"AI" Rejection

I understand that a confrontational response can harden opposition, but I think you should also recognize that your own contributions to the thread have been pretty inflammatory. You've acted very dismissive about authors getting trapped in a Kafkaesque process of false and insulting accusations that they have no way of defending against, and being arbitrarily blocked from the site on those unfair grounds. Despite all the evidence that AI detectors are junk that regularly (not just in rare cases, but frequently) classify human-written texts as AI-generated, you completely ignore that and persist in arguing as if they can generally be trusted. You've suggested that people who disagree with you must be arguing in bad faith, surreptitiously using or wanting to use AI more than they admit. And like many of those defending the policy, you go pretty far towards implying that even if authors haven't actually used AI, it's no great loss and they deserve to be kicked out anyway because their writing must not be good enough.

Do you not see how that leads to heated responses?

You've written some of my favorite stories on this site, so I'm sorry to find you on the other side of this controversy. I hope you reconsider.
Well said and thank you for adding a bit of perspective.

Please know that I’m not at all upset or angry about the issue, or even the responses in this thread. I do need to do a better job of accounting for the fact that many here are. Doesn’t excuse some of the responses, but it goes a long way to explain them. As they say, “the axe forgets, but the tree remembers.”

But if you’re reading anything I wrote in a dismissive, condescending or angry tone, then what you think I’m saying is being colored by your feelings about the issue. I am genuinely trying to understand what is going on, but maybe that’s not possible by asking questions of those who are in the midst of a conflict.

I’m going to bounce out of this thread without trying to do any more harm to those who are sincerely angry over getting their work rejected, but I would only share these observations with the sincere hope they help find a way through and get back to writing about what you love:

The examples posted here don’t read very well as organic dialogue. As I’ve said, I’ve been there, am still, and will always be, trying to get better at it. In many ways you have to imagine each person in depth, have a clear sense of character identity and be scrupulously willing to live within the constraints of the world that you’ve built around them. None of that is easy, nor are there any shortcuts that I’ve found. I think there may be a few people who have a natural gift for it, but most of us do not. I’ve never seen anyone invest a good amount of effort over a long period of time who didn’t get better at it. Just like any skill.

Portraits are especially hard to paint because we have built in templates for things like feature placement and spatial relationships in the shapes that tell us that what we are looking at is ‘human’. Ask any visual artist who has made the transition from another subject matter and they will tell you that it’s brutal. Especially as we are first starting out.

I think the same is true with dialogue. It has a rhythm to it that is hard to fake. It shifts constantly as the feeling of the participants change. The words we choose have to reflect the vocabulary, life experience and mood of those we have saying them. There are pauses, additional emphasis, non verbal cues that need to be added and it all has to flow in a way that our mind recognizes as ‘human’. It’s really, really hard to fake. The better we do that the more the story becomes immersive. We stop ‘reading the words’ and we start hearing the voices, imagining the expressions and getting caught up in the feelings of those who say them. That’s no small thing.

Again, the good news is that the same things that make us a better writer will likely also make us less likely to get caught up in the AI filter. I don’t think there’s a formula for dialogue. Think of Sheldon trying to write an algorithm for humor on Big Bang Theory. Or the bartender in “Passengers” saying, “These are not robot questions”. Everything I know about predictive analytics, mining large datasets, assessing software application and writing algorithms tells me they are great at distilling and clarifying complex patterns, but they have limited value in solving what are essentially ‘human’ problems.

We’re more predictable than we imagine and far less unique than we believe, but we are also completely irrational about certain subjects and our patterns are constantly disrupted by responses that look inappropriate to those who don’t, and honestly can’t, know our whole history. Add just one or more other humans to that exercise and it sets off a chain reaction of disruptions, inappropriate and irrational behaviors. Factor in fatigue, physical pain, uncomfortable temperatures and a smell with a strong memory association and it’s like shaking up a snow globe and you have a complete reset.

All of which is to say, give yourselves a lot of grace in developing a way to mimic those behaviors. Keep grinding. Slow down to work (and rework) the dialogue of your characters until it flows and the words disappear because you are too caught up in the story to see it any longer. Read. A lot. Go back and look at the blocking and tackling of what that writer did to make the words disappear and suck you into the story. Develop your ear and take the time to listen to your own voice. It’s all there. It just takes time.

Take comfort in the fact that if these tools are as bad as you imagine, large numbers of writers here will quickly join your camp and you are more likely to make headway in changing the submission process. But that really is the key. You need converts. A lot of them. You need to be able to explain and demonstrate your position in a way that solidifies their support. Mocking them for asking questions until they leave the debate will only hurt your cause. Having clarity and unity in what you are asking for is critical to that process. As is policing your own supporters. If one of them gets frustrated and decides they are going to take a whizz off the end of the diving board, you’ll have a harder time convincing the rest of those at the pool party that your goal in convincing the host o stop using the dye isn’t just a ploy to make peeing in the pool less detectible.

Be well and good luck in working through the next steps in the process.
 
Like you, @jhealy55, I understand, appreciate, and even empathize with everyone's views. And yes, @TheArsonist I have said I don't use AI several times. It is writing for me, it isn't what gives me pleasure if I allow a program to write my stories.

We are shotting in the dark since there is no clear way to understand what they are judging AI to be. We don't know what criteria is being used, if it is a smell test, an eye test, software assisted or a coin toss. I don't suggest for anyone to change how they write, but how one edits (creative editing) should be considered. Even so, if you mind works a certain way, what you produce (to Laurel or her software) may always look AI-ish. In world filled with AI, your phone, TV, internet radio, search functions, and social media mining your preferences to give you what your looking for, or give you what the sellers want you to buy, I'm not sure we will every be able to know what is human and what is machine, for certain, writing.

That last paragraph is rather complicated, perhaps I should run it through AI to uncomplicate it. :kiss: :heart: :ROFLMAO:
 
What I find the most irritating aspect of grammar and writing software assistants is the constant desire to make everything standardized. It wants to remove introductory or concluding clauses from sentences or move them to the opposite end. It desires to remove clarification words or phrases, such as--some of them, most of them, each and every one of them, from the sentence or truncate it, in some way, to one word to take its place. In short, it wants my writing to be like everyone else's work. To remove that which makes a sentence in fiction unique and make it more like a business letter than a story.

As an example, Grammarly wants to replace, "What I find most irritating," with, "The most irritating." It doesn't matter that I have Grammarly and ProWritingAid set to creative and General Fiction. They want to simplify and remove creativity from the equation. Yes, sometimes, being creative creates passive sentences, and I deal with 95 percent of those by rewriting them myself. The other five percent are left passive. I only want grammar and structure aids to aid me with grammar and structure, not nitpick every fucking thing.

I think structuring some things, like a thousand writers before you, makes the AI detection say, "Oh, yeah, we stole this for you. It must be AI if we've seen it this way before." For crying out loud, how many ways can say, "He's a tall man."
 
Last edited:
What I find the most irritating aspect of grammar and writing software assistants is the constant desire to make everything standardized. It wants to remove introductory or concluding clauses from sentences or move them to the opposite end. It desires to remove clarification words or phrases, such as--some of them, most of them, each and every one of them, from the sentence or truncate it, in some way, to one word to take its place. In short, it wants my writing to be like everyone else's work. To remove that which makes a sentence in fiction unique and make it more like a business letter than a story.

As an example, Grammarly wants to replace, "What I find most irritating," with, "The most irritating." It doesn't matter that I have Grammarly and ProWritingAid set to creative and General Fiction. They want to simplify and remove creativity from the equation. Yes, sometimes, being creative creates passive sentences, and I deal with 95 percent of those by rewriting them myself. The other five percent are left passive. I only want grammar and structure aids to aid me with grammar and structure, not nitpick every fucking thing.

I think structuring some things, like a thousand writers before you, makes the AI detection say, "Oh, yeah, we stole this for you. It must be AI if we've seen it this way before." For crying out loud, how many ways can see "He's a tall man."
And "Very." Come on, sometimes you need two or three for emphasis... :)
 
I think you should resubmit your text, no changes, with a similar Note to the Editor as you have here.

If the site is introducing a tool to "recognise" AI and it's false finding, then the onus should be on the site to explain what is triggering. After all, only a human mind will be able to actively respond and address that, given that AI is artificial, but not intelligent. But there needs to be something to respond to, not a generic statement. @Laurel to consider, please (asked on behalf of the frustrated folk in this tread - who sound human to me).

This is the biggest issue. I work in academia, specifically in the ed tech space, and I can tell you that there isn't a single software option out there that's even 25% accurate in identifying AI writing. We've even instituted a video proctored writing sample for incoming graduate students because the current software is useless and we know many students aren't doing their own writing on applications.
 
Meanwhile, an AI-written story has won a Chinese SF contest: https://cybernews.com/news/ai-novel-wins-prize-china/

A few caveats on that story:

- 40k characters in Chinese would usually equate to around 24-28k words in English, which is more "novella" territory than "novel"
- Looks like the story was judged based on an excerpt of 6k characters (~3600-4200 words eq.; about 1 Lit page equivalent) rather than the whole text, so this doesn't tell us whether the software can generate a coherent novella-length story; IME the longer generative AI goes on the more likely it is to have continuity glitches.
- Article mentions 66 prompts to construct the whole thing, so the excerpt would've been based on about 10 prompts, which still leaves the human operator a fair bit of room to shape the story within that excerpt.
- One of the judges did spot it as AI, and "Despite winning the award, judges agreed that The Land of Machine Memories could not be published as is and would require significant revisions".

But still a very interesting development.
 
They need to publicly explain what's happening here. AI detection software is utterly useless at this point in time. I submitted a story 2 weeks ago that has no AI or even grammar software used. I wrote it 7 years ago. It was posted here previously under a different username. I took that profile down, along with my profile in other writing forums, due to a moderator at Freecatfights who was threatening several people with doxing (sadly she's still one of the primary mods over there).

I waited for two weeks for this story to get through pending, and then it was sent back as suspected AI. If what others here are saying is true, I'm going to be waiting a stupid amount of time for the resubmission to be looked at, even though the story was literally edited once (I added a comma and changed the tense of two words) since it was originally written years ago. There is no excuse for what's going on here and it's telling that Laurel and Manu aren't being open with the process. It's okay to have things be flawed. It's not okay to be secretive about it to the point that it's clearly got a lot of regular writers agitated about how the process has declined precipitously in just a few months. The sheer number of threads dedicated to the false positives in the AI feedback process should be enough for them to scrap it, or just be honest about why it's happening and what they're doing to improve it.
 
Uh oh, this discussion makes me worried for the next time I submit a story. I work on them on and off for months and changing my writing style to try to not sound like AI (should I be flagged) would ruin the enjoyment for me. Was anyone able to get the issue resolved with their stories?

I don't think getting flagged for AI is a sign of bad or dull writing. The training data for language models is an enormous corpus of human-written text. Its outputs are based on probabilities, so AI should sound like human-written text (especially when asked to write in a specific style), which might be why tools for detecting AI have so many false positives.

The site owners are in a tough spot. AI-generated content is everywhere and we don't have a reliable way to detect it. And because tools like ChatGPT make it so effortless to generate a lot of content quickly, it can sting if you're accused of using it when you haven't. When you put a lot of effort into something, you don't want someone thinking, "oh that took five minutes with an AI."

I'm still undecided on how I feel about people using AI for writing. I don't want to, because I like figuring out what words I want to use. But it's a tool that is already changing jobs (including mine) and making people more efficient.

My bigger fear at the moment is some AI identifying me as having written my stories because I've written other things online under my real name. Just like with facial recognition, is it possible we all have a unique, identifiable writing style in which our diction and structure and quirks could be used by AI to output a probability of our authorship of a piece of text? Then we would have to use another AI to scramble our writing to stay anonymous!
 
My bigger fear at the moment is some AI identifying me as having written my stories because I've written other things online under my real name. Just like with facial recognition, is it possible we all have a unique, identifiable writing style in which our diction and structure and quirks could be used by AI to output a probability of our authorship of a piece of text? Then we would have to use another AI to scramble our writing to stay anonymous!

People have been interested in using computer analysis of text to identify authors for decades - questions like "did Shakespeare write this play?" and "does the letter from this serial killer match this suspect's writing style"? And yes, the recent large language model "AI" technologies have been adapted to this use, e.g.: https://www.nature.com/articles/s41598-022-13690-4

That article reports accuracy of about 75% in picking the correct author when dealing with a sample of 20 authors. Still some way to go before it'd be useful for identifying authors out of a field of millions, I guess, and it'd depend on how much material is available on each author. But this seems like the kind of thing LLMs probably would be good at.
 
Three stories on the queue. Thirteen days between submission and rejection for the first two. Both resubmitted with notes saying they're not AI, and that no program has edited my stories. The third is on day four, and I fear it'll go the same way.

I write on my Notes app, and run the stories through a random online checker at the end, to find errors I didn't spot or to add/remove commas (and I ignore the comma suggestions if they don't sound good). I then manually change the story on the Notes app, never letting the checker do it for me.

I don't use AI for my real job, I sure won't use it for something I do for free and for fun (and for the numbers that go up, and the comments, and the red H).

They're all short stories, with almost no plot, a severe lack of description of the characters and the scenes, and my POV characters don't even get a name (I quite like this nameless part though). I'm not a good writer, of that I'm sure, but the readers seem to be enjoying my stuff.

It's a bit discouraging to read replies in this thread saying that I might continue to get rejections on the stories that I resubmitted, and even more now that I'm writing one for the Valentine's Day contest, since contests have time limits.

Well, at least I'm building up a backlog of stories that will eventually get published. Right? Right?
 
Three stories on the queue. Thirteen days between submission and rejection for the first two. Both resubmitted with notes saying they're not AI, and that no program has edited my stories. The third is on day four, and I fear it'll go the same way.

I write on my Notes app, and run the stories through a random online checker at the end, to find errors I didn't spot or to add/remove commas (and I ignore the comma suggestions if they don't sound good). I then manually change the story on the Notes app, never letting the checker do it for me.

I don't use AI for my real job, I sure won't use it for something I do for free and for fun (and for the numbers that go up, and the comments, and the red H).

They're all short stories, with almost no plot, a severe lack of description of the characters and the scenes, and my POV characters don't even get a name (I quite like this nameless part though). I'm not a good writer, of that I'm sure, but the readers seem to be enjoying my stuff.

It's a bit discouraging to read replies in this thread saying that I might continue to get rejections on the stories that I resubmitted, and even more now that I'm writing one for the Valentine's Day contest, since contests have time limits.

Well, at least I'm building up a backlog of stories that will eventually get published. Right? Right?
Happened to me, too. After the first rejection, VE told me to re-submit. noting that it's not AI - I did. Reject again after 12 days pending. VE said he'd try to restructure sentences to make it through their detector. I decided not to - didn't want to waste his time. Free site and I don't need the viewership for selling stories elsewhere, so it wasn't a tough decision. Small-time author - only had 37 stories posted, 5 mil views. Deleted my account and moved to another site. Now, when I post a story, it's published within 10 minutes. And if I have a question, the site owner/webmaster answers within an hr or so. I feel much more appreciated, but the viewership is way lower. Had 4k followers here, 140 on new site, but that's not a biggie for me.
 
Happened to me, too. After the first rejection, VE told me to re-submit. noting that it's not AI - I did. Reject again after 12 days pending. VE said he'd try to restructure sentences to make it through their detector. I decided not to - didn't want to waste his time. Free site and I don't need the viewership for selling stories elsewhere, so it wasn't a tough decision. Small-time author - only had 37 stories posted, 5 mil views. Deleted my account and moved to another site. Now, when I post a story, it's published within 10 minutes. And if I have a question, the site owner/webmaster answers within an hr or so. I feel much more appreciated, but the viewership is way lower. Had 4k followers here, 140 on new site, but that's not a biggie for me.
That's sort of cutting of your nose to spite your face, though, isn't it? To toss out 4k followers and five million views just because you didn't have the patience to cope with a new technology problem, the same one a bunch of other writers are suffering? I'm sure those 4k followers appreciated you, just a little bit - where's your appreciation of them in return? Not much, clearly.

Deleted your account? So you've come in here with an alt account, just to complain? That's a weird way to treat a publication platform that gave you five million views and 4k followers. But hey, those 140, must be worth every penny.
 
I just had two stories rejected. One was a resubmit, that I made minor changes to after being accused of AI usage and the other was the sequel to that story.

I did use Grammarly to check for grammar but I didn’t change wordings or anything. I don’t know what to do. When I resubmitted the story I attested that I didn’t use Ai but it still got rejected.

The rejection mentioned possibly using an editor to fix the issue. Has anyone ever gone this route?
 
That's sort of cutting of your nose to spite your face, though, isn't it? To toss out 4k followers and five million views just because you didn't have the patience to cope with a new technology problem, the same one a bunch of other writers are suffering? I'm sure those 4k followers appreciated you, just a little bit - where's your appreciation of them in return? Not much, clearly.

Deleted your account? So you've come in here with an alt account, just to complain? That's a weird way to treat a publication platform that gave you five million views and 4k followers. But hey, those 140, must be worth every penny.
How? If he can’t publish new stories, those fans mean nothing.
 
How? If he can’t publish new stories, those fans mean nothing.
I do have some doubts that the author with 4k followers and 5 million views, so, presumably a long history here, suddenly couldn't get anything posted, and deleted his account before coming here with a brand new one to complain about it. Anyone disappear from the most favorited authors list recently?

But yes, if it happened, it would be very irritating and his response would be understandable.
 
I do have some doubts that the author with 4k followers and 5 million views, so, presumably a long history here, suddenly couldn't get anything posted, and deleted his account before coming here with a brand new one to complain about it. Anyone disappear from the most favorited authors list recently?

But yes, if it happened, it would be very irritating and his response would be understandable.
I know there are a few stories that used to be here in LW that are suddenly gone, and their authors appear to be as well, but that doesn’t necessarily mean anything.
 
4k followers is down below the 100 mark on that list. Only one of those followers or somebody close who watches their position would be likely to notice.

I doubt I'd pull my stuff for one or even two, but repeated erroneous rejections would be enough to make me stop submitting here until things returned to a semblance of normalcy. Different circumstances, but that's what I did on another site not too long ago.
 
I didn't post a reply to complain, as some have stated. I answered someone who experienced a similar problem as I encountered. My followers can google my name (MrCurrie) or the story they liked and find me in seconds - many have already done it and PM'd me. It took 8 yr to get 4k followers - 140 at the new site in 2 weeks. The new site doesn't have views, but they do log ebook downloads, and some of the stories already have 20k downloads. Top story was 4.86 so no one is going to see a difference in the lists. A lot of views were from one story. You stick 'Breed' and 'Mom' in the title or description and it'll get a lot of attention. The reader base here is comprised of a lot of sick motherfuckers - literally.

1703874697627.png
 
You appear to have had this account for some time, but you've only made two posts, both on this page. I don't know what new site you're talking about and don't care about it. But must assume, since the screen cap is from an author's control panel, you're kink-shaming your own readers at this site.
I didn't post a reply to complain, as some have stated. I answered someone who experienced a similar problem as I encountered. My followers can google my name (MrCurrie) or the story they liked and find me in seconds - many have already done it and PM'd me. It took 8 yr to get 4k followers - 140 at the new site in 2 weeks. The new site doesn't have views, but they do log ebook downloads, and some of the stories already have 20k downloads. Top story was 4.86 so no one is going to see a difference in the lists. A lot of views were from one story. You stick 'Breed' and 'Mom' in the title or description and it'll get a lot of attention. The reader base here is comprised of a lot of sick motherfuckers - literally.

View attachment 2301194
 
I didn't post a reply to complain, as some have stated. I answered someone who experienced a similar problem as I encountered. My followers can google my name (MrCurrie) or the story they liked and find me in seconds - many have already done it and PM'd me. It took 8 yr to get 4k followers - 140 at the new site in 2 weeks. The new site doesn't have views, but they do log ebook downloads, and some of the stories already have 20k downloads. Top story was 4.86 so no one is going to see a difference in the lists. A lot of views were from one story. You stick 'Breed' and 'Mom' in the title or description and it'll get a lot of attention. The reader base here is comprised of a lot of sick motherfuckers - literally.

View attachment 2301194
I stand corrected!

Yes, very strange they would think an 8 year old account would suddenly start being produced by AI. Bizarre and infuriating!
 
Back
Top