My story gets flagged as being written by AI

[JuliusKai]

I recently submitted 3 chapters of my first ever story and everything went well, all three chapters were accepted and are now available. However, the first chapter of my second story was rejected.

At first I didn't pay enough attention to the message about why it was rejected, because I was kinda anticipating a rejection, because the way it was initially it was very noncon (I was planning on going more towards reluctant in later chapters). I did my edits and made it a bit less dark.

To my great surprise it was rejected again and now paying closer attention I realised that my story was deemed to be written by an AI. Full disclosure, I do use AI tools to help with grammar and especially punctuation (English is my second language). Sometimes I ask ChatGPT for suggestions about flow and pacing, a list of potential names for characters or places, if I should be more or less descriptive, that kinda thing. What I don't do is use any AI generated text or ideas. Everything I write is my own, based on my own ideas (Even names from a list I tend to modify because I rarely like the exact ones ChatGPT suggests).

After being rejected twice, I resubmitted my chapter unchanged, in hopes that my explanation will suffice (still pending), I do have a very rough first draft available if need be.

What got me curious is, when I found one of the online checkers I submitted all of my stories and the rejected chapter got flagged as AI, while the accepted chapters were certified completely human. I used the exact same process to write my stories, so now I don't know what precautions I should take to not get my final edited chapter flagged as AI. Does anyone have any ideas? Is the only way to be safe is by writing with no assistance at all?

 

[Bamagan]

This is becoming such a common thread lately.
The simplest advice is probably not to trust chatbots to have a good grasp of flow and pacing, as those are two of the things that are most likely to be flagged as artificially generated. Asking it for lists of names or places is unlikely to be a problem, but taking its advice about being more or less descriptive might potentially make your work seem less 'organic,' especially if it advises you on what to cut or add.
Best thing to do for now is wait to see if your explanation is accepted and rely less on chatbots as beta readers in the future.

 

[MrPixel]

This is becoming such a common thread lately.

It is. My suspicion is that the online grammar checkers are getting too smart for their own good, and are bending prose to their will with a mechanical consistency.

OTOH, it should be considered that TPTB (the powers that be) are using an AI content checker that flunks perfect grammar. Language pedants beware!

 

[JuliusKai]

This is becoming such a common thread lately.
The simplest advice is probably not to trust chatbots to have a good grasp of flow and pacing, as those are two of the things that are most likely to be flagged as artificially generated. Asking it for lists of names or places is unlikely to be a problem, but taking its advice about being more or less descriptive might potentially make your work seem less 'organic,' especially if it advises you on what to cut or add.
Best thing to do for now is wait to see if your explanation is accepted and rely less on chatbots as beta readers in the future.

Thanks for the advice, when it comes to descriptiveness I tend to rush things along to progress the plot, because that's the way I read, I often catch myself skipping entire paragraphs or even sections if they go too much into detail. So I felt like it was a good way to check myself. Probably should make some kind of checklist instead.

 

[nice90sguy]

It's a bit ironic that humans are using AI tools to detect AI tools.

 

[Bamagan]

It is. My suspicion is that the online grammar checkers are getting too smart for their own good, and are bending prose to their will with a mechanical consistency.

OTOH, it should be considered that TPTB (the powers that be) are using an AI content checker that flunks perfect grammar. Language pedants beware!

To err is human, to forgive divine. To weld college-level vocabulary to fifth-grade story skeletons is AI.

 

[12oclocktales]

It's a bit ironic that humans are using AI tools to detect AI tools.

Pretty soon AI detection tools will be rejecting stories as written by AI out of sheer jealousy: my AI is so much better than that one! It will be the beginning of the literary bot wars.

 

[StillStunned]

It will be the beginning of the literary bot wars.

Robot Wars, devolved into a poetry slam.

 

[Bamagan]

Pretty soon AI detection tools will be rejecting stories as written by AI out of sheer jealousy: my AI is so much better than that one! It will be the beginning of the literary bot wars.

The Autocorrects wage their battles to destroy the evil forces of the Descripticons.

 

[OverconfidentSarcasm]

Here's the advice that seems to work for whatever crap detection software Lit seems to use:
- Consider using "though" instead of "however".
- Avoid "the" and "that" to make it sound less grammatically correct, and more like actual spoken words. Example:

"I resubmitted my chapter unchanged, in hopes that my explanation will suffice..."
to
"I resubmitted my chapter unchanged, hoping my explanation will suffice..."

 

[JuliusKai]

Here's the advice that seems to work for whatever crap detection software Lit seems to use:
- Consider using "though" instead of "however".
- Avoid "the" and "that" to make it sound less grammatically correct, and more like actual spoken words. Example:

"I resubmitted my chapter unchanged, in hopes that my explanation will suffice..."
to
"I resubmitted my chapter unchanged, hoping my explanation will suffice..."

Thanks, sounds like something worth trying

 

[Altissimus]

Here's the advice that seems to work for whatever crap detection software Lit seems to use:
- Consider using "though" instead of "however".
- Avoid "the" and "that" to make it sound less grammatically correct, and more like actual spoken words. Example:

"I resubmitted my chapter unchanged, in hopes that my explanation will suffice..."
to
"I resubmitted my chapter unchanged, hoping my explanation will suffice..."

Ugh. I'm not changing my grammar to avoid sounding like an AI.

I doubt I'll have any problems, though. AI doesn't use any sentence fragments or run-on sentences, and I've got both of those nailed.

(In before someone else mentions my propensity for the semi-colon.)

Actually, that's the answer: just add semi-colons everywhere.

 

[MillieDynamite]

AI, I've been telling you guys for months now: AI's out to destroy us. Beware, the end is near! Or, maybe, the end is nigh! Let's ask AI which we should use here.

 

[Devinter]

To weld college-level vocabulary to fifth-grade story skeletons is AI.

I wonder if this is why so many of these "my story has been flagged as AI" seems to come from people who's first language isn't English - because I imagine that this group of writers use Google and dictionaries to find synonyms for words at a greater frequency than a native speaker, and often end up picking a fancy-sounding word that they perhaps would never use in everyday speech. I've certainly done that many times, and thus perhaps if my 'story skeleton' was weaker I'd also be falsely flagged for throwing in words like "staccato" into a story, which I only know with the help of Google. So a non-native English writer might therefore write more like the AI would? Food for thought.

 

[Bamagan]

I wonder if this is why so many of these "my story has been flagged as AI" seems to come from people who's first language isn't English - because I imagine that this group of writers use Google and dictionaries to find synonyms for words at a greater frequency than a native speaker, and often end up picking a fancy-sounding word that they perhaps would never use in everyday speech. I've certainly done that many times, and thus perhaps if my 'story skeleton' was weaker I'd also be falsely flagged for throwing in words like "staccato" into a story, which I only know with the help of Google. So a non-native English writer might therefore write more like the AI would? Food for thought.

ESL writers are also likely to have a detailed knowledge of formal English grammar as opposed to colloquial grammar, which means their dialogue is likely to seem somewhat off to native speakers, at least in many contexts. AI is in the same boat at the moment, I think, which contributes to false flagging. It's better than average humans at being technical, but inexperienced with being 'casual' or whatever you'd like to call it.

 

[StillStunned]

ESL writers are also likely to have a detailed knowledge of formal English grammar as opposed to colloquial grammar, which means their dialogue is likely to seem somewhat off to native speakers, at least in many contexts. AI is in the same boat at the moment, I think, which contributes to false flagging. It's better than average humans at being technical, but inexperienced with being 'casual' or whatever you'd like to call it.

I wrote 19k words in a very formal, stylised language to simulate a 19th century Gothic/Ruritanian novel, and it was approved without any issue.

 

[Bamagan]

I wrote 19k words in a very formal, stylised language to simulate a 19th century Gothic/Ruritanian novel, and it was approved without any issue.

Okay? I'm not sure anyone has ever asked a chatbot to attempt that, but I'd be surprised if one would succeed without a lot of coaching and reprompting.

 

[OverconfidentSarcasm]

I wonder if this is why so many of these "my story has been flagged as AI" seems to come from people who's first language isn't English

Because those assisting AIs like Grammarly are all geared towards UK English, and that is the version of the English language that most of the world learns. Native English speakers are able to recognize when that AI makes your sentences sound weird or unnecessarily fancy. The rest of the users just accept the suggestions without questioning them.

 

[MillieDynamite]

Gammarly is geared to American English on the North American Continent.

 

[SimonDoom]

If we've gotten to the point that we actually have to think about writing so as to avoid coming across as AI, something is wrong.

SkyNet is becoming self-aware.

 

[alohadave]

Because those assisting AIs like Grammarly are all geared towards UK English, and that is the version of the English language that most of the world learns. Native English speakers are able to recognize when that AI makes your sentences sound weird or unnecessarily fancy. The rest of the users just accept the suggestions without questioning them.

Something that many people don't know is that Grammarly is geared to business communications, not prose or dialogue. So it'll suggest more formal changes than what we may want as fiction writers.

 

[ElectricBlue]

If we've gotten to the point that we actually have to think about writing so as to avoid coming across as AI, something is wrong.

SkyNet is becoming self-aware.

Will it be able to finish stories, though?

Just askin', for a friend .

 

[SimonDoom]

Will it be able to finish stories, though?

Just askin', for a friend .

"A friend." Uh huh.

My next story will be live in about 7 hours.

 

[ElectricBlue]

"A friend." Uh huh.

My next story will be live in about 7 hours.

Good man .

 

[Rustyoznail]

Gammarly is geared to American English on the North American Continent.

Gammarly is geared to check out and critique ladies' legs... Like in your pic.

I'll head back to work now.