Grammarly

[gunhilltrain]

My suspicion is the site is being swamped with AI garbage, and the folk who are showing up here with (one assumes) genuine tales of woe are being caught up in it.

Why would the site be doing anything if there wasn't a problem? I don't think it's paranoia, I think it's trying to cope with a deluge of AI junk. Organisations respond to problems, but they don't go making them up.

Except if it's a government agency! I have no faith in them any longer. Anyway, I don't know a reliable way to deal with this problem. What would you do if you ran a site?

 

[ShelbyDawn57]

Except if it's a government agency! I have no faith in them any longer. Anyway, I don't know a reliable way to deal with this problem. What would you do if you ran a site?

Protect it and its integrity as best I could with whatever tools I had, even if it inconvenienced some of my members or patrons. In other words, pretty much exactly what Laurel and Manu are doing.

 

[cadeauxxx]

This site has been plagued for months with authors getting rejected due to the faulty AI detection tools being used. Every week, there are new names coming here to talk about their most recent rejections. It's the only reason I even check these forums.

It's a waste of time coming to this forum and making complaints. Authors have been coming on here going back to late last year. What tends to happen is concerns and frustrations are voiced, then we're met with a bunch of armchair experts who think they know what AI is, and they just lecture us about how bad your writing must be if it's getting flagged as AI. Goalpost shifting makes the arguments useless since it moves from point A, B, C, and to D just arguing what they think Ai is. It always ends with a form of being told "well, your writing just sucks. Learn to write better", cause smug elitism wins at the end of the day.

The biggest problem is that the site owner and whoever moderates this place are completely silent in all of this. I think it would help if Laurel and Manu started talking to us openly in public and we could engage in open discussion with them.
These threads always turn into arguments on what AI is, and telling people their writing is crap if it's being false flagged. This is truly a moral panic over AI, and it's gotten ridiculous. Search the forums on here and you'll see some of the insanity and also how unreliable these AI detection tools are that we are being held to. Someone ran historical US government documents through one and it came up over 70% AI detected on the basis of James Madison's writing. The biggest problem with the arguments is people don't want to meet in the middle on a shared belief and it turns into just endless debating over what AI even is.

The only advice I can really give you based on my frustrations and getting a few stories pushed through; is to lean on a thesaurus and constantly check for repeated word usage. An overuse of key words tends to trigger it, and the only solution is to change some of those words out by using a thesaurus. Something as simple as patterns in your sentence structure tends to trip the detection tool.

Hopefully this madness ends sooner than later, cause it has become completely ridiculous. You are not alone. Many, many, many of us have had these issues.

 

[ElectricBlue]

Protect it and its integrity as best I could with whatever tools I had, even if it inconvenienced some of my members or patrons. In other words, pretty much exactly what Laurel and Manu are doing.

Agree this. I'd be looking at the issue on a total inputs basis, and categorising what I'm seeing into some kind of like vs like. Somewhere there's going to be a grey area, and that's what we're seeing here in the forum. The thing is, we're not seeing everything going through the pipeline.

 

[geronimo_appleby]

I only use word's editor. I don't have a beta reader and I have some issues with grammar that my readers are always pointing out. But with all these bogus rejections going on triggered by programs I'm happy to be a grammar Nazi's nightmare, and never worry about rejections.

those 'errors' in grammar are style.

on a tangential note ref grammarly, et al. i wonder how it would fare with some of cormac mccarthy's (RIP) text! that could be a fun one to test out.

 

[StillStunned]

It always ends with a form of being told "well, your writing just sucks. Learn to write better", cause smug elitism wins at the end of the day.

Besides some lighthearted comments in the early days of these complaints, I've seen none of this in any of the ongoing threads.

People have offered commiseration and sympathy, they've explained what's happened with other writers facing the same problems, they've offered to analyse texts to see where the problem might be.

Overall, people have tried to be helpful, even in the face of seeing the same threads appear again and again.

 

[geronimo_appleby]

Besides some lighthearted comments in the early days of these complaints, I've seen none of this in any of the ongoing threads.

People have offered commiseration and sympathy, they've explained what's happened with other writers facing the same problems, they've offered to analyse texts to see where the problem might be.

Overall, people have tried to be helpful, even in the face of seeing the same threads appear again and again.

incest- and LW-themed threads are a regular occurrence too.

 

[OverconfidentSarcasm]

My suspicion is the site is being swamped with AI garbage, and the folk who are showing up here with (one assumes) genuine tales of woe are being caught up in it.

I'm still not sure where that suspicion comes from. I mean, yes, Amazon does have a problem with bogus AI-submissions, but Amazon is a market. Someone's motivated by money to create all that crap and put it up there. But here on Lit?

I already said so in another thread, but I took the liberty of asking around. I asked the people running SOL, AO3, Lushstories, BDSM Library, and CHOYA. And, according to the return emails I got, none of them are being "swamped with AI garbage". None of them seem to even be bothered by it.

So... why would Lit have that problem? What would you, as an AI content "creator", gain by publishing garbage on a free site?

 

[StillStunned]

I'm still not sure where that suspicion comes from. I mean, yes, Amazon does have a problem with bogus AI-submissions, but Amazon is a market. Someone's motivated by money to create all that crap and put it up there. But here on Lit?

I already said so in another thread, but I took the liberty of asking around. I asked the people running SOL, AO3, Lushstories, BDSM Library, and CHOYA. And, according to the return emails I got, none of them are being "swamped with AI garbage". None of them seem to even be bothered by it.

So... why would Lit have that problem? What would you, as an AI content "creator", gain by publishing garbage on a free site?

It's human nature. People take credit for other people's work all the time, right down to repeating jokes as their own to get the laughs and cutting Gary Larson's signature out of Far Side cartoons.

A few days ago, someone posted that there are something like 150 new stories every day. Someone else said that Laurel mentioned getting 200-250 submissions every day. I don't remember the precise numbers, but it was something like that.

Either way, that means 50-100 stories that don't make it through, for whatever reason. Let's assume that AI is one of those reasons. And as soon as it becomes easy for people to swamp the site with effortless AI-generated stories, it will only get worse.

So a big thank you to @Laurel for holding back the flood!

 

[OverconfidentSarcasm]

It's human nature. People take credit for other people's work all the time, right down to repeating jokes as their own to get the laughs and cutting Gary Larson's signature out of Far Side cartoons.

And that human nature somehow doesn't apply to the users on all the other sites?

 

[StillStunned]

And that human nature somehow doesn't apply to the users on all the other sites?

I have no idea. Except that everyone's always saying that Lit is the biggest by a vast margin, and so probably attracts more people.

 

[ElectricBlue]

So... why would Lit have that problem? What would you, as an AI content "creator", gain by publishing garbage on a free site?

If Lit doesn't have a problem, the site wouldn't have come up with the AI rejection notice, which is less than six months old. The fact that it did, suggests something has changed significantly - hence my suspicion there's been an increase in AI derived content.

I have no idea why someone would want to publish AI content as "their own work", but you only have to look at the deluge of visual AI junk to know that people are using these "tools" - more "toys" in my opinion - because they're there.

 

[BobbyBrandt]

Whatever is (over)happening in rejection here for AI appears to have started within the last two months.

I know. I have submitted Grammarly-reviewed stories during this period. That was what I was trying to explain.

 

[BobbyBrandt]

If Lit doesn't have a problem, the site wouldn't have come up with the AI rejection notice, which is less than six months old. The fact that it did, suggests something has changed significantly - hence my suspicion there's been an increase in AI derived content.

I have no idea why someone would want to publish AI content as "their own work", but you only have to look at the deluge of visual AI junk to know that people are using these "tools" - more "toys" in my opinion - because they're there.

There are people who take pride in the creative aspects of designing and manipulating prompts within the AI engines to produce output that they believe is acceptable.

It does take skill, but the value of that skill is debatable.

 

[pink_silk_glove]

Goalpost shifting makes the arguments useless since it moves from point A, B, C, and to D just arguing what they think Ai is. It always ends with a form of being told "well, your writing just sucks. Learn to write better", cause smug elitism wins at the end of the day.

That's not true. Personally I do not recall anyone in this AH forum at least ever telling a rejected author to solve their problem by writing better. Not once. I do not agree with your assertion here in the slightest. I have however, seen an almost 100% basis, the suggestion to re-submit without any edits whatsoever and add an author's note to the admin declaring your innocence. So this flies directly in the face of your assertion.

The biggest problem is that the site owner and whoever moderates this place are completely silent in all of this.

That I can agree with.

 

[ShelbyDawn57]

That's not true. Personally I do not recall anyone in this AH forum at least ever telling a rejected author to solve their problem by writing better. Not once. I do not agree with your assertion here in the slightest. I have however, seen an almost 100% basis, the suggestion to re-submit without any edits whatsoever and add an author's note to the admin declaring your innocence. So this flies directly in the face of your assertion.



That I can agree with.

I’ll step in here and take @cadeauxxx’s side in that I personally have suggested to multiple writers with this issue that it is dryly based and perhaps adjusting their style could help. In my defense, that is not qualitative in any way, nor is it a slight on anyone’s writing.

Rather, I see it as pragmatic. If the way you write displeases the mystical AI gods or goddesses, then it’s simple, change the way you write, or continue to suffer their wrath.

 

[pink_silk_glove]

I’ll step in here and take @cadeauxxx’s side in that I personally have suggested to multiple writers with this issue that it is dryly based and perhaps adjusting their style could help. In my defense, that is not qualitative in any way, nor is it a slight on anyone’s writing.

Rather, I see it as pragmatic. If the way you write displeases the mystical AI gods or goddesses, then it’s simple, change the way you write, or continue to suffer their wrath.

that's not what he was saying. He was saying that those being rejected for possible AI have been hurled with insults to their craft. That is not true. I have not seen one single comment (perhaps I've missed one?) where someone said "you got rejected 'cause you suck!" so that is not at all true.

 

[EmilyMiller]

This site has been plagued for months with authors getting rejected due to the faulty AI detection tools being used. Every week, there are new names coming here to talk about their most recent rejections. It's the only reason I even check these forums.

It's a waste of time coming to this forum and making complaints. Authors have been coming on here going back to late last year. What tends to happen is concerns and frustrations are voiced, then we're met with a bunch of armchair experts who think they know what AI is, and they just lecture us about how bad your writing must be if it's getting flagged as AI. Goalpost shifting makes the arguments useless since it moves from point A, B, C, and to D just arguing what they think Ai is. It always ends with a form of being told "well, your writing just sucks. Learn to write better", cause smug elitism wins at the end of the day.

The biggest problem is that the site owner and whoever moderates this place are completely silent in all of this. I think it would help if Laurel and Manu started talking to us openly in public and we could engage in open discussion with them.
These threads always turn into arguments on what AI is, and telling people their writing is crap if it's being false flagged. This is truly a moral panic over AI, and it's gotten ridiculous. Search the forums on here and you'll see some of the insanity and also how unreliable these AI detection tools are that we are being held to. Someone ran historical US government documents through one and it came up over 70% AI detected on the basis of James Madison's writing. The biggest problem with the arguments is people don't want to meet in the middle on a shared belief and it turns into just endless debating over what AI even is.

The only advice I can really give you based on my frustrations and getting a few stories pushed through; is to lean on a thesaurus and constantly check for repeated word usage. An overuse of key words tends to trigger it, and the only solution is to change some of those words out by using a thesaurus. Something as simple as patterns in your sentence structure tends to trip the detection tool.

Hopefully this madness ends sooner than later, cause it has become completely ridiculous. You are not alone. Many, many, many of us have had these issues.

Not being adversarial, but what would you do about AI submissions if you were running the site (I don’t mean communications about the subject, I agree on that point, I mean actual submissions)?

Emily

 

[STrent]

Not being adversarial, but what would you do about AI submissions if you were running the site (I don’t mean communications about the subject, I agree on that point, I mean actual submissions)?

Emily

Are there really that many stories made with AI though that can be proven?

Virtually all of mine that were rejected have been flagged as having potential AI generated content, not that they actually had any AI generated content.

Which is part of the overall problem, the stories are submitted in good faith, there’s not sufficient evidence than what has been said in messages back to us.

 

[ShelbyDawn57]

that's not what he was saying. He was saying that those being rejected for possible AI have been hurled with insults to their craft. That is not true. I have not seen one single comment (perhaps I've missed one?) where someone said "you got rejected 'cause you suck!" so that is not at all true.

So, we agree.

 

[EmilyMiller]

Are there really that many stories made with AI though that can be proven?

Virtually all of mine that were rejected have been flagged as having potential AI generated content, not that they actually had any AI generated content.

Which is part of the overall problem, the stories are submitted in good faith, there’s not sufficient evidence than what has been said in messages back to us.

My question was - assuming, as seems reasonable given the world we now live in, that there are a significant number of actual AI submissions - what would you do?

Emily

 

[STrent]

My question was - assuming, as seems reasonable given the world we now live in, that there are a significant number of actual AI submissions - what would you do?

Emily

Maybe have the discussion with the membership about what should be implemented and why it’s being implemented before putting something in place that creates a seemingly large amount of false positives.

Would I put in AI detection software? Yes, if one existed that didn’t seem in some way flawed.

I’ve had a number of these suggested to me for checking my own work before submitting here and none of them marry up to the other on analysing the same work.

The only thing that has helped me is not using grammarly before submission. That runs the risk of sub-par formatting, in my view, for those of us who haven’t got the time or maybe the skills to self edit in the same way.

 

[EmilyMiller]

Maybe have the discussion with the membership about what should be implemented and why it’s being implemented before putting something in place that creates a seemingly large amount of false positives.

Would I put in AI detection software? Yes, if one existed that didn’t seem in some way flawed.

I’ve had a number of these suggested to me for checking my own work before submitting here and none of them marry up to the other on analysing the same work.

The only thing that has helped me is not using grammarly before submission. That runs the risk of sub-par formatting, in my view, for those of us who haven’t got the time or maybe the skills to self edit in the same way.

There is no world in which AI detection is perfect. There is no world in which it’s even close to something reasonable, say 80% accuracy. So the only other option is do nothing.

Think of it as a medical test that isn’t wholly accurate. It has flaws positives and false negatives. Do you not take the test if you know you are at risk?

I agree that communications could be better. I think a better way to handle people’s appeals would be good.

But I don’t support doing nothing.

Emily

 

[STrent]

I understand why the scans need to be done and I understand the reasoning behind it all, but the lack of communication and the inconsistencies between the rejections are the issue.

I've done everything recommended of me - not using Grammarly, messaging Laurel, putting it through an AI detector and making changes - but I'm still stuck on the rejection merry-go-round. I don't know what is causing the kick-backs and I'm not getting told how I can fix it.

It needs to be done, yes, but when people are following the rules to the letter but are still stuck with no way of knowing if the part they put their heart and soul into will ever be published, that's where the issues are coming from. I shouldn't have to wait over a month to have a part published because no one is communicating with me.

That is where the issue is.

@EmilyMiller the above sums up my thoughts, the length of time waiting plus the lack of significant insight on what’s causing the AI rejections are the two biggest issues.

I think “do nothing” actually has its own attractions as I doubt it’s anywhere near 80% accurate, if actually accurate in any way.

But hey ho, it’s Laurel’s site. The final decisions lie with her. I just wish I’d had more of a steer on the AI rejections.

 

[CockSparrow]

I had a chat with one of my young tech-savvy friends. He suggested that perhaps Laurel is using an automated pre-screening tool (which would make sense). And if the author has used Grammarly, or some similar machine-trained editing software, the pre-screen tool is likely to shout ‘AI off the starboard bow!’ Just a thought.