The AI Rejection Conversation Matters

[ShelbyDawn57]

I usually make my tags more elaborate...

"I agree with you," he concurred, slowly undoing the buckle of his jeans as John slipped to his knees..

"I'm happy," Sara smiled, her eyes wide as the scene before her unfolded.

"Let me tell you how it is," Hector mansplained, grasping the sides of John's head firmly between his powerful hands.

"I spit at you," John expectorated, spewing Hectors hot seed on the more dominant man's shoes.

Wait, did I just violate your implied copyright on this snippet??? Shit. I need to call my attorney...

 

[Thanatos_X13]

To be fair, having given some such advice myself, I think most of that advice was given in good faith and with a desire to help. But what I realize now is that you can follow all that advice and still get caught in the jaws of the AI Filter Monster. So, some humble pie is appropriate, as is a reassessment of what the site is doing and how to navigate it.

I'm all ears.

I'm curious how people have responded to the rejection, what they've done, and how their efforts have been responded to. I don't see how a volunteer editor can help. I don't want to change anything I've written, because I know it's all my own.

I understand and concur with the Site's policy of trying to exclude AI, but the policy obviously is not quite working.

If it helps, I documented my own experiences with this problem and what I did to get around it back in April/May. The top post was updated with everything I did. I don't claim that I have "solved" anything, or that my steps will work for everyone, but I hope it at least helps. For me, once I got around the AI issue once, I've been fine ever since. https://forum.literotica.com/threads/another-ai-rejection-post-5-30-update.1609076/

The thing to keep in mind is that once you've been flagged by whatever system Lit uses, you need to look for a "detector" that is flagging you and go from there. It doesn't matter that 9/10 don't flag you, b/c that just means those are definitely not the ones Lit is using. For me, I kept getting flagged as 40%+ AI generated by https://gptzero.me/ so I used that as my baseline, created a bunch of free accounts to get around the scan limits, and worked to get that % as low as I could. Unfortunately, that does mean you have to compromise yourself stylistically. For me, I was happy if I could keep 90-95% of my style intact and just break things in a few places. That ultimately worked out for me, but it's a personal choice how far you want to go.

I'm sorry you got caught in it. It sucks, and it's unfair, and it's bullshit.

 

[SaraLab]

Welp, I just had another 2 stories rejected today for being "composed largely of AI-generated prose", despite scoring 0% on ZeroGPT as well.

 

[SaraLab]

My suggestion: press your case with messages to Laurel and cite the results you get from the AI detectors you are using. My experience in the past is that Laurel doesn't always respond immediately but she does eventually and is usually reasonable and courteous. This might be a case where the pleasantly squeaky wheel gets the grease.

Has that ever worked for someone in the case of mistaken AI detection? I saw someone else saying that there submission sat pending for 14 days, only to be rejected for AI once more.

 

[Katie_Mae]

But that would be using AI to correct AI. It might get me past the filter, but it wouldn't solve the problem in a meaningful way, right? (I think that's your point)

It seems absurd that I should have to use AI tools to make my authentically personal prose sound like it's not AI.

When I was looking through the rules the other day, I was surprised to see them encouraging the use of AI for generating cover images for series.

We know there's AI generated stories which Laurel accepted, knowing they were AI and she hasn't bothered to try and take down. I'm not sure if actual AI use bothers her as much as apparent AI use.

I'm not sure what to do. The rejection notice recommends having a volunteer editor look at the story but I don't see how a volunteer editor can check for AI compliance or help me. If I were to take the volunteer editor's suggestions the story would be less my own than it is now.

The question is, have you been rejected for using AI, or for your style being similar to AI?

If it's the former, you can show not AI use by showing proof such as drafts, version history, a screen recording of you writing it* etc

If it's the latter, then you can use an editor to help you not sound like what Laurel thinks is AI.

*don't laugh, I know of students who do this now to fight false AI claims.

 

[SaraLab]

When I was looking through the rules the other day, I was surprised to see them encouraging the use of AI for generating cover images for series.

We know there's AI generated stories which Laurel accepted, knowing they were AI and she hasn't bothered to try and take down. I'm not sure if actual AI use bothers her as much as apparent AI use.



The question is, have you been rejected for using AI, or for your style being similar to AI?

If it's the former, you can show not AI use by showing proof such as drafts, version history, a screen recording of you writing it* etc

If it's the latter, then you can use an editor to help you not sound like what Laurel thinks is AI.

*don't laugh, I know of students who do this now to fight false AI claims.

How are you supposed to submit the drafts for evidence to Laurel? The 'Note to Admin (optional, not published)' section doesn't have the ability to upload files.

 

[Katie_Mae]

How are you supposed to submit the drafts for evidence to Laurel? The 'Note to Admin (optional, not published)' section doesn't have the ability to upload files.

That was rather my point. Laurel's rejection claims it's due to AI use, but then the remedy suggested is to change your style, not prove it's not AI.

 

[Thanatos_X13]

Welp, I just had another 2 stories rejected today for being "composed largely of AI-generated prose", despite scoring 0% on ZeroGPT as well.

This happened to me, too. I'm going to repeat what I wrote to SimonDoom

The thing to keep in mind is that once you've been flagged by whatever system Lit uses, you need to look for a "detector" that is flagging you and go from there. It doesn't matter that 9/10 don't flag you, b/c that just means those are definitely not the ones Lit is using. For me, I kept getting flagged as 40%+ AI generated by https://gptzero.me/ so I used that as my baseline, created a bunch of free accounts to get around the scan limits, and worked to get that % as low as I could. Unfortunately, that does mean you have to compromise yourself stylistically. For me, I was happy if I could keep 90-95% of my style intact and just break things in a few places. That ultimately worked out for me, but it's a personal choice how far you want to go.

Unfortunately, there is no way to "prove" you didn't use AI, since that's not the process Lit is using. There's no appeal or adjudication process. If your work gets flagged, it's up to you to "fix" it and resubmit. Which is total bullshit when your work is 100% human-written, but that's the reality of the process Lit has chosen to use to try and combat AI submissions. Your only options are to rework your manuscript to try to pass their filter, or pick up your ball and go home.

 

[MillieDynamite]

It is worded as suspected AI usage. You don't get arrested for suspected drug usage, suspected grand theft auto, or ticketed for suspected jaywalking. Why is suspicion enough to reject a story wholesale? Just saying, it's a very subjective type of suspicion.

This happened to me, too. I'm going to repeat what I wrote to SimonDoom



Unfortunately, there is no way to "prove" you didn't use AI, since that's not the process Lit is using. There's no appeal or adjudication process. If your work gets flagged, it's up to you to "fix" it and resubmit. Which is total bullshit when your work is 100% human-written, but that's the reality of the process Lit has chosen to use to try and combat AI submissions. Your only options are to rework your manuscript to try to pass their filter, or pick up your ball and go home.

 

[SaraLab]

I wonder if the best option at this point is to find someone (maybe an editor) who is willing to help "improve" parts of the story to break whatever patterns are triggering the AI detector.

My editor suggested resubmitting them as is while he sent Laurel a message, but I haven't heard of that working for people?

 

[SimonDoom]

I wonder if the best option at this point is to find someone (maybe an editor) who is willing to help "improve" parts of the story to break whatever patterns are triggering the AI detector.

My editor suggested resubmitting them as is while he sent Laurel a message, but I haven't heard of that working for people?

This is what I have done. I'm waiting for a response.

Part of the problem is that Lit gives you no information about what triggered the AI filter. I submitted it to ZeroGPT, and it flagged some text as suspected, but I carefully reviewed everything flagged and every single word of it is my own, without any input from AI. There's nothing "wrong" or "odd" about it. It's the same style I've been writing in for 8 years.

 

[MillieDynamite]

It's probably just your turn. Mary had one rejected. Our editor/publisher went through and changed a few awkward sentences, and Mary (@theWollstonecraftWoman) re-submitted it. It was up in a day. He changed five sentences. Out of however many words it was, that's nothing.

This is what I have done. I'm waiting for a response.

Part of the problem is that Lit gives you no information about what triggered the AI filter. I submitted it to ZeroGPT, and it flagged some text as suspected, but I carefully reviewed everything flagged and every single word of it is my own, without any input from AI. There's nothing "wrong" or "odd" about it. It's the same style I've been writing in for 8 years.

 

[Thanatos_X13]

It is worded as suspected AI usage. You don't get arrested for suspected drug usage, suspected grand theft auto, or ticketed for suspected jaywalking. Why is suspicion enough to reject a story wholesale? Just saying, it's a very subjective type of suspicion.

Oh, I fully agree. It's a crappy situation especially considering that AI "detectors" are all smoke and mirrors anyway. Unfortunately, until Lit makes a change to this process, suspicion is enough. I'd feel less irked by it if A) it didn't take 14+ days to get a rejection over AI, B) the feedback was more insightful, and C) there was a process to dispute or try to prove your case. But at this point, that seems like a pipe dream >_>

 

[MillieDynamite]

What the checkers look for is the repetition of certain keywords and phrases that are common with AI writers. The over-repetition of words such as blue eyes and brown hair in descriptions of what characters are doing. "She gazed at him with her blue eyes twirling a strand of her brown hair." They also look at types of sentence structure, over long and complicated sentences, or shot commonly used sentences. AI also uses lots of alternate sentence tags. My standard go-to tag (if one is needed) is he said, she said, she asked, he asked, and then rarely she answered, or he answered.

Oh, I fully agree. It's a crappy situation especially considering that AI "detectors" are all smoke and mirrors anyway. Unfortunately, until Lit makes a change to this process, suspicion is enough. I'd feel less irked by it if A) it didn't take 14+ days to get a rejection over AI, B) the feedback was more insightful, and C) there was a process to dispute or try to prove your case. But at this point, that seems like a pipe dream >_>

 

[RejectReality]

I can tell you I'm not wasting time fighting with some garbage AI detector. If something gets erroneously sent back to me for AI, it's simply not going on Lit. If it's a chapter of an in-progress story, I'll pull the whole series. I'm not going to dance with some dumb program. I'll just post them on one of my other venues.

I originally said I'd give it one re-up with a note, but that's out the window now. Too many reports have said that's a waste of time. I'm not rewriting a single one of my own words to please a hallucinating algorithm.

 

[Omenainen]

…over long and complicated sentences…

That’s it, I’m done for

 

[SimonDoom]

Well, hmmmm. Shucks. I just heard from Laurel, and the story had been rejected in error. They meant to approve it. The decision has been reversed and the story has been published.

Never mind.

 

[StillStunned]

I want to make a funny comment, but nothing actually seems that funny when I type it out. All I can say is that I can only imagine your relief after what must have been a shitty few hours. Congrats!

 

[MillieDynamite]

Many, many overly long and complicated sentences, not one here and there.

That’s it, I’m done for

EDIT: I did know you were joking. I wanted to clarify what I was saying.

 

[StillStunned]

Many, many overly long and complicated sentences, not one here and there.

What always strikes me about AI-generated text, and the snippets that have been shared in the various threads, is how many sentences are structured the same. Two clauses, separated by a comma. Sentence after sentence after sentence. Like a first-year uni student who's trying to write "grown up", but can only present their thoughts in single bites.

 

[SimonDoom]

What always strikes me about AI-generated text, and the snippets that have been shared in the various threads, is how many sentences are structured the same. Two clauses, separated by a comma. Sentence after sentence after sentence. Like a first-year uni student who's trying to write "grown up", but can only present their thoughts in single bites.

This is my impression as well.

 

[MillieDynamite]

What's that word I've used about AI writing? MMM, oh, yes, it sucks.

Sucks is the word, is the word that you heard
It's got a vacuum, it's got a meaning
Sucks is the time, is the place, is the notion
Sucks is the way we are feeling

AI sucks,
We start believin' now that we write what were feeling
and AI sucks, Sucks is the word

 

[Britva415]

I'll change all my dialogue tags to colorful, clever, overly descriptive terms that step all over the dialogue. Maybe that will help.

"I agree with you," he concurred.

"I'm happy," Sara smiled.

"Let me tell you how it is," Hector mansplained.

"I spit at you," John expectorated.


Bet that gets by the filter.

Ah, the old Tom Swifty.

 

[Britva415]

have you been rejected for using AI, or for your style being similar to AI?

I haven’t seen an example of a “reasons for rejection” notification which seems to indicate any meaningful difference.

Besides: It’s clear that “style” isn’t a reason for rejection around here. Grammar, spelling and punctuation? Sure. But not style, unless you mean the “Chicago Manual” or “Strunk and White” kind of “style.”

 

[StillStunned]

Besides: It’s clear that “style” isn’t a reason for rejection around here. Grammar, spelling and punctuation? Sure. But not style, unless you mean the “Chicago Manual” or “Strunk and White” kind of “style.”

I don't know about that. Most of the complaints seem to be about AI writing a flat, monotonous, dead style. See my post above, about every sentence having the same structure without any effort to flow from one to another.