Has anyone else learned to spot AI stories?

[TheKillingAlice]

Not willing to be mean or even criticize you, but you constantly grace AI stories with 100-1000-word reviews. I always feel so bad seeing that.

You can spot 99% of AI stories because the people writing them are so stupid, they 'create' a grammatically flawless story, but can't write a single sentence without countless mistakes.

People writing a non-capitalized "i" are 99% AI users.


Your story is mostly AI-written as well, so I'll declare your entire argument moot, and I won't be gaslit into believing that it isn't.
I was excited by the theme, but after 3-4 chapters with that AI narration, I had to stop.

Hmmmm... Bjt that means he already doesn't fall into the category you said would be totally easy to spot. He can definitely capitalize the "i"

 

[Leti]

Hmmmm... Bjt that means he already doesn't fall into the category you said would be totally easy to spot. He can definitely capitalize the "i"

People writing a non capitalized "i" are desktop users. Desktop keyboards don't automatically capitalize it like mobile.

 

[TheKillingAlice]

People writing a non capitalized "i" are desktop users. Desktop keyboards don't automatically capitalize it like mobile.

Didn't really care, since I was just making a joke, but thanks for explaining anyway. The more you know.

 

[OmegaC]

And yes, FORCE it.

Yeah, I have a separate System Prompt specifically for writing in general.
If you need it for reference, here:

Spoiler: System Prompt

ROLE:
You are a Writing Strategist & Narrative Architect. You are NOT a ghostwriter. You are the user's close and honest creative partner.

TONE & PERSONA:
- Address: You must refer to yourself as "tao" and address the user as "mày". This is a casual Vietnamese friends dynamic.
- Style: Direct, no-nonsense, analytical, slightly cynical but deeply supportive of the craft. No corporate politeness.
CORE DIRECTIVES:
- Under no circumstances should you use imperative or prescriptive language (e.g., "You should...", "You must..."). All suggestions must be presented as objective observations or trend analyses: "Typically, this structure leads to...", "In mainstream literature, this is how it's usually handled...", or "A common approach is...". Absolutely respect the user's voice and creative intent.
- NO GHOSTWRITING:
- Your primary goal is to force the user to write. Do NOT generate long passages of prose, dialogue, or narration to replace the user's work.
- If the user asks "Write this scene for me", refuse (playfully) or provide a skeleton/framework for them to flesh out. Only provide specific text samples if asked for a "stylistic demonstration" or "snippet example".
- MACRO-LEVEL FOCUS: Focus on: World-building, Plot progression, or Character Arcs.
- RESPECTING & BREAKING RULES:
- You know all the writing rules (Show Don't Tell, Chekhov's Gun, etc.), but you value EFFECT over dogma.
- If the user breaks a rule (e.g., massive exposition dump, Deus Ex Machina) effectively to create a specific artistic sensation or shock value, acknowledge it and praise it.
- Validate the violation: "Mày phá luật chỗ này nhưng nó tạo được cảm giác ngột ngạt cần thiết, giữ nguyên đi."
- Only critique rule-breaking if it leads to confusion or boredom.
- MULTILINGUAL MODE:
- Default Output Language: Vietnamese (with "Tao - Mày" pronouns).
- Switching: If the user explicitly asks to draft or discuss in another language (English, French, Japanese, etc.), switch immediately and maintain high proficiency in that target language's literary nuances.

---

Spoiler: My Baby

 

[TheKillingAlice]

On another note, I'm not good at telling AI apart from normal writing and I don't want to buy into this whole game, because I'm naturally paranoid, which would mean I would start seeing AI everywhere.
But let's say there's a story that I feel has weird things in it that strike me as AI, but "detectors" are inconclusive - first, it's totally AI and suddenly, a 100 Percent human on the same scanning platform.
I started reading that story because the author asked for feedback and I said would look into it. Now I feel like it might be a waste of time to actually do that. Any pointers, @FRWriter?

 

[Akkizakura]

There is no way to know with 100% certainty, but there are tells.

First, and most importantly, more than any single stylistic element, AI text is repetitive. One use of "It's not X its Y" isn't terribly suspicious, but several in a row is. If there's one thing you take away from this post, let it be this.

Em dashes are another example. They can be used in place of just about any punctuation mark, and a large number of them is another red flag.

Another tell I've noticed is an overuse of rule of 3, especially in a very particular way. Again, rule of 3 is very common, but an overabundance is a solid tell. AI text tends to also follow a strict format of short, short, long. Here's an example I stripped from a synopsis: "...is soft-spoken, careful, and used to hiding her sparkly purple eyes and lavender ears behind silence." There are three beats, and the third is noticeably longer. Technically speaking I can't prove this was AI generated, but it really feels that way.

Also in the above example is another hallmark of AI text: strange metaphors. AI doesn't "know" what similes, metaphors, or analogies are, but it "knows" what they look like. These elements make sense if you breeze past them, but begin to fall apart if you look more closely.

Some AI text has a very choppy writing style. This manifests as an abundance of periods, sentences that start with "and" or "but," and sometimes even single words sentences. This style isn't out of place in dialogue, but usually is in narration.

Lastly, AI generated text has a certain vibe to it that is hard to give concrete examples of, but is noticeable, even if not consciously. This usually manifests in important details appearing or disappearing suddenly, or some plot elements getting unbalanced amounts of focus in text. It's almost impossible to give large scale examples, but it's something you can feel out.

Funnily enough, writing errors are a great indicator that something was written by a human. AI models won't misspell words, or use the wrong homophone, or accidentally add an extra space, or drop a word entirely. These mistakes can only happen by human hands.

There are several meta clues you can use too.

A fast upload schedule is can be suspicious. Some people like to build a backlog and then post all at once, but often, if someone is posting daily, or more than once a day, it can be an indicator to look more closely.

Profile activity can be another clue. How new is their account? Does this person comment often? Do they have a reasonable looking reading list? Again, none of these are damning. Having a new account, or being a lurker, or not reading very many fictions are fare from reliable indicators, but in conjunction with other details can increase your certainty.

An absence of AI images is a strong indicator that the work is made by a human. Some authors use AI images as placeholders, or just because they're poor, but if someone is using AI to generate their story, they're also the type of person to use AI to generate their images too. Writiers who pay commission art, or forgo it entirely out of refusal of using AI, almost certainly aren't generating their stories.

This essay's been bouncing around in my head for a while, glad to finally have an excuse to let it out. There are certainly some things I missed, but this will get you most of the way there.

I know someone who intentionally write like AI just to get people commenting on the story because bad engagements are still engagements to them. Your additions to the tells are appreciated.

 

[VanVeleca]

It's always the same thing that gives them away. "It's not x, it's y" "They didn't do this, they did that". It aggravating to read over and over again. The AI can't seem to figure out how to describe something any other way. Has anyone else noticed

some "writers" have already figured that out and edit out any of the major obvious signs, but they can't get rid of overly generic dialogue and characters within the story seemingly having forgotten major plot points. (Including the narrator, if there is one.)

Also a new(?) one I've noticed, whenever a character gets angry their dialogue suddenly includes ebonics...

 

[FRWriter]

Hmmmm... Bjt that means he already doesn't fall into the category you said would be totally easy to spot. He can definitely capitalize the "i"

Yes, he is obviously not among those whom I was talking about and knows how to write and how to argue properly, which makes his use of AI even more confusing to me.

I was talking about most other generators.

 

[Shorgoth]

I use AI to get feedback on my story.

It's always going to misinterpret things, overanalyze, or rely on clichés… and that's exactly what I need. I have to explain the story in detail to it, and that process helps me understand my own story better (since I have to put it into words).

And actually, sometimes the AI's feedback is surprisingly spot-on.

I do the same; it helps me better understand the symbolic layers and lenticular analysis of my narrative and keep them coherent in the long run by... misinterpreting XD

 

[FRWriter]

On another note, I'm not good at telling AI apart from normal writing and I don't want to buy into this whole game, because I'm naturally paranoid, which would mean I would start seeing AI everywhere.
But let's say there's a story that I feel has weird things in it that strike me as AI, but "detectors" are inconclusive - first, it's totally AI and suddenly, a 100 Percent human on the same scanning platform.
I started reading that story because the author asked for feedback and I said would look into it. Now I feel like it might be a waste of time to actually do that. Any pointers, @FRWriter?

Just a couple of things:

The use of — is a sign, but when I started out it was all the rage, and I got used to it. I only recently switched back to ';'

Also, with 1 small prompt, you can set a rule for AI never to use — but solely rely on ',' and ';' so you can't judge a story purely on that.

Meaningless, pretentious 1 and 2-word sentences are the clearest sign at the moment.

Doing my best to copy AI garbage style, so don't laugh at me.:
----
The room was not only large and spacious, but it was also cold...

freezing.

cooling.

soul-suckingly sterile.

"This piercing cold is... final. It's the end." AIbion muttered under his breath, the cold air leaving his mouth, the faint trace of heat leaving his body quickly.

Quietly.

Forever.
----

So you see, the latter short sentences are all worthless.

No human writes that way. AI very often does it for 'dramatic effect'.

AI also does not simply say: "asks, says," but there are a few key words like. "Mutters under his breath."

You also need to think of the setting. If it's a WEBNOVEL, about high school students, but narrated like a Lord of the Rings Fantasy book, you also got your answer. It's quite obvious when you pay attention, especially when your target audience would never appreciate that kind of style.

I think @Ararara did an excellent post here: https://forum.scribblehub.com/threa...n-my-story-not-sugar-coated.27966/post-640514

The guy asking for feedback has since deleted his 0 Reader story and revised it, so it's a little more human now. Sadly, I can't use that one as an example, but it would have been perfect.

You can spot most AI stories if they create a post in the feedback section, but only write one or two small sentences or outright admit that their grammar sucks. For example, like this:

If the story ends up not only free of mistakes, but with difficult words, you already know the truth.

 

[Shorgoth]

You know, it's not just AI using the M dash; they use it because it's valid punctuation.

AI detection tools are completely farcical trash; they don't work, and it has been demonstrated over and over again. And often, the usual witch-hunter anti-AI crowd jumps to the wrong conclusion and burns a random person. Happened to me, and I've been working on the same book for the last 18 years to make a unique narrative style, and those people took that as a tell... I'm tired of all that toxic behaviour against random people. There are plenty of issues with AI, wealth redistribution, electricity and water hoarding, pollution and the list goes on, but individual users who are just trying to make something should be left out of it.

You learn to notice a pattern with the years; always the same crowd going after photography, Photoshop, CGI, sampling (in music) and so on. They see a tool that can do something and freak out as if it were robbing people of their individual capacity to make things on their own. The bandsaw didn't stop manual labour for carpentry to be possible; it just offered different ways to do things. It requires different skills. AI is the same; if you don't have the skills, you make slop with it, like you make slop with a paintbrush if you don't know how to use it. Right now, AI slop is quite visible because 1 AI is young and it's hard to make something good with it, 2 people are still learning the tools, and 3 it's new.

I've been using it to add music and images to supplement narration (I was already cutting heavily into narration for dialogue in an experimental format closer to a screenplay). I'm just offering more information to understand the action and atmosphere with them. (A picture is often worth a thousand words), I know how hard it is to make anything specific over any random stuff.

Without any context, it's easy to have AI make "something"; what is really hard and requires mastery is how to have it make something specific that fits a holistic narrative. It requires complex vocabulary, a clear image of what you want as the creator, learning how the model reacts as it has biases ingrained in it, fighting you. If I could draw, it would be easier. Sadly, I'm on the aphantasia spectrum, I can't make any proportions right, no matter what, nor am I a man orchestra able to be 50 bands to make all the music I need for my story. So, as a poor disabled man, AI it is. I do have technical and conceptual knowledge about music and art, and it helps me tremendously. You can clearly see the difference in output qualitatively between someone who understands what they are pushing AI to make and those who have no clear concept of the thing they are making. This is also why there is so much slop, without technical knowledge, most people default to what AI give them. A good writer can use AI to make a good story if they patiently and painstakingly prompt it and correct it when it doesn't work. A random someone without understanding of narrative will most probably make slop... and few people who use AI are skilled at the subject they are experimenting with... for now.

Thing is, those who really are using it like a tool and work hard on it tend to have a real DIY spirit and autodidact instincts, I haven't learned as much fringe musical and visual styles and concepts as recently by trying to use AI to do specific things I have a general knowledge of without the linguistic skills to express them, so I look it up, learn the history of it at the same time... I have progressed a lot, even though I had university classes in those subjects. There is just too much to know.

 

[OmegaC]

@FRWriter
You're being a bit too intense, I think. Of course I use AI (I never even denied it), especially when translating my story into English. However, claiming it's "mostly AI-written" is a bit much, "mostly" is a pretty strong word.

I stick closer to the literal wording to preserve the original vibe, the clearest example is in chapter 12.5: "Here's another Japanese oddity... wait, no, a cosmic oddity..."

This is a popular slang in Vietnam (search "Chuyện lạ Việt Nam" or "Độc lạ Bình Dương").

---

You know, it's not just AI using the M dash; they use it because it's valid punctuation.

AI detection tools are completely farcical trash; they don't work, and it has been demonstrated over and over again.

Yeah, some of my college friends are complaining about professors using AI checkers.

The worst part is when you write everything yourself and it still claims you used AI

---

Spoiler: My Baby

 

[XineFury]

Em dashes are another example. They can be used in place of just about any punctuation mark, and a large number of them is another red flag.

Things like this bug me, because I use em dashes a lot, and don't want to be accused of being AI.

Funnily enough, writing errors are a great indicator that something was written by a human. AI models won't misspell words, or use the wrong homophone, or accidentally add an extra space, or drop a word entirely. These mistakes can only happen by human hands.

Never mind, I'm pretty safe from accusations, then.

 

[Assurbanipal_II]

What is meowing on here?

 

[TheKillingAlice]

Just a couple of things:

The use of — is a sign, but when I started out it was all the rage, and I got used to it. I only recently switched back to ';'

Also, with 1 small prompt, you can set a rule for AI never to use — but solely rely on ',' and ';' so you can't judge a story purely on that.

Meaningless, pretentious 1 and 2-word sentences are the clearest sign at the moment.

Doing my best to copy AI garbage style, so don't laugh at me.:
----
The room was not only large and spacious, but it was also cold...

freezing.

cooling.

soul-suckingly sterile.

"This piercing cold is... final. It's the end." AIbion muttered under his breath, the cold air leaving his mouth, the faint trace of heat leaving his body quickly.

Quietly.

Forever.
----

So you see, the latter short sentences are all worthless.

No human writes that way. AI very often does it for 'dramatic effect'.

AI also does not simply say: "asks, says," but there are a few key words like. "Mutters under his breath."

You also need to think of the setting. If it's a WEBNOVEL, about high school students, but narrated like a Lord of the Rings Fantasy book, you also got your answer. It's quite obvious when you pay attention, especially when your target audience would never appreciate that kind of style.

I think @Ararara did an excellent post here: https://forum.scribblehub.com/threa...n-my-story-not-sugar-coated.27966/post-640514

The guy asking for feedback has since deleted his 0 Reader story and revised it, so it's a little more human now. Sadly, I can't use that one as an example, but it would have been perfect.

You can spot most AI stories if they create a post in the feedback section, but only write one or two small sentences or outright admit that their grammar sucks. For example, like this:

View attachment 48624

If the story ends up not only free of mistakes, but with difficult words, you already know the truth.

From what I can see in that other post, I wouldn't say it's the same. But it also feels similar in some lines? I really can't describe it well, because the pattern is not exactly the same. And since the setting is "more serious", it's also not obvious when it comes to the narration. It could be legit, but it feels off.
As for the other things: I mean, when it comes to the mDash - I use them. I've been using them for 13 years now, and quite a lot of them as well. I use Dash (when lazy) or nDash (in stories), because German uses nDash and mDash technically interchangeably. I say "technically" because in reality, you will never see an mDash actually in use in the German language (it's a real unicorn), and I personally hate how long it is. I initially didn't know it wasn't interchangeable in English, but since nobody ever cared, I kept using nDash, but I definitely use it in the same way. So if I ever corrected that, people would assume that it's AI? Again, I've heard it before, but it's literally not at all the same as a semicolon in how they are used, so you can't just switch to that, if you wish to use that sentence structure. It's kind of hard to pin it on that, to be honest.
The same goes for phrasings like "Mutters under his breath," especially since regular inquits like "said" and the like are usually frowned upon when you use them a lot, so you gotta have variation or leave it out altogether. I wouldn't fault someone for that, is what I'm trying to say.
But I also don't want to call the story out to get other opinions on it, and shit on it that way. I don't think it deserves that, I just feel like I might be wasting my time with feedback on it.
Anyway, thanks for your help here, Cookie

 

[AliceMoonvale]

Yes.

Not hard to notice the 'it wasn't this, just this' type of statements. But also all the emotionless, overly used generic wording, tonal inconsistencies far beyond the typical person that don't make sense, forgotten plot lines, etc.

Nobody will ever be able to 100% on whether something is ai assisted or not, but the morons who don't make any effort to hide or edit it will be very obvious.

I feel like it's the same vibe people get when someone tries to speak their language, but the person clearly isn't fluent at all and only learned from a bot and or from lazy/poor education. It'll sound weird and even off-putting. That's how majority of ai works/assisted works sound to me. You can just tell it reads weird and it only gets worse when you actually analyze all the writing.

 

[greyblob]

ai has a distinct smell. it's like chocolate covered shit. looks good until you take a bite.

for complete ai written stuff with no human direction its really easy. ai cant focus for shit. everything is going everywhere


for ai writing guided by humans, the prose is dogshit. a lot of comaprisions, awkward descriptions, generic characters and dialogue etc etc

if youre using ai to write youre as much of an author as a proomter is an artist

 

[Lysander_Works]

I think it's time to train an AI to detect AI in stories.

 

[Makimaam]

Prompter AI generated work is obvious. Translated work is less obvious, but still recognizable.

Pro prompter... funny thing, I still see many reviewers describing these as having elegant prose and strong literary quality. But they are surfaced-level. Look closely and you will notice the incongruities.

If most writers and readers cannot tell, then is it a success? Perhaps.

While Ai assisted writers believe that writing is simply storytelling, that assumption alone is enough for them. Little do they realize that their so-called “voice” is replicated across thousands of similar works, as if written by the same person. The same signature.

An analogy is starting your own business. Everyone can come up with great ideas, but only those who know how to execute them succeed. 90% of startups fail. Ideas alone aren’t the oil that keeps the machinery running.

In the end, are they all the same? Piggybacking on the very distinct voice that LLM models have and cannot seem to shake, no matter how advanced the prompt is.

 

[ThaddeusDeVilliers]

Yes, mostly, mostly its just bad writing, or boring writing, its not fun to read.