do a the vast majority, if not all of the writing.

I don't know if they all use the same model or if AI in general just do 'creative writing' the same way but all AI written stories just sound so samey with tone, cadence, dialogue, etc, etc.
 
xihute said:
Anything beyond spell checkers, imo.

There's no clear-cut divide, more like spectrum and arbitrary points on this spectrum that people consider to be their line, when the thing stops being human writing.

AI assisted is vibe based tag and it depends on the author what it means because of how vague this is.

Even using AI for ideas and writing completely "free ranged human writing" can be easily called an assistance from AI.
Click to expand...
Click to shrink...
Personally, I feel like I wouldn't count AI involvement in ideas and outlining as "AI assisted", in the sense of it being a useful label for people to filter out stories they don't want to see, if the actual text is human-written. It might be really formulaic or whatever, but a huge amount of modern fiction is literally formulaic, as in, a lot of really popular writers literally have what amounts to mad libs templates for the overall plot. Dan Brown's books are a good example of that, where you can often see the exact same pattern in chapters between different books (numbers of chapters, tempo, pacing, character perspective swaps, etc). The human factor comes in with the actual writing: you could give Dan Brown and Stephen King exactly the same story outline and plot beats and nobody would mistake the two for each other.
 
Last edited:
Narcissist said:
"I used AI to point out grammar mistakes and repetitive word choices."
Click to expand...
Click to shrink...

This, but I also bounce ideas off of AI, ask it for suggestions on a characters potential actions, stuff like that.

I deliberately avoid asking it for generated text and instead simply ask it for corrections and ideas.
 
Avernus said:
Yeah. Thing is, "AI" writing isn't popular at all; as far as I can tell there's two basic positions on it, "don't care" and "hate it". Which in turn means there's no positive reason to admit using it (outside of, perhaps, being too honest for your own good), since it will be either ignored or drive potential readers away. From a self-interested point of view it only makes sense as a means of evading accusations of lying.
Click to expand...
Click to shrink...

Brother, I just skimmed a story that's the literal definition of AI generated slop and the author has a patreon which many people have joined. So, all this hate for AI seems like the loud minority stuff because as unbelievable and mind boggling as it seems, people are making significant money off of AI generated stuff.

I wish that weren't the case but the story I just mentioned skimming, even had praise for the prose which was as AI as one could get. Hell, AI sex chatbots are more refreshing and less repetitive in their prose than that story but it was hailed as peak People will chug down anything these days.
 
Does grammerly count as someone who is almost dyslexic. Was assessed just scrapped in as not it very appealing.
 
dylanredefined said:
Does grammerly count as someone who is almost dyslexic. Was assessed just scrapped in as not it very appealing.
Click to expand...
Click to shrink...
Imo, it doesn't matter *why* you are using something for a question *if* you are using it.

I would consider grammarly as a very-weak AI-assistance. If it helps, use it and don't worry what other think.

The practice and seeing corrections should also help with writing.
 
dylanredefined said:
Does grammerly count as someone who is almost dyslexic. Was assessed just scrapped in as not it very appealing.
Click to expand...
Click to shrink...
That's a very low level AI. Hell, it's more of an editor tool. The main prose is still your own so there's no issue there.

The problems start when people are generating entire excerpts and portions of a story using AI.

Think of the difference as such: you sculpt/design something by yourself and then have someone polish it for you. That's still your effort at the core. But if you download a design from the internet to 3D print and call it your own original work with some assistance, then that's fraud.

So, at the price of being redundant: corrections are fine. Wholesale generation of passages is not.
 
ForestDweller said:
What's the harm?
Click to expand...
Click to shrink...
When I read a story, I want to see the intent of the author behind the words. AI generated passages do not have that. It's one thing to use a small generated passage as an inspiration to build things off of but it's another issue entirely when large portions or the entire story is generated and copy pasted as some original work.

I'll get much more entertainment by using a chatbot to imagine having personal adventures than reading that kind of stuff.
 
ForestDweller said:
Really? I just want to read a good story.
Click to expand...
Click to shrink...
I have yet to read a story written mainly by AI that doesn't feel off somehow. You can always feel that something is slightly wrong, it emphasises things a human writer wouldn't, and it forgets stuff.

Take this as an example:

Someone who can't speak English very well decided they had a good idea for a story, wanted to see it, gave minimal guidance to an AI, looked at the output, decided it was going in a good direction overall and uploaded it here without editing the output at all. It is entirely readable and the premise is actually pretty good (to me).
You can tell it's all AI just by how the dialogue is structured pretty quickly. There are many instances where characters should know something and don't show it, as well as instances where they know things they shouldn't.
Example for the first of these: we have Malfoy and Harry, both aware of the horcrux in the diary, talking about what to do about second year. Neither of them mentions it because Malfoy doesn't want to talk about Lucius being maybe the slightest bit evil and the AI thinks Harry doesn't know. Example for the second case: Harry thinks about a scene that happened with Malfoy and family while reading a letter from her, but what he thinks about wasn't in that letter and he'd have no other way to know about it.

Issues like that happen with AI all the time and I'd say they count as generally detrimental to the quality of a story.

As for what the thread actually is about, I'd separate it into three levels: a story counts as ai-edited or beta'd as soon as the AI gets a chapter and gets told "check if this section needs to be rewritten to sound better" and any of its input is accepted. AI-assisted should be if a writer lets it create a section or two from a storyboard and then checks it over (like if they're not making progress on a scene; this is probably the most useful thing AI can do and it shouldn't detect from the quality). The story I have as an example above and anything else where the AI gets a storyboard and then just writes everything should count as AI-written or something. Right now, all of these cases are tagged as AI-assisted if they're tagged at all, which is pretty annoying.
 
Last edited:
You must log in or register to reply here.

Users who are viewing this thread

Total: 1 (members: 0, guests: 1)
Back
Top