How to Avoid AI Rejection

[SimonDoom]

There have been many threads lately by authors whose stories have been rejected by the Site for not passing the AI-generation screening. There seem to be a great deal of false positives.

I'm not an expert in AI, but I've also never had this problem with my stories, and I've read enough AI-generated stories and seen samples of text that apparently were rejected that I think I have some idea why these stories are being rejected, and I have some recommendations that I think may help avoid rejection. Keep in mind, I don't KNOW for sure that these ideas will avoid rejection. I welcome any additional suggestions people may have. The purpose of this thread is to offer positive suggestions. So here goes:

1. The number one suggestion is NOT to use true AI tools to write your story. Don't use things like ChatGPT. Just don't do it, period.
2. If you use tools like Grammarly and similar tools (I use them all the time), use them to identify problems, but don't accept all of their proposed solutions. Correct clear grammatical, spelling, and punctuation problems (but see below). But DON'T regularly follow their suggestions for rewording things or for restructuring your sentences.
3. In fact, make a point of REJECTING the tool's suggestions, and add a distinctive word of your own here and there.
4. Be conscious of whether your story reads like it's written as an essay. This, to me, is what a lot of AI-generated prose reads like. The sentences are too long, and the structure and flow are too monotonous. Too many commas.
5. Consciously mix up the length of your sentences and paragraphs. Mix short ones with long ones. I've noticed that with AI-generated text there's a tendency for sentences and paragraphs to feel the same.
6. Sometimes make your dialogue grammatically incorrect. People do not speak in a grammatically perfect manner, and the SUREST sign that your writing is overinfluenced by AI is that your dialogue sounds smooth and perfect just like your narrative. Mix in filler words like "like" and "you know," because that's the way real people, but not machines, talk.
7. Spice up your prose. Have a distinctive style. Add descriptive words and phrases here and there that are unusual. Add metaphors and similes -- but try to make them yours; don't just add cliches, i.e., "pleased as punch."
8. Re-read your text, and replace some bland verbs with more interesting ones. DON'T use AI for suggestions for replacement words. Look them up or think them up yourself.
9. Make your prose vivid and specific. AI-prose tends to have a vague, abstract, generalized quality. There's a lack of concreteness. Be concrete and specific. I think the more you show rather than tell, the less likely you are to be tagged for AI. If your character is climbing a mountain, don't just tell the reader that the character was challenged by the experience; show the reader exactly how their hands were bruised and dirtied.
10. Do things that good editors know are perfectly fine in fiction, but that machine editors might chide you for--like starting a sentence with "And" or "But." There's nothing wrong with that. End sentences with prepositions. Avoid writing like a pedant. ESPECIALLY in dialogue.

That's it for now. I welcome other suggestions. Based on my limited experience with the AI phenomenon, I think these ideas will help. I'll add more if I think of more ideas.

 

[StillStunned]

My suggestion:

Think about how your sentences and paragraphs are structured. Identify what elements belong together, and split your sentences accordingly. Transition to the next sentence or paragraph like a camera panning over the scene. Every element that you present should follow logically from the one before.

 

[karina_jayde]

My suggestion:

Think about how your sentences and paragraphs are structured. Identify what elements belong together, and split your sentences accordingly. Transition to the next sentence or paragraph like a camera panning over the scene. Every element that you present should follow logically from the one before.

That's definitely an uncanny thing I notice with AI writing. Each sentence on its own may be correct and coherent, but the pace, rhythm, and flow of the bigger picture is off.

 

[AchtungNight]

I just submitted a story last night with AI antagonists (thug androids) commanded by a pair of villains who act very illogical. The heroes aren’t sure what they’re up against until the final battle but take several different steps to counter potential threats. This approach may be worth repeating. Don’t have all your characters act predictably or the same way. Insane chaotic aligned characters are not something easily written by AI.

 

[KeithD]

Is rejection happening to anyone who was having stories accepted into the file a year ago, or is it newer authors being scrutinized? Following writing adjustment suggestions such as Simon is giving might well keep you from being rejected at Literotica, but I wouldn't accept limitations to this degree in writing a story just to get it in the Literotica file. I already curb my writing to a slight extent to be published here, but at some point you are sacrificing your voice too much.

 

[Priscilla_June]

Honestly I think that dialogue bit that you mentioned has saved me above all else. I make it a point to write my dialogue very clunky, stuttering, unsure and with less absolutes because its more human. I wish more people wrote like that sometimes.

 

[MrPixel]

Mix in filler words like "like" and "you know,"...

Or natural conversational lead-ins such as "Yeah, but...", "Uh huh," "Oh, I dunno," or "Really?".

 

[Kumquatqueen]

That's definitely an uncanny thing I notice with AI writing. Each sentence on its own may be correct and coherent, but the pace, rhythm, and flow of the bigger picture is off.

It's also a thing with a lot of inexperienced writers that I've seen online - every sentence has much the same rhythm, subject did object to indirect object, little variation in sentence length or order.

I think part of it is related to how essay writing is taught in American schools, with short opinionated topic sentences, then related sentence, another related sentence - there's a rather bland-textured result that's very detectable as American in online fora such as Quora or Reddit - anywhere someone writes multiple paragraphs but may not be very used to it.

AI imitates that blunt straightforward style very well. Get drunk and look sideways at things, and pretend to be a queer Brit or a piss-taking Australian, and the AI checker will likely run away.

 

[SimonDoom]

Is rejection happening to anyone who was having stories accepted into the file a year ago, or is it newer authors being scrutinized? Following writing adjustment suggestions such as Simon is giving might well keep you from being rejected at Literotica, but I wouldn't accept limitations to this degree in writing a story just to get it in the Literotica file. I already curb my writing to a slight extent to be published here, but at some point you are sacrificing your voice too much.

I'm curious that you saw my advice as imposing "limitations" or "curbing" an author's voice. I thought it was the opposite. I think the problem stems from authors using tools that stifle their own voice and make them sound generic.

I suppose I should add the caveat to my remarks that these are just guidelines, and there are almost innumerable ways to make one's prose distinctive, but distinctiveness is the point. I'm offering ideas about ways that one can avoid common problems that get stories flagged for AI use. Authors who already have their own distinctive voices are unlikely to have this problem or have use for this advice.

 

[Duleigh]

My suggestion:

Think about how your sentences and paragraphs are structured. Identify what elements belong together, and split your sentences accordingly. Transition to the next sentence or paragraph like a camera panning over the scene. Every element that you present should follow logically from the one before.

That's just basic writing. How is that different from AI Bust Avoidance?

 

[Trionyx]

“Uh…”
“Hmm?”
“Uh uh.”
“Uh huh.”
“Uh hmm.”
“I, uh, I wanna…”
“I’m gonna…”
“They’ll…”

And listen to how people talk. In conversations we rarely say more than a sentence or two before the other speaks. Use contractions in dialogue as people do. Microsoft Word wants me to eliminate all contractions and even in non-dialogue sections I tend to ignore most of its little blue underlines if it sounds clunky or formal.

 

[SimonDoom]

That's just basic writing. How is that different from AI Bust Avoidance?

Because that's not what AI does. AI crafts sentences that superficially appear like they follow a connecting logic, but it's obvious that the connecting glue isn't quite there.

 

[KeithD]

I'm curious that you saw my advice as imposing "limitations" or "curbing" an author's voice. I thought it was the opposite. I think the problem stems from authors using tools that stifle their own voice and make them sound generic.

Any time, including whether or not to use something like Grammarly for help, that the advice is "do this/avoid that" messes around with original voice.

Your advice on what to do to get around the AIing ax at Literotica is helpful. But that anyone has to do it to get published here is limiting their voice.

 

[Duleigh]

The number one suggestion is NOT to use true AI tools to write your story. Don't use things like ChatGPT. Just don't do it, period.

This is the very best suggestion

But that anyone has to do it to get published here is limiting their voice.

Which makes me wonder why someone would do it in the first place. It's not like anyone is being forced to write here, this is were amateur writers come to hone their skills and improve. AI is a step back, it's like learning to drive by watching YouTube videos.

 

[AwkwardlySet]

I'll just repeat what I said in the other thread. Any author should try to improve their writing, and in that sense, some of your advice is very sound and helpful. That is why I would support this thread had you named it "Simon's tips for better writing" rather than "Simon's tips on how to avoid rejection by some arbitrary tool Laurel is using." Even if someone accepts adjusting their style to suit some AI detection tool, the tips you posted are no guarantee that the story will get accepted because it's all based on speculation. Reasonable speculation, for sure, but still speculation. No one here knows how that mystical tool works.

 

[SimonDoom]

I'll just repeat what I said in the other thread. Any author should try to improve their writing, and in that sense, some of your advice is very sound and helpful. That is why I would support this thread had you named it "Simon's tips for better writing" rather than "Simon's tips on how to avoid rejection by some arbitrary tool Laurel is using." Even if someone accepts adjusting their style to suit some AI detection tool, the tips you posted are no guarantee that the story will get accepted because it's all based on speculation. Reasonable speculation, for sure, but still speculation. No one here knows how that mystical tool works.

I think you are right, but I think you and I disagree on the degree to which this is true. This is true about other topics we've discussed. I think you overstate how unresponsive and mystifying this Site is; you think I understate it. I know this: what I do works for me. I've had stories rejected two or three times, and each time I sent a courteous and reasonably detailed and clear message to Laurel, which cleared things up, and my story got published. I've never had a story rejected for AI concerns, and there are other authors here who have not encountered that problem, either, and based on sample submissions I've seen I think it's pretty clear that there are things authors CAN DO to at least minimize the problem, if not eliminate the risk of it altogether.

 

[Candy_Kane54]

2. If you use tools like Grammarly and similar tools (I use them all the time), use them to identify problems, but don't accept all of their proposed solutions. Correct clear grammatical, spelling, and punctuation problems (but see below). But DON'T regularly follow their suggestions for rewording things or for restructuring your sentences.
3. In fact, make a point of REJECTING the tool's suggestions, and add a distinctive word of your own here and there.

I've been using Grammarly since I first started in 2020 and I've never had a problem with rejection by following SD's above two suggestions. I use the problems found by Grammarly to fix any issues but I very seldom use its suggestions for rewriting sentences (usually only when it points out sentence that are passive and I can make them active). I almost never use its suggestions for replacing words that are overused unless its suggestion is a synonym that makes my meaning clearer (for some reason, Grammarly really dislikes the use of the word 'nice').

 

[StillStunned]

That's just basic writing. How is that different from AI Bust Avoidance?

Because people who learn their writing from the same informative web sources that are used to train AI don't seem to know this.

Every sample I've seen from someone whose story has been rejected seems to follow the same pattern of "piece of information, unassociated piece of information, third random piece of information."

If this thread is about providing tips to avoid sounding like AI, this is where I'd begin. And it might be basic writing to you and me, but we're not the ones getting flagged, because we know this.

 

[StillStunned]

(for some reason, Grammarly really dislikes the use of the word 'nice').

I had a professor at uni who hammered on and on about avoiding "nice" in academic writing. Even now I shy away from it.

 

[AwkwardlySet]

I think you are right, but I think you and I disagree on the degree to which this is true. This is true about other topics we've discussed. I think you overstate how unresponsive and mystifying this Site is; you think I understate it. I know this: what I do works for me. I've had stories rejected two or three times, and each time I sent a courteous and reasonably detailed and clear message to Laurel, which cleared things up, and my story got published. I've never had a story rejected for AI concerns, and there are other authors here who have not encountered that problem, either, and based on sample submissions I've seen I think it's pretty clear that there are things authors CAN DO to at least minimize the problem, if not eliminate the risk of it altogether.

That's fair. While we can have different impressions about the way the website does certain things and discuss them politely as we always have, this is about other people mostly. You posted a guide of a sort. In that sense, neither mine nor your personal experiences with the website are a proper measure of the website's responsiveness and attitude. You clearly had a very positive experience with it while I had an unsatisfactory one. We can't know which one is closer to the truth for an average website user. Yet you are the one here who is posting a guide for other users and basing it on speculation - hence my reply.
I believe I know you well enough now to be certain that you are doing this with the best possible intentions, but I felt I had to point out the fact that even if an author followed every advice you posted here to the letter, it's no guarantee that all of that work would result in their story being approved. I believe your advice is helpful but it shouldn't lead to false expectations.

 

[SimonDoom]

I had a professor at uni who hammered on and on about avoiding "nice" in academic writing. Even now I shy away from it.

Me, too. But it sounds just right in this snippet of screenplay dialogue in a great movie: "I ate his liver with some fava beans and a nice chianti."

If you replaced "nice" with another word offered by an AI, it wouldn't sound as good.

 

[forgotmyantidepressants]

Is there any hope for an author who's work already flagged the system? The stories I've already published are still getting views and I got a nice message from somebody this morning. I'd love to keep posting stories and maybe even enter a contest, but all I get are rejections. I don't even know why I'm still hanging around at this point. It really sucks.

 

[KeithD]

Reasonable speculation, for sure, but still speculation. No one here knows how that mystical tool works.

And I keep asking, without much response, whether this is something happening to new authors here rather than ones established before the AI problem became a problem. Is the site letting established authors through on this and only putting the thumbs to new author attempts? Does the site, in fact, have enough handle on how to be right most of the time on the call of what is too much AI support (whatever that is) or is it knee-jerk insulting writers on how much originality they are putting into their writing--and only challenging new submitters on it because they realize established writers were doing their thing before AI support was a thing?

 

[lustychimera]

Because people who learn their writing from the same informative web sources that are used to train AI don't seem to know this.

Fucking Brando Sando.

 

[SimonDoom]

And I keep asking, without much response, whether this is something happening to new authors here rather than ones established before the AI problem became a problem. Is the site letting established authors through on this and only putting the thumbs to new author attempts? Does the site, in fact, have enough handle on how to be right most of the time on the call of what is too much AI support (whatever that is) or is it knee-jerk insulting writers on how much originality they are putting into their writing--and only challenging new submitters on it because they realize established writers were doing their thing before AI support was a thing?

This is a good question. I'd like to hear from a) experienced Lit authors if they've encountered an AI rejection, and b) new Lit authors who have NOT encountered this problem.