Clarification regarding AI policy

[Questianor]

I understand there's no 100% surefire way to distinguish between AI-written/assisted and purely human writing and that enforcement would inevitably never be complete and ocassionally suffer from false positives/negatives and subjective determinations of the moderators. I also think, however, that this describes most moderation activity anywhere on the internet. I think it would be well worth it for the site's usability and long-term appeal to at least try to crack down on or identify and filter the really obvious AI slop. You know, the stuff that updates with a new full-length chapter every single day where every other sentence is "it wasn't just X— it was Y". All this slop crowding up the pages is just a bad user experience, at least in my humble opinion. If it's not as obvious as that it's not so big a deal so there's not much need to try to enforce anything anyway.

That's my two cents, had to throw it out there even if it doesn't seem likely to make a difference.

 

[NuclearBird]

I understand there's no 100% surefire way to distinguish between AI-written/assisted and purely human writing and that enforcement would inevitably never be complete and ocassionally suffer from false positives/negatives and subjective determinations of the moderators. I also think, however, that this describes most moderation activity anywhere on the internet. I think it would be well worth it for the site's usability and long-term appeal to at least try to crack down on or identify and filter the really obvious AI slop. You know, the stuff that updates with a new full-length chapter every single where every other sentence is "it wasn't just X— it was Y". All this slop crowding up the pages is just a bad user experience, at least in my humble opinion. If it's not as obvious as that it's not so big a deal so there's not much need to try to enforce anything anyway.

That's my two cents, had to throw it out there even if it doesn't seem likely to make a difference.

Click to shrink...

Literally all you have to do is to look out for emdashes and reddit prose (for some godforsaken reason they are all trained on reddit style prose) and then a quick sweep of any positive results for Grammarly users (why anyone uses that garbage service is beyond me) and that's most of the work done.

 

[Lady Naga]

I will never give up my emdashes, you can take them from my cold dead hands.

 

[Lady Naga]

Unironically though AI is an issue for two people: paranoid schizos who can't read a chapter without thinking the person has been replaced by a robot, and people who write so poorly that most people arent bothering to begin with. Its a non-issue on both sides and I'm glad QQ is being sane about their rules. You read fanfiction. If you want perfect media tailored to you, find a bunch of goodread reviews and buy a book.

 

[LazyAutumnMoon]

Literally all you have to do is to look out for emdashes and reddit prose (for some godforsaken reason they are all trained on reddit style prose) and then a quick sweep of any positive results for Grammarly users (why anyone uses that garbage service is beyond me) and that's most of the work done.

Click to shrink...

To add on:

Go about 1.5 years back, and the em dash was a sign of real writing. It's useful in many different situations - adding details to a sentence, interrupting sentences, placing actions in between dialogue. All the complicated stuff that makes writing fun and engaging. Best of all, iterations of AI programs back then just didn't know what to do with them, and most of them don't.

But then, if that's the case, how come you've never seen it before? Because by the time a writer is confident enough to start using em dashes, they tend to know the rules and conventions behind writing. They use these marks sparingly and in ways that flows naturally, thus leading your your eyes to glaze over them in the same way that a reader's eyes pass over the 'said' dialogue tag.

Recent versions of AI changed that, causing an explosion of em dashes in the past year. Suddenly, they're everywhere!

Of course, the programs still don't understand the proper way to use them, and like an amateur the AI overuses it to the point where the average reader notice them. Thus, everybody jumped on it as the sign of AI writing, launching a hundred articles and a thousand reddit threads telling people to look for the em dash.

Except, that's not quite right. What you'd need to consider when you read a story are questions like:
Does the writer show that they know how to use the em dash?
Does the writer know when to stop using the em dash?
Do they not know what a comma is? It would have been so much better there!
...And such things.

My advice is to find a good writer who you know is not using AI to write their chapters, and ask them to explain the many signs of AI writing rather than relying on advice condensed to the layman-level, which, though containing a grain of truth, simplifies it to the point that the advice becomes unusable.

Basically, there's no quick and dirty way to spot AI writing, hence the creation of this thread in the first place.