Practical advice for dealing with AI rejection.

[Awbb]

In my experience, when I was messing about with ChatGPT, it overloads passages with what it refers to as "emotional shorthand". Stuff like "Her stomach flipped" or "His throat tightened". There's nothing wrong with that (other than I think it's appalling writing) but it's everywhere. It's a default mechanism for emotion, regardless of character, scene or tone. It's simply "emotion happens here" without having to worry about a specific physical or behavioural consequence.

It's partial to one-line stingers too. A thoroughly unnecessary "And for the first time, they could see their life clearly before them." at the end of a chapter is a regular feature. Ultimately it's a pseudo-profound summary sentence that (usually) doesn't arise from the scene. It tries to create a manufactured sense of significance to what could be a thoroughly mundane ending (see all my stories, mundane endings are a particular speciality).

As has been mentioned. Em dashes. It loves em dashes. Again, nothing wrong with an em dash here or there, but every paragraph? I don't know why it does this. It's fond of a staccato list (see below) when it wants, but it uses an em dash as its preferred punctuation. Don't get me wrong, I have my own punctuation habits (the em dash being an absolute no-no, for no reason other than I was never taught about it at school and probably use the princely semi-colon less than I should).

And the most vomit inducing of all: The "Not X. Not Y but Cabbage" phrase. "Time passed. Not fast. Not slow. But resolutely." What? Sometimes it makes more sense than that, but not often. Pass the bucket.

Obviously this is only ChatGPT, I've heard that other LLMs are available and will probably have their own nonsense "tells".

Most importantly, I couldn't begin to tell you if Lit's AI filter would ping on any of the things I've mentioned above.

Glad to be of (time-wasted) service.

 

[jsmiam]

Tip: Stories are made up of paragraphs. Paragraphs are made up of sentences. Profound, right?

Autocomplete at the sentence completion level, means you could inadvertently have dozens of sentences where the second half or closing words of that sentence really were written by AI. Turn it off.

 

[AwkwardlySet]

Using a tool or service that generates or modifies text using a large language model

From what I can tell, very nearly every single post about AI rejections that we see here are from people who have either:

• Used Grammarly or Quillbot or some other writing tool to "check their grammar," which turns out to mean they let those tools rewrite sentences and words for them
• Used translation software to translate from their home language into English
• Used Word's copilot writing tools to correct for grammar or other mechanical elements

People don't think of those tools as AI, but they are. Every single one of them now uses an LLM to function. Any time you let a tool modify your text, it's mechanically not any different from letting ChatGPT generate text for you.

And the part about it that sucks is that if you used one of these tools from the very beginning throughout your writing process, there's not really anything you can do to undo it, there's no "original" version that you can revert to.

So of course people get mad and frustrated about that, because the only answer we can give them is "write a new story from scratch and don't use tools."

Yes, this is all quite obvious, but also largely irrelevant for this thread. I'll explain why.

This thread is for authors whose stories were already rejected by the site. Those stories are already written. And whether the author used Grammarly's or ProWritingAid's suggestions about rewriting sentences, or they used AI to fully generate the story, or they simply wrote the story in a style that triggered the rejection, is now pretty much irrelevant.

Your advice is excellent for all those who are starting to write their story. It fits perfectly as pre-writing advice for all new authors. But when the story is already written and rejected, we must deal with the things that triggered the rejection. And the triggers are specific sentence constructions, key words, dialogue especially, etc. They might have been put there by the AI or by the author's own style of writing, fuck knows. In that sense, the guide doesn't make a difference.

So as you see, we aren't dealing with the cause, as that doesn't help the authors at this point. They hardly have the keys to a DeLorean to go back in time and tell their younger self not to use Grammarly suggestions, or to simply avoid writing the constructions that look like something that the AI might write. The guide is all about fixing the already existing story.

Now, I can already see people claiming that, along with legit authors, the guide might help "the cheaters" and potentially make their AI-generated story pass the screening algorithm.

I beg to differ. I believe that all of them who truly wrote their story, and even those who mostly wrote their story but foolishly followed Grammarly's advice to rewrite, will have the writing capacity to rewrite their story in the way the guide advises, and hopefully pass the screening, while true cheaters will lack the skill to do all that. It's not just a matter of changing a word or two, but of rewriting dialogue, whole sentences and constructions.

The only thing the guide doesn't help with is the frustration of the honest authors, and the fact that they must change their potentially legit writing style just so some algorithm wouldn't flag them. That can't be helped since we don't make the rules or policies on Lit. But if they want to be published here, they'll need to go for it, and only they can answer for themselves whether that's worth it or not.

 

[Bazzle]

My work is overbrimming with these and I’ve never had a problem.

thats because you write with a dash of Em?

 

[PennyThompson]

Yes, this is all quite obvious, but also largely irrelevant for this thread. I'll explain why.

This thread is for authors whose stories were already rejected by the site. Those stories are already written. And whether the author used Grammarly's or ProWritingAid's suggestions about rewriting sentences, or they used AI to fully generate the story, or they simply wrote the story in a style that triggered the rejection, is now pretty much irrelevant.

Your advice is excellent for all those who are starting to write their story. It fits perfectly as pre-writing advice for all new authors. But when the story is already written and rejected, we must deal with the things that triggered the rejection. And the triggers are specific sentence constructions, key words, dialogue especially, etc. They might have been put there by the AI or by the author's own style of writing, fuck knows. In that sense, the guide doesn't make a difference.

So as you see, we aren't dealing with the cause, as that doesn't help the authors at this point. They hardly have the keys to a DeLorean to go back in time and tell their younger self not to use Grammarly suggestions, or to simply avoid writing the constructions that look like something that the AI might write. The guide is all about fixing the already existing story.

Now, I can already see people claiming that, along with legit authors, the guide might help "the cheaters" and potentially make their AI-generated story pass the screening algorithm.

I beg to differ. I believe that all of them who truly wrote their story, and even those who mostly wrote their story but foolishly followed Grammarly's advice to rewrite, will have the writing capacity to rewrite their story in the way the guide advises, and hopefully pass the screening, while true cheaters will lack the skill to do all that. It's not just a matter of changing a word or two, but of rewriting dialogue, whole sentences and constructions.

The only thing the guide doesn't help with is the frustration of the honest authors, and the fact that they must change their potentially legit writing style just so some algorithm wouldn't flag them. That can't be helped since we don't make the rules or policies on Lit. But if they want to be published here, they'll need to go for it, and only they can answer for themselves whether that's worth it or not.

I totally get all that, and I can only try to empathize with what it feels like to get caught up in AI purgatory

I don't think most of them intentionally did anything wrong, they're victims of predatory AI speculation by companies that keep pushing these tools into every possible workflow because they don't know how to turn a profit with them and are over-leveraged.

But I don't believe nebulous writing style elements are what triggers Lit's AI detector, and I don't believe that bending and butchering a writer's style to meet someone else's guesses about what "sounds human," is effective at "untriggering."

 

[TheLobster]

I believe that all of them who truly wrote their story, and even those who mostly wrote their story but foolishly followed Grammarly's advice to rewrite, will have the writing capacity to rewrite their story in the way the guide advises, and hopefully pass the screening, (...)

Absolutely. But this also means that @PennyThompson's advice is actually rather useful to those authors, as they can put it into practice when they attempt their second draft and so increase the likelihood of it eventually passing through the filter.

Assuming they care enough about their story to write it more than once, that is.

 

[AwkwardlySet]

I totally get all that, and I can only try to empathize with what it feels like to get caught up in AI purgatory

But I don't believe nebulous writing style elements are what triggers Lit's AI detector, and I don't believe that bending and butchering a writer's style to meet someone else's guesses about what "sounds human," is effective at "untriggering."

There are people who said that their story was first rejected, and then they rewrote some parts, and then it went through. Who knows.

Either way, it's the best we can offer to all the frustrated authors. It's certainly better than ignoring their posts or telling them that their writing style sucks anyway. It's something.

 

[PennyThompson]

There are people who said that their story was first rejected, and then they rewrote some parts, and then it went through. Who knows.

Either way, it's the best we can offer to all the frustrated authors. It's certainly better than ignoring their posts or telling them that their writing style sucks anyway. It's something.

Building replicas of radio transmitters out of coconuts and bamboo is doing something too, but it's not strongly correlated with how many airplanes fly overhead

 

[AwkwardlySet]

Absolutely. But this also means that @PennyThompson's is actually rather useful to those authors, as they can put it into practice when they attempt their second draft and this increase the likelihood of it eventually passing through the filter.

Assuming they care enough about their story to write it more than once, that is.

Sure. We can include that as something to pay attention to in their future writing projects, or make a pre-writing guide too, and then link it for good measure.

 

[AwkwardlySet]

Building replicas of radio transmitters out of coconuts and bamboo is doing something too, but it's not strongly correlated with how many airplanes fly overhead

Right.
No one is forcing you to be in this thread. You stated your beliefs, and that's all fine, and I'm sure we can add all the advice about Grammarly and PWA here. But you seem to be more focused on shooting down our approach and advice than anything else. To be honest, I don't see where your confidence and expertise in this sense come from. And it has to be that, if you feel like mocking our approach.

If you have a better idea on how to help all the rejected authors and their already written stories, then put it forward, and we'll all chime in. Otherwise, you're just trolling.

 

[PennyThompson]

If you have a better idea on how to help all the rejected authors and their already written stories, then put it forward, and we'll all chime in.

Rewrite the story from scratch and reject the LLM tools. Which, I 100% understand is a painful answer to give someone, and I 100% understand why people desperately want a different answer

I'll leave the thread now, because I don't really have anything else to say about it. I feel people's frustration, and it makes me sad, and it's nobody's fault but the tech companies.

 

[crookedletter]

Rewrite the story from scratch and reject the LLM tools. Which, I 100% understand is a painful answer to give someone, and I 100% understand why people desperately want a different answer

I'll leave the thread now, because I don't really have anything else to say about it. I feel people's frustration, and it makes me sad, and it's nobody's fault but the tech companies.

Applying a whole new writing style to a story is going to require a rewrite from scratch anyway. I would say this is a useful suggestion for a starting point. It might actually be less painful than a rewrite applying all the other guidelines laid out in this thread.

 

[EmilyMiller]

Too much wishful thinking on this thread. Humans always want a solution - ideally one with minimal effort - I’ll leave as well as I feel nixing em dashes is giving the poor people who are stuck false hope.

 

[Britva415]

I'm not going to quote the several people whose comments scream for the following habit:

Version control.

If you save the draft before a copy of it is revised, you don't have to strip anything out of it later. If you get rejected after revision by AI, going back to the unaltered copy is practical advice for dealing with AI rejection.

Don't erase or alter the early drafts. Changes should be done on a copy. AI isn't even the only or the best reason for keeping version history.

 

[BobbyBrandt]

I thought that it might be beneficial to share an actual AI-rejection success story. This is from a feedback e-mail that I received last night:

"Hi,

You provided a comment to my posting about my chapter revisions being rejected for AI after a few months pending. You suggested it was due to the length of chapters throwing the percentage off, so I had the site delete the chapters, combined them alltogether (21 lit pages) and resubmitted. It was published the very next night after submitting.

Thanks so much for your suggestion. You were spot on and it was the perfect solution in my case.

MrC"

This is obviously not a solution for all cases, but it does offer a glimmer of hope for those who have had only parts of their work rejected.

 

[RocketGrunt]

I have a belief that some word processor addons are leaving computer gibberish in the files that automated AI detection is getting triggered by. I would suggest saving as plain text if your word processor is getting to handsy with your document. I'm using a pretty no frills OpenOffice, but enough people with these complaints mention that they write in Word that I think that might have something to do with it. Certainly some of the Microsoft Office features right now are giving Co-Pilot suggestions and shit, and if there's Co-Pilot metatext in your file it would obviously be detected as an AI document immediately by even the softest AI scan.

 

[BobbyBrandt]

I have a belief that some word processor addons are leaving computer gibberish in the files that automated AI detection is getting triggered by. I would suggest saving as plain text if your word processor is getting to handsy with your document. I'm using a pretty no frills OpenOffice, but enough people with these complaints mention that they write in Word that I think that might have something to do with it. Certainly some of the Microsoft Office features right now are giving Co-Pilot suggestions and shit, and if there's Co-Pilot metatext in your file it would obviously be detected as an AI document immediately by even the softest AI scan.

I have used MS Word for decades, and view it as an advantage over plain text applications specifically due to the metadata embedded in the documents. The metadata can be used to demonstrate authenticity of the writing.

Then, if a story is rejected for suspected AI, the metadata can be reviewed to identify where Grammarly or Word itself made any changes. It can be burdensome, but far less so than rewriting an entire story.

 

[RocketGrunt]

I have used MS Word for decades, and view it as an advantage over plain text applications specifically due to the metadata embedded in the documents. The metadata can be used to demonstrate authenticity of the writing.

Then, if a story is rejected for suspected AI, the metadata can be reviewed to identify where Grammarly or Word itself made any changes. It can be burdensome, but far less so than rewriting an entire story.

MS Word for the last 18 months or so has been a pretty different animal than it was decades ago or even two years ago. I think it's putting a lot more "AI Was Here" crap into peoples' manuscripts than it used to.

I'm not positive this is it, but the substantial number of people writing in MS Word who are suddenly getting AI rejections leads me to believe this is the case. There might be a secret setting you can make that keeps it from happening, but I don't know what it is or where it might be. Microsoft has been pushing Co-Pilot slop into things HARD. And I really think that the simplest answer for people getting rejected for AI generated material whose texts have no AI generated material is that their files are getting AI generated metadata that they can't see.

 

[Writer61]

MS Word for the last 18 months or so has been a pretty different animal than it was decades ago or even two years ago. I think it's putting a lot more "AI Was Here" crap into peoples' manuscripts than it used to.

Although I write in Word (specifically the latest version of Word for Mac under a 365 subscription), I have switched to generating an RTF version for uploading to Lit.

 

[RocketGrunt]

It's a comedy bit, but it kind of hints toward what I was talking about. I think the current Microsoft Office Co-Pilot insertions are genuinely inserting AI "stuff" into peoples' manuscripts that the AI detector is detecting, even when the actual writing is being done like they were using WordPad.

 

[BobbyBrandt]

MS Word for the last 18 months or so has been a pretty different animal than it was decades ago or even two years ago. I think it's putting a lot more "AI Was Here" crap into peoples' manuscripts than it used to.

I'm not positive this is it, but the substantial number of people writing in MS Word who are suddenly getting AI rejections leads me to believe this is the case. There might be a secret setting you can make that keeps it from happening, but I don't know what it is or where it might be. Microsoft has been pushing Co-Pilot slop into things HARD. And I really think that the simplest answer for people getting rejected for AI generated material whose texts have no AI generated material is that their files are getting AI generated metadata that they can't see.

I can't speak on anything later than a locally installed version of MS Word 2019, which is what I currently use. Windows 11 keeps trying to add Copilot and other features that I disable to any extent possible, and it doesn't seem to be affecting anything in my Word setup, so far.

 

[BobbyBrandt]

I think the current Microsoft Office Co-Pilot insertions are genuinely inserting AI "stuff" into peoples' manuscripts that the AI detector is detecting

If I'm not mistaken, Microsoft's Copilot exists at the Operating System (OS) level, essentially replacing the "Cortana" feature that was released with Windows 10.

Copilot doesn't interface with all Windows applications, but you are correct in stating that virtually all Microsoft Office applications can be affected by it if the user isn't careful or neglects to disable it. Personally, I keep it turned off.

 

[MillieDynamite]

Excellent suggestions from an exceptional writer.

Some practical advice based on what has been working for me so far.... Other may disagree, but I have yet to see anyone answer this question with concrete examples, and no guarantees. My next story could get kicked back, too...

1. Vary your sentence length. Short is good. Sometimes, though, you need to go longer, espouse on what you're describing, dig into the feeling of the moment to get to the core of the character's motivation. Consistent sentence length is not only boring, it's what AI does, so don't do that. Got it?
2. Inject some personality, some emotion into your writing. "Tom hated his job and only stayed for the money." is flat, boring, mechanical. "The last fifteen minutes felt like hours as Tom watched the clock, waiting for five o'clock and his escape from the hellhole he called his job, cursing the bi-weekly check that shackled him to the tedium." says the same thing but paints a deeper picture of what Tom feels.
3. Use big words, or as @EmilyMiller might say, sesquipedalienate(is that a word?). View attachment 2591428Get creative with metaphors and other figures of speech. Use Similes and personification, hyperbole, irony, euphemisms. Throw in a pun or two. Add an oxymoron. AI doesn't do these things.
4. When it comes to grammar, use proper punctuation and spelling, but break all the other rules. Use. One. Word. Sentences. for emphasis, use dialect, y'all; do stuff like that. In dialog, remember people don't always talk in complete thoughts. Get creative with dialog tags, too.
   "Sometimes they, um, uh, you know, they, um, sort of, shit..." Derick shook his head, frustrated that the words just disappeared.
   "Lose track of what they're trying to say?" Emily chucked, showing her amusement at his discomfort.
   "Fuck you." He shook his head trying not to laugh with her.
   "Later." She winked.
   Notice none of my tags are "he said" or "she said" but you know exactly who is speaking.