Seeking Advice - German Literotica Community Facing Mass AI Rejections

[ChloeTzang]

I already gave Djinn68 the only advice I've been able to figure out.

And you can't do that for every individual affected - and THAT's the issue. LIT has given no meaningful advice on how to avoid this and what to do if it IS a false -positive. PM's to Laurel go nowhere

The only real action I can see that is viable is open a Feedback Report / Bug Ticket along with a link to the story and wait for Manu / Laurel to take action.

A lot of people are reporting AI Rejections, but I don't believe they are necessarily providing that feedback / bug report to the site so Literotica may well not be aware of the extent of the impact.

I did the best I could to give people a path foward if they hit this - as an alternative to giving up and moving to another site. I think everyone would welcome steps they could take, however generic, to avoid this issue.

I first worked out Lit's AI detector in mid-2024, and sat on my hands saying nothing for over a year because I couldn't figure out how talk about it. It is a black box that only works because it is a black box.

My head explodes just thinking about it! I would have no idea even where to start. And obviously from all the people raising this as an issue (and those are the ones that speak up) no-one else really has any clues.

I just go back to - everyone who hits this should raise a Feedback Report with the site and open a bug report - at the very least, this highlights the impact

And LIT really needs to explain what authors should do to (a) avoid this and (b) what to do if you get a false-positive


Personally I have never had this - but I don't even use Grammarly. Spellcheck is it for me.

 

[AwkwardMD]

Unfortunately, I seem to be the only person to reverse engineer it (or at least the only one who will say anything). All anyone has is me standing here with my sandwich board sign saying "Trust Lit's AI detector", and I understand why some people wouldn't.

 

[Djinn68]

All anyone has is me standing here with my sandwich board sign saying "Trust Lit's AI detector, even if it gives you false positives", and I understand why some people wouldn't.

FTFY.

 

[AwkwardMD]

My head explodes just thinking about it! I would have no idea even where to start. And obviously from all the people raising this as an issue (and those are the ones that speak up) no-one else really has any clues.

I lucked into privileged information for a starting point

I just go back to - everyone who hits this should raise a Feedback Report with the site and open a bug report - at the very least, this highlights the impact

And LIT really needs to explain what authors should do to (a) avoid this and (b) what to do if you get a false-positive


Personally I have never had this - but I don't even use Grammarly. Spellcheck is it for me.

I have already addressed A. B is the tricky part, because how to get around a false positive and how it failed and how the detector works are all intertwined.

 

[ChloeTzang]

Unfortunately, I seem to be the only person to reverse engineer it (or at least the only one who will say anything). All anyone has is me standing here with my sandwich board sign saying "Trust Lit's AI detector", and I understand why some people wouldn't.

It shouldn't be on you to do that.

Literotica really needs to step up here and explain, and give authors some guidelines.

 

[AwkwardMD]

FTFY.

You're not wrong, unfortunately.

 

[ChloeTzang]

I lucked into privileged information for a starting point


I have already addressed A. B is the tricky part, because how to get around a false positive and how it failed and how the detector works are all intertwined.

So what is (a) - I guess I missed that.

 

[AwkwardMD]

It shouldn't be on you to do that.

Literotica really needs to step up here and explain, and give authors some guidelines.

I suspect they haven't for the same reason I didn't at first. Any amount of talking about "it" is a terrific risk.

 

[AwkwardMD]

I'm talking now because I believe in the site and its platform, but I'm just one grumpy old lady.

 

[Djinn68]

Any amount of talking about "it" is a terrific risk.

As ist NOT talking about "it". If Lit can't help honest authors overcome these kind of obstacles, I won't remain the only one to leave...

 

[ChloeTzang]

So what is (a) - I guess I missed that.

Ohhhhhh - this thread?

https://forum.literotica.com/threads/awkwardmd-ai-rejection-help-desk.1635750/#post-101209655

 

[AwkwardMD]

(a) avoid this

Write it yourself. Edit it yourself (or get someone you trust to help you). Submit what you wrote.

and (b) what to do if you get a false-positive

This is the part I can't do. I mean, I *know*, but I can't talk about it.

 

[PennyThompson]

Literotica really needs to step up here and explain, and give authors some guidelines.

As with all kinds of op sec, explaining the process risks making the process ineffective

Telling good actors how to bypass the system inevitably lets bad actors do the same.

It seems like the only solution is individualized human review and appeals, and the human bandwidth that the amount of content flowing through Lit daily requires doesn't seem to exist.

 

[AwkwardMD]

As ist NOT talking about "it". If Lit can't help honest authors overcome these kind of obstacles, I won't remain the only one to leave...

You're not wrong, and it’s unfortunate. The threat of the site being overwhelmed with slop is real.

 

[Djinn68]

You're not wrong, and it’s unfortunate. The threat of the site being overwhelmed with slop is real.

As is the threat of losing human authors. Scylla and Charybdis.

 

[ChloeTzang]

Write it yourself. Edit it yourself (or get someone you trust to help you). Submit what you wrote.

Yep, that's what I do and I have had no issues with AI detection. Yet.

Seems a lot of people have had AI Rejections and have also written it themselves tho, so that doesn't help them.

 

[AwkwardMD]

As is the threat of losing human authors. Scylla and Charybdis.

The really unfortunate part of this equation is that, especially in the years since the pandemic, there is no shortage of new writers trying their hand at fantasy. Losing a few veterans here and there is a risk worth taking as far as the site is concerned. It's brutal calculus, and I don’t like it, but I also know I don't matter to the site either. Accepting that helped me.

 

[AwkwardlySet]

Unfortunately, I seem to be the only person to reverse engineer it (or at least the only one who will say anything). All anyone has is me standing here with my sandwich board sign saying "Trust Lit's AI detector", and I understand why some people wouldn't.

I'm sure you know this isn't a matter of trust, really. I trust you when you say you believe you know how Lit's AI detector works. I trust you when you say you believe it works well.

What I don't "trust" in this case is your expertise to make that judgment, especially in the light of what we see every day.
Top software companies and top universities can't seem to figure out a reliable AI detector. But I am somehow supposed to believe that Laurel and Manu came up with it? Come on.

If the two are such top-notch programmers, they should patent their detector and make millions selling it, since it would be the only one working well.
Then again, one would think that two such programmers could make some basic functionalities work on their very website, for a change.

Once again, this isn't a matter of trusting that you're being truthful. We've had both clashes and fruitful discussions, and even at its worst, I never thought you were being less than truthful from your standpoint. But I'm also quite sure you understand how your claims must seem to anyone who can think critically.

 

[AwkwardMD]

Seems a lot of people have had AI Rejections and have also written it themselves tho, so that doesn't help them.

Like I said (and this includes Djinn), every instance of a false positive that I've come across (that I believe is genuine) involved someone adding steps to my very basic outline for how to avoid a rejection.

 

[AwkwardMD]

I'm sure you know this isn't a matter of trust, really. I trust you when you say you believe you know how Lit's AI detector works. I trust you when you say you believe it works well.

What I don't "trust" in this case is your expertise to make that judgment, especially in the light of what we see every day.
Top software companies and top universities can't seem to figure out a reliable AI detector. But I am somehow supposed to believe that Laurel and Manu came up with it? Come on.

If the two are such top-notch programmers, they should patent their detector and make millions selling it, since it would be the only one working well.
Then again, one would think that two such programmers could make some basic functionalities work on their very website, for a change.

Once again, this isn't a matter of trusting that you're being truthful. We've had both clashes and fruitful discussions, and even at its worst, I never thought you were being less than truthful from your standpoint. But I'm also quite sure you understand how your claims must seem to anyone who can think critically.

I know. I can only do what I can do.

 

[Djinn68]

Accepting that helped me.

As it did for me - and that is why I move on. Because, in the end, no one here could really help me. I have accepted that. I had a good time here at Lit, two wonderful years. Many encouraging comments, over a hundred followers, which is - as I only wrote erotic sci-fi, not that popular on the german side of LIT, is (at least for me), quite an achievement. I won't be able to gain new readers here at Lit, because I'll let them delete all my stories here - except the last one, which is erotic, but also an explanation, why I left. With a hint to my followers, likers and commenters, about the new whereabouts of my stories...

 

[PennyThompson]

I know. I can only do what I can do.

 

[ChloeTzang]

So to summarize, we really are no clearer than we were at the start of this thread.
- Write it yourself.
- Don't use a tool like GRammarly or whatever.
- Spellcheck it and that's about it.
- Submit and Cross your fingers.

If you get an AI Rejection after all that,
Step One:
(a) After getting a rejection, resubmit with a note to the admin saying "I did not use AI to write this" (please note, not an option for German-language users)
(b) if rejected again, move to Step Two

Step Two
- open a Feedback Form,
- Open a Bug Ticket,
- in the above, state your case, provide the URL to the rejected story and
- hope Literotica gets around to looking at it
- and if you believe in miracles, sprinkle holy water over your laptop and PM Laurel (the equivalent of praying to the Lord for intercession - sometimes it works LOL)

 

[AwkwardMD]

So to summarize, we really are no clearer than we were at the start of this thread.
- Write it yourself.
- Don't use a tool like GRammarly or whatever.
- Spellcheck it and that's about it.
- Submit and Cross your fingers.

If you get an AI Rejection after all that,
- open a Feedback Form,
- Open a Bug Ticket,
- in the above, state your case, provide the URL to the rejected story and
- hope Literotica gets around to looking at it
- and if you believe in miracles, sprinkle holy water over your laptop and PM Laurel (the equivalent of praying to the Lord for intercession - sometimes it works LOL)

After getting a rejection the first step is to resubmit with a note to the admin saying "I did not use AI to write this"

Unfortunately, that step is functionally useless to the Germans asking for help. I think non-English reviews are beyond Laurel's abilities.

 

[ChloeTzang]

After getting a rejection the first step is to resubmit with a note to the admin saying "I did not use AI to write this"

Unfortunately, that step is functionally useless to the Germans asking for help. I think non-English reviews are beyond Laurel's abilities.

Updated my steps to reflect that