Thoughts on AI checkers

[StillStunned]

Clearly the mature category. It’s too perfect to edit and correct.

Not necessarily "Mature". Kevin Bridges* has a bit about "people who represent Glasgow", where he talks about "the children all need new tracksuits for Grandma's thirtieth birthday party."

* A comedian from one of Glasgow's poorer parts.

 

[BeechLeaf]

Speaking of grandma. Last time I checked this thread, no-one had made the point that AIs can now beat chess grandmasters, so checkers should be a doddle.

 

[dmallord]

Sorry about your experience

It only hurt for a little while.

I’ve lived long enough, and gone through enough, that an AI-related rejection barely registers compared to the real griefs I’ve carried—and one very present one that outweighs them all.

At this point, writing stories for Lit doesn’t feel as cathartic as it once did. With the time I have left, it makes more sense to focus on putting my affairs in order than on spinning out musings that get brushed aside like crumbs on the supper table.

Life isn’t about AI debates or appeasing grammar purists. There’s more to living than poking at those bears. We shouldn’t be bashing people for using tools to help them write; all that does is add more negativity to our lives. And bodies don’t age well under that kind of constant, needless stress.

Go forth and do good works, someone said long ago. I’m glad to see you walking that path soon as well.

 

[EmilyMiller]

It only hurt for a little while.

I’ve lived long enough, and gone through enough, that an AI-related rejection barely registers compared to the real griefs I’ve carried—and one very present one that outweighs them all.

At this point, writing stories for Lit doesn’t feel as cathartic as it once did. With the time I have left, it makes more sense to focus on putting my affairs in order than on spinning out musings that get brushed aside like crumbs on the supper table.

Life isn’t about AI debates or appeasing grammar purists. There’s more to living than poking at those bears. We shouldn’t be bashing people for using tools to help them write; all that does is add more negativity to our lives. And bodies don’t age well under that kind of constant, needless stress.

Go forth and do good works, someone said long ago. I’m glad to see you walking that path soon as well.

There is more than enough negativity for sure

 

[dmallord]

I'm convinced that this advice would solve the vast majority of "false," AI rejections. And I'm not putting in scare quotes because I think those writers are intentionally lying, I think they genuinely and honestly feel that Grammarly doesn't or shouldn't count as using AI. There are posts in this thread that basically say exactly that.

But regardless of what a writer feels in their heart about the provenance of their work, if they're letting Grammarly or a similar tool modify their text in any way, it's mechanically not any different from just running it through ChatGPT. Grammarly is now an LLM. It's doing the same things behind the scenes to your text, and Lit is flagging it in the same way it flags obvious and blatant AI-generated text.

And it's not about writing style or grammar or vocabulary, or em dashes, or linguistic triplets, or any of those things. There is endless daily evidence in New Stories that Lit will publish mechanically perfect writing, atrocious eye-melting writing, and everything in-between.

Points taken. I suppose it goes to intent, doesn’t it? Should we all return to pencil and lined paper and eschew the CPU as a tool, thereby ensuring no outside influences are brought to bear?

If you scribble down a thought and submit it to some machine to evolve into a story, I think that would be grounds to label it a bad-faith writing effort. A writer, in this example, deliberately chooses not to put in the effort to be that creative writer, and as such, perhaps deserves chastisement if the work is accurately described as insipid AI-generated.

On the other hand, if you find you’ve diligently written your epic tome and find along the way an AI tool suggests a correction or two in that voluminous work, is that any different from having a peer editor offer a suggestion and you accept and incorporate that into your work? No one here has yet to address collaborative editing as something to be frowned upon as well. Would taking advice from a human be the exception to modifying a story by an MMLLL or whatever they are at present?

Don’t get me wrong here. I’ve no problem with peer editing for those who have access to peers. My problem is with individuals who sit in judgment of someone they’ve never met, never read their work, nor have any idea of how the AI influenced their content. It smacks of judgment in a gladiator arena with all thumbs down for the fallen, regardless of their best efforts.

Asking for a friend.

 

[PennyThompson]

On the other hand, if you find you’ve diligently written your epic tome and find along the way an AI tool suggests a correction or two in that voluminous work, is that any different from having a peer editor offer a suggestion and you accept and incorporate that into your work?

I think it's a valid question, I don't have a super strong opinion on it either way!

I think there's a lot of value in relying on a combination of your own abilities plus a trusted beta reader or three, and I think that tools like Grammarly tend to sand down the interesting human edges of one's writing, a little bit at a time.

On the other hand I don't think it's the biggest AI harm around, and tools like Grammarly can be a genuinely helpful communication tool for people who are dyslexic or struggle with written tone or things like that

But the facts on the ground are that from a mechanical document perspective, there's no difference between an LLM that corrects your grammar and an LLM that writes for you. Whatever tool or technique Lit uses to check for AI sees them as the same thing, and can't differentiate, it doesn't see intent.

Not perfect, and frustrating for people who get caught in its filter... But it also seems to catch and filter a LOT of genuine, bad faith, AI slop.

 

[CalBishForNow]

Don’t get me wrong here. I’ve no problem with peer editing for those who have access to peers. My problem is with individuals who sit in judgment of someone they’ve never met, never read their work, nor have any idea of how the AI influenced their content. It smacks of judgment in a gladiator arena with all thumbs down for the fallen, regardless of their best efforts.

If someone puts a prompt into an LLM and calls the output their own writing, I judge them just as harshly as if they had put a plastic-wrapped TV dinner in the microwave and called it cooking. Maybe a little more harshly because writing is sorta kinda close to my actual day job, but they're both bad in similar ways.

I'm not saying I've never used a microwave, to be clear. Sometimes I'm tired and in a hurry and want food anyway, maybe even something cheap and more complicated than a sandwich. But it would be pretentious and dishonest even in daily life to call cutting the plastic foil and inputting the numbers in the microwave "cooking." When we aren't just talking about daily life, when we're talking about things people put time and energy and money into, it's worse.

If you're on a dating site and someone says that their hobbies include cooking and you later find out they mean heating up Lean Cuisines, would you think better or worse of them? If you're looking at candidates for a job and one resume says, "Served as personal chef for a demanding family of four, producing varied and nutritional meals at least six days a week," do you picture filling the cart in the freezer aisle? If you're a guest and the host serves you something they describe as their mother's favorite recipe, do you think there's a handwritten note card with the recipe and a personal story about learning to make it, or a wrapper in the trash with nutritional information?

Likewise, I'm not saying I've never used an LLM. It's hard to avoid; they're the default first answer from any search engine these days. But I wouldn't call using an LLM writing, and anyone who would is either stupid (a lot more goes into good writing than just selecting a prompt!), dishonest, or both. At most it's an optional part of the process, neither the first nor the last step. (Likewise, if a recipe calls for softening butter, sure, I might do it in the microwave depending on what else is going on and still call it part of cooking.)

 

[Actingup]

At this point, writing stories for Lit doesn’t feel as cathartic as it once did. With the time I have left, it makes more sense to focus on putting my affairs in order than on spinning out musings that get brushed aside like crumbs on the supper table.

It must suck, particularly when you're racing the clock as you are. This might not be exactly the right place for the conversation (but where is?), but I hope you decide to leave your stories up when you go, alongside your 'real life' legacy. I hope you're coping okay physically.

 

[dmallord]

I think it's a valid question, I don't have a super strong opinion on it either way!

I think there's a lot of value in relying on a combination of your own abilities plus a trusted beta reader or three, and I think that tools like Grammarly tend to sand down the interesting human edges of one's writing, a little bit at a time.

On the other hand I don't think it's the biggest AI harm around, and tools like Grammarly can be a genuinely helpful communication tool for people who are dyslexic or struggle with written tone or things like that

But the facts on the ground are that from a mechanical document perspective, there's no difference between an LLM that corrects your grammar and an LLM that writes for you. Whatever tool or technique Lit uses to check for AI sees them as the same thing, and can't differentiate, it doesn't see intent.

Not perfect, and frustrating for people who get caught in its filter... But it also seems to catch and filter a LOT of genuine, bad faith, AI slop.

I’m not conversant with the intricacy of LLM or grammar checking, other than the Pro version of Grammarly I subscribe to. I thought the purposes were different in how they function. The early version of Grammarly I used at first didn’t have an AI bent, now it does. I ignore that so I’m not relying on it to ‘write my story.’ Perhaps others do, I can’t say without some definitive study on the subjject. I can see it as a useful tool for ESL students, though.

I’m a little confused by your statement that there is no difference mechanically between the two, while you say one corrects grammar and the other writes a script for you. Those are not the same things by any stretch of the imagination. I’ll accept your word that intent is not seen by Lits AI, whatever flavor that is.

But color me curious: how do you know it seems to be filtering a LOT of genuine bad-faith AI slop? Did Lit say so, or is that musings among the locals of Lit? Or a little expansion on an opinion?

Without data on the rejections, one can’t make that leap in conjecture very soundly.

 

[EmilyMiller]

how do you know it seems to be filtering a LOT of genuine bad-faith AI slop? Did Lit say so, or is that musings among the locals of Lit? Or a little expansion on an opinion?

The site told me that they were getting a lot of AI submissions long before it became a mainstream topic of conversation here. The context then was translating work into English. But they said they got AI submissions natively written in English too.

No, I don’t have stats, it was a series of PMs about a specific story that had been rejected.

Clarification: it wasn’t one of my stories that got rejected. I was trying to help someone out.

 

[SmilingLez]

But color me curious: how do you know it seems to be filtering a LOT of genuine bad-faith AI slop? Did Lit say so, or is that musings among the locals of Lit? Or a little expansion on an opinion?

Without data on the rejections, one can’t make that leap in conjecture very soundly.

Your questions have one fatal flaw, you are assuming that the AI checkers (including the one(s) Literotica uses) are accurate 100% all of the time. Just because an AI checker rejects a story, it doesn't mean that AI was used to write the story.

I am willing to go on record and say at least 5% of the stories currently live on Literotica (submitted from 2020 to 2026) were either written by AI (either completely or sectionally) and bypassed Literotica's AI checker because it is faulty. Call it luck of the draw.

 

[dmallord]

It must suck, particularly when you're racing the clock as you are. This might not be exactly the right place for the conversation (but where is?), but I hope you decide to leave your stories up when you go, alongside your 'real life' legacy. I hope you're coping okay physically.

I will commit to leaving them up. A number of them have breadcrumbs to my life in them with some embellishment over an 87-year-old’s lifetime.

As to real-life legacy, that’s perhaps a story worth writing as a last cathartic exercise. Maybe one of the WIWAW story-types. Perhaps, there is enough energy to do that. I had planned to exit writing after this last story posted this week. Fatigue is getting to me. Words are harder to come by.

But thanks for the suggestion on a real-life legacy piece.

At present, I'm tethered to an oxygen machine and a low dose of morphine for comfort. It’s foggy at times, and I push my limits to stay off it as much as possible. The hospice nurses aren’t pushing me. It’s my pace, they tell me, and honor my control over the meds at this point. My last hospital doctor gave me two to six months. I fooled him. I’ve pushed past a year. But I know there has been a huge drop in that time.

I am okay with not waking up one day. What I haven’t figured out is how to make sure the nurse don’t find out I write ‘stuff’ for Lit.

 

[EmilyMiller]

I don’t think anyone is asserting that AI checkers are foolproof. They are notorious for false positives and false negatives, and this is widely known.

The question is: what to do instead? Drop the site’s pro-human policy? Make each author attest their work is human written, on pain of permanent ban and all stories being deleted?

 

[dmallord]

The site told me that they were getting a lot of AI submissions long before it became a mainstream topic of conversation here. The context then was translating work into English. But they said they got AI submissions natively written in English too.

No, I don’t have stats, it was a series of PMs about a specific story that had been rejected.

Clarification: it wasn’t one of my stories that got rejected. I was trying to help someone out.

This is a source I can rely upon as validation.

 

[PennyThompson]

I’m a little confused by your statement that there is no difference mechanically between the two, while you say one corrects grammar and the other writes a script for you.

Oh, I just meant in terms of what's happening "behind the scenes" so to speak. Input (your text) is being fed to a Large Language Model, which is generating output (new text) and putting it back into your document.

From the human's perspective that is different from what ChatGPT or something might do when you ask it to write a paragraph of text for you, but as far as the computer systems involved are concerned they're basically the same, and by all appearances they trigger Lit's detection process in the same way

 

[EmilyMiller]

I will commit to leaving them up. A number of them have breadcrumbs to my life in them with some embellishment over an 87-year-old’s lifetime.

As to real-life legacy, that’s perhaps a story worth writing as a last cathartic exercise. Maybe one of the WIWAW story-types. Perhaps, there is enough energy to do that. I had planned to exit writing after this last story posted this week. Fatigue is getting to me. Words are harder to come by.

But thanks for the suggestion on a real-life legacy piece.

At present, I'm tethered to an oxygen machine and a low dose of morphine for comfort. It’s foggy at times, and I push my limits to stay off it as much as possible. The hospice nurses aren’t pushing me. It’s my pace, they tell me, and honor my control over the meds at this point. My last hospital doctor gave me two to six months. I fooled him. I’ve pushed past a year. But I know there has been a huge drop in that time.

I am okay with not waking up one day. What I haven’t figured out is how to make sure the nurse don’t find out I write ‘stuff’ for Lit.

 

[dmallord]

Your questions have one fatal flaw, you are assuming that the AI checkers (including the one(s) Literotica uses) are accurate 100% all of the time. Just because an AI checker rejects a story, it doesn't mean that AI was used to write the story.

I am willing to go on record and say at least 5% of the stories currently live on Literotica (submitted from 2020 to 2026) were either written by AI (either completely or sectionally) and bypassed Literotica's AI checker because it is faulty. Call it luck of the draw.

I didn’t assume Lit’s AI was accurate at all. It rejected two of my stories that I wrote there. I didn’t use AI except for the spell check in Grammarly.

Luck has nothing to do with it.

 

[EmilyMiller]

Pertinent to this thread:

https://forum.literotica.com/threads/article-on-the-atlantic-about-grammarly-and-ai.1651710/