"AI" Rejection

[SoCalAshley]

I did both. I messaged Laurel last Friday, and I reposted my story today with exactly those words. Hopefully it will work.

 

[BelleCanzuto]

I did both. I messaged Laurel last Friday, and I reposted my story today with exactly those words. Hopefully it will work.

I hope so too.

 

[FieryLion]

Imo having to email the site's creator and wait weeks/months for a response is probably not a valid solution, and even then you may need to do more things to get one story submission to go through. Rather than experiencing such frustration, I'd rather wait until Literotica implements a more practical system. So far (according to this thread), simply resubmitting and saying that it is not written with AI does not help and the submission will be rejected again (hence messaging Laurel seems to be the only way).

 

[HeyAll]

Just a suggesiton:

I think what can get writing in trouble is when you use use phrases like "___ and ___" and both words have the same meaning.

For example: "He was about the leave, then looked at her with pain and heartache." Or "apprehension and hestiation."

That seems to be a hallmark of AI writing, using two words with the same/similar meaning like that.

Something to consider and avoid.

 

[FieryLion]

Of course, we can change the writing to avoid tripping the AI detector. But why should we change the way we write just because some crappy AI language model also writes in a similar manner? What if AI writes erotica in more varied ways in the future, do we have to stop writing in those styles/phrasing also? Why do we have to stop writing in certain ways just to please an overzealous AI filter?

With the way the automated filter works right now, I expect past stories and stories that passed today to be taken down in the future for AI use too, as the AI filter becomes even more sensitive. As in, Lit will take down stories that have already been published in the future when the AI detector adds in new arbitrary detection rules. You could publish a story today about one of your characters having a freckle butt, and 5 years from now the AI detector decides that stories with freckle butts are AI-generated and poof, there goes your 5-year-old story. Don't believe me? Many cases in this very thread show the exact thing happening. Even better, if your story has been rejected for AI and your petition works and it gets accepted, nothing is stopping the AI filter from taking it down again in the future following an update to its code.

 

[StillStunned]

Sometimes the only way to win is not to play.

 

[SyleusSnow]

I got rejected for a third time on this story xD This time they asked what grammar checkers I used so I just mentioned Grammerly and ProWritingAid since I use them for really small issues

ProWritingAid (and, I suppose, Grammarly... I don't use business writing tools for fiction) offers to rewrite some sentences to avoid repetition, glue words, passive voice and so on. Sometimes I accept one of the suggested rewrites, but usually they ain't great, so use the choices displayed to make my own changes.

Since rewriting features like that are "AI", it's possible that accepting too many of them tips the scale on whatever crap AI detector Lit has bought into. Normal grammar checking functions for punctuation, etc. certainly shouldn't. Nor should regular use of sentence rewriting.

This AI stuff is getting silly, though. This site https://contentatscale.ai/ai-content-detector/ not only offers a detector for AI writing, but an AI-based "paraphrasing tool" to "rewrite sentences and paraphrase paragraphs in a way that is so humanlike it bypasses even the toughest and most accurate AI detection tools"!

 

[MillieDynamite]

And it's only 2500 characters at a time, so if you have a 10,000-word story, that's a lot of trips through the mill. But for the low, low price of $XXX a month, you can check 5500 characters at a time. to be honest, I don't know what it costs or how many characters it checks on the paid version. However, whatever it is, I'm not using it.

ProWritingAid (and, I suppose, Grammarly... I don't use business writing tools for fiction) offers to rewrite some sentences to avoid repetition, glue words, passive voice and so on. Sometimes I accept one of the suggested rewrites, but usually they ain't great, so use the choices displayed to make my own changes.

Since rewriting features like that are "AI", it's possible that accepting too many of them tips the scale on whatever crap AI detector Lit has bought into. Normal grammar checking functions for punctuation, etc. certainly shouldn't. Nor should regular use of sentence rewriting.

This AI stuff is getting silly, though. This site https://contentatscale.ai/ai-content-detector/ not only offers a detector for AI writing, but an AI-based "paraphrasing tool" to "rewrite sentences and paraphrase paragraphs in a way that is so humanlike it bypasses even the toughest and most accurate AI detection tools"!

 

[FieryLion]

Do not use those crappy "Ai paraphraser" services to avoid tripping the AI filter. Aside from butchering your writing, its entirely possible that the paraphasing itself will cause the filter to reject the submission. Don't waste your money.

 

[MillieDynamite]

My thoughts on AI writing tools are that AI is designed to remove money from your pocket and place it in the provider's pocket. In doing that, it is excellent on the SI scale, Scam intelligence.

 

[SoCalAshley]

Do not use those crappy "Ai paraphraser" services to avoid tripping the AI filter. Aside from butchering your writing, its entirely possible that the paraphasing itself will cause the filter to reject the submission. Don't waste your money.

So, are you suggesting that using a tool to create a better experience for your reader is “butchering your writing?” So maybe writers should not be allowed to use the thesaurus. Perhaps dictionaries should be banned on this site. Better yet, let’s ban computers and typewriters and submit handwritten manuscripts.

 

[MillieDynamite]

You're making an assumption that a software engine will write better than a human. Sure, some humans, maybe, but it's just a machine. It pulls bits and pieces from other works, reworks the words used, and vomits out a collection of words. It isn't a human, has no understanding of the emotions it's writing about, and since the owner of the engine is paid by the produced words, the AI uses as many words as possible to say what you want. I don't want AI to create my work. Then it isn't my work. If you want it to create for you, that's great, you and AI can write away.

So, are you suggesting that using a tool to create a better experience for your reader is “butchering your writing?” So maybe writers should not be allowed to use the thesaurus. Perhaps dictionaries should be banned on this site. Better yet, let’s ban computers and typewriters and submit handwritten manuscripts.

 

[FieryLion]

So, are you suggesting that using a tool to create a better experience for your reader is “butchering your writing?” So maybe writers should not be allowed to use the thesaurus. Perhaps dictionaries should be banned on this site. Better yet, let’s ban computers and typewriters and submit handwritten manuscripts.

by Ai paraphraser services, I meant the new services that claim to paraphrase your writing to no longer trip up AI text detection. To do this they will mess up your writing in whatever ways they want, as long as the writing doesn't trip AI detectors. That is what I mean by butchering your writing.

 

[SyleusSnow]

by Ai paraphraser services, I meant the new services that claim to paraphrase your writing to no longer trip up AI text detection. To do this they will mess up your writing in whatever ways they want, as long as the writing doesn't trip AI detectors. That is what I mean by butchering your writing.

Not having used one (and never will), I couldn't say how much they butcher the text. But the concept of using an AI-based tool to evade AI-based detection of AI-based writing is absurd. Almost as absurd as Lit taking the output of an AI detector at face value in the first place. Bots battling bots with all of it, as Millie says, putting money in the provider's pocket. Where does it end?

AI might suck at creative writing, but it can be pretty good at images. Let's see how OpenAI's own DALL-E does at illustrating the concept:

View attachment 2288649

 

[FieryLion]

the concept of using an AI-based tool to evade AI-based detection of AI-based writing is absurd

Imo it's not absurd, it's the only natural conclusion to this insane AI arms race. As AI language models improve and better mimic human writing, detecting it gets harder and AI detectors will falsely "detect" more human writing as AI text, and to avoid this you have to rely on AI models that are trained specifically to avoid detection. This is why relying on AI detectors in the first place is so disastrous. Why would you want to submit your work to a place like Lit that will, at any moment from now to the far future, deem your already submitted work as AI-generated and remove it with you being none the wiser?

Not having used one (and never will),

I fear that while we all say this at first, imagine the frustration of having your submission rejected 3+ times due to some bullcrap AI-generated detection. Anyone can go back on their word and use a service to twist and alter their own original writing to some cursed text produced by deeply flawed AI models designed to circumvent other deeply flawed AI detectors.

Ironically, the prolific use of automated AI detection will lead to the extinction of genuine human-created content because that content will be falsely flagged, and to pass these cursed filters one will have to make use of AI to alter their own human-made original creations. In this worst-case scenario where automated AI-detectors are rampant everywhere on the internet, one will always feel anxious when posting anything online because it can always be falsely flagged as ai-generated. So to ease this anxiety one can use an AI service to alter the original writing in a way that will not trigger such AI filters, and the writing is hence stripped of its human-made originality.

 

[Beholders_Eye]

The future of writing:

 

[SyleusSnow]

Ironically, the prolific use of automated AI detection will lead to the extinction of genuine human-created content because that content will be falsely flagged, and to pass these cursed filters one will have to make use of AI to alter their own human-made original creations. In this worst-case scenario where automated AI-detectors are rampant everywhere on the internet, one will always feel anxious when posting anything online because it can always be falsely flagged as ai-generated. So to ease this anxiety one can use an AI service to alter the original writing in a way that will not trigger such AI filters, and the writing is hence stripped of its human-made originality.

Electronic calculators were predicted to inevitably lead to no one being able to do math, and spell-checkers would mean no one could spell. Going back further, radio was the demise of books and newspapers, recordable cassette tapes the end of the music industry, and so on.

It's all part of the hype cycle that happens every time there's some technical innovation. Excitement is matched by dire predictions and pearl-clutching overcorrections... the perceived need for AI detectors and companies rushing them to market despite their inaccuracy part of that.

If Lit loses writers because of a shit AI detection service, at some point they'll ditch it or fix it... or writers will go elsewhere. Amazon, the kingpin of shit writing, tried AI detection and backpedaled quickly. And they learned... now they only require authors to declare they used AI, and have distinguished between "AI generated" (AI is used to create actual content... scammers were flooding KDP with nonsensical chatgpt crap) and "AI assisted" (AI was used to brainstorm or refine writing.) They don't care about AI assisted. I suspect Lit will quickly learn to follow suit.

 

[MillieDynamite]

I suspect that my publisher will be marking that box of AI-assisted at Amazon if only because sometimes his cover designer uses AI to generate or agument covers. I don't know if he has declared it on any yet or not. I know two recent covers were at least enhanced by AI. But to be 100 percent honest, I can't actually tell if they have been created or improved by AI.

Electronic calculators were predicted to inevitably lead to no one being able to do math, and spell-checkers would mean no one could spell. Going back further, radio was the demise of books and newspapers, recordable cassette tapes the end of the music industry, and so on.

It's all part of the hype cycle that happens every time there's some technical innovation. Excitement is matched by dire predictions and pearl-clutching overcorrections... the perceived need for AI detectors and companies rushing them to market despite their inaccuracy part of that.

If Lit loses writers because of a shit AI detection service, at some point they'll ditch it or fix it... or writers will go elsewhere. Amazon, the kingpin of shit writing, tried AI detection and backpedaled quickly. And they learned... now they only require authors to declare they used AI, and have distinguished between "AI generated" (AI is used to create actual content... scammers were flooding KDP with nonsensical chatgpt crap) and "AI assisted" (AI was used to brainstorm or refine writing.) They don't care about AI assisted. I suspect Lit will quickly learn to follow suit.

 

[SyleusSnow]

I suspect that my publisher will be marking that box of AI-assisted at Amazon if only because sometimes his cover designer uses AI to generate or agument covers. I don't know if he has declared it on any yet or not. I know two recent covers were at least enhanced by AI. But to be 100 percent honest, I can't actually tell if they have been created or improved by AI.

Huh. Reading the fine print, I see Amazon includes "AI-generated images include cover and interior images and artwork" in their scope.

Again though they don't care about "AI assisted." Good thing, otherwise every designer using Photoshop with the new generative fill and expand features would have to declare.

A cover art service I've been eyeballing (goonwrite.com) now generate all their covers using Midjourney... way beyond "assisted"... so would need to be declared.

 

[MillieDynamite]

I don't know what she uses. I just know she has great covers.

Huh. Reading the fine print, I see Amazon includes "AI-generated images include cover and interior images and artwork" in their scope.

Again though they don't care about "AI assisted." Good thing, otherwise every designer using Photoshop with the new generative fill and expand features would have to declare.

A cover art service I've been eyeballing (goonwrite.com) now generate all their covers using Midjourney... way beyond "assisted"... so would need to be declared.

 

[u1u1u1]

Sigh...
Submitted the next part of my ongoing saga. Had used a couple of AI detectors (Sapling and GPTZero) to check my work. They helped me detect a few areas where I could tighten things up, but after doing so, they still detected my story as largely AI generated.
I included a pre-emptive note on my submission saying "I wrote this, I am the copyright owner, I am not an AI" but it still got rejected.

This is the third part in a trilogy, it's almost 20,000 words long, I've been keeping track of character development, descriptions, preferences, maintaining a consistent style and tone, and writing graphic BDSM scenes (which AIs will run away from!) where the physical configurations of characters and objects are very important. I'd be shocked if an AI could generate this stuff. My experience is that it only takes ChatGPT about 10 messages to start forgetting key details, it would have no chance of remembering a character's eye colour or favourite tipple over 20,000 words.

My working theory is that using ChatGPT has led to me picking up some of its mannerisms without noticing. But I'm really questioning myself now. Does this (part 2) really read as AI generated? Am I that bad at writing?

 

[SyleusSnow]

Not that we should have to change our writing styles, but some of the detection services highlight exactly what they think looks AI generated. This one for example: https://contentatscale.ai/ai-content-detector/

 

[MillieDynamite]

The problem with the free checkers is the 2500-character limit at one time. What checks as human in small chunks might check as AI written when everything is in aggregate.

Not that we should have to change our writing styles, but some of the detection services highlight exactly what they think looks AI generated. This one for example: https://contentatscale.ai/ai-content-detector/

 

[u1u1u1]

I have accidentally stumbled upon a major revelation. I was using Sapling.ai to check my WIP. An excerpt was showing an 81% chance of being fake. I tried inserting a new sentence, initially hitting space before considering what to write... and suddenly the probability shot up to 99.9%.
I went through the whole excerpt and removed three "trailing spaces" at the end of paragraphs. Removing those three spaces caused the probability to drop from 99.9% fake to a mere 21%. That's still 21% higher than any other detectors, but still. I sometimes instinctively throw two spaces at the end of paragraphs because some sites (most notably Reddit) will not display line breaks unless you do. I'm now wondering if I could use that same technique on my previous story which is back for review...

 

[MillieDynamite]

Multiple spaces between words are anomalies that humans don't often use. So perhaps there's your answer.

I have accidentally stumbled upon a major revelation. I was using Sapling.ai to check my WIP. An excerpt was showing an 81% chance of being fake. I tried inserting a new sentence, initially hitting space before considering what to write... and suddenly the probability shot up to 99.9%.
I went through the whole excerpt and removed three "trailing spaces" at the end of paragraphs. Removing those three spaces caused the probability to drop from 99.9% fake to a mere 21%. That's still 21% higher than any other detectors, but still. I sometimes instinctively throw two spaces at the end of paragraphs because some sites (most notably Reddit) will not display line breaks unless you do. I'm now wondering if I could use that same technique on my previous story which is back for review...