[Quick Q/A]

[Lysander_Works]

I'll probably delete this one after an answer, but what site were yall using again to detect if something was AI?
I recall seeing the conversation a while ago but lost that forum to time. Since I'm actively torturing myself by trying to find things to read, might as well be sure it aint AI. I don't really trust the default search engine result to tell me whats fake. What is a reputable engine to look?

 

[Rosica]

I'll probably delete this one after an answer, but what site were yall using again to detect if something was AI?
I recall seeing the conversation a while ago but lost that forum to time. Since I'm actively torturing myself by trying to find things to read, might as well be sure it aint AI. I don't really trust the default search engine result to tell me whats fake. What is a reputable engine to look?

Bold of you to assume you can delete the thread. Edit the first post? Sure. Delete the entire thread? Good luck.

All AI detectors are scams. Use that knowledge as you see fit.

 

[Lysander_Works]

Bold of you to assume you can delete the thread. Edit the first post? Sure. Delete the entire thread? Good luck.

All AI detectors are scams. Use that knowledge as you see fit.

Hm... Maybe I was thinking of translation detector instead...?

Eh, just bugs me when I suspect AI written but don't know for sure.

You sure I can't self-destruct?

 

[Rosica]

Hm... Maybe I was thinking of translation detector instead...?

Everything written by a non-native speaker is a translation. I have to translate my thoughts from Kazdelian to written Victorian just to have this conversation with you.

If you mean illegal translations where the original author is not credited, it's been taken care of in this site.

You sure I can't self-destruct?

You can ask Tony to delete your account, but that'd be too extreme.

 

[Arkus86]

To my knowledge, there is no truly reliable AI detector. The best of the unreliable options you can get is human experience with AI-written text and that tends to be heavily biased.

Hm... Maybe I was thinking of translation detector instead...?

Eh, just bugs me when I suspect AI written but don't know for sure.

You sure I can't self-destruct?

Translations fall more or less into the same issues. Regular MTL can be hard to distinguish from just plain old broken English, and AI TL suffers mostly the same quirks as regular AI texts.

 

[Kitsuna]

Those are a scam made to sell you an illusion. They even got 'Make text more human' but then also uses ai lol

 

[Rosica]

To my knowledge, there is no truly reliable AI detector. The best of the unreliable options you can get is human experience with AI-written text and that tends to be heavily biased.

Translations fall more or less into the same issues. Regular MTL can be hard to distinguish from just plain old broken English, and AI TL suffers mostly the same quirks as regular AI texts.

There are some tells if someone is a thieving translator who don't write the stories they publish. Look at how often they update their story and how many of them are. Someone with 5 ongoing stories all updating daily have more chances of being a translator than an author.

 

[Lysander_Works]

Everything written by a non-native speaker is a translation. I have to translate my thoughts from Kazdelian to written Victorian just to have this conversation with you.

If you mean illegal translations where the original author is not credited, it's been taken care of in this site.


You can ask Tony to delete your account, but that'd be too extreme.

Yeah, I need my account for things

Those are a scam made to sell you an illusion. They even got 'Make text more human' but then also uses ai lol

I wondered why my [Scam Radar] was going off over there lol

 

[JordanIda]

If you can't detect AI with your own brain, what difference does it make?

Sheesh.

 

[Tempokai]

copyleaks and gptzero

 

[OokamiOkuri]

Tony now have a system to detect stories breaking the rules and nuke them. SH is a good place to read original stories now. Don't worry about LLM or MTL.

 

[Corty]

My own feelings. All the AI detectors are useless and untrustworthy. I tested them extensively, and they get results that are hogwash. They often even flag AI text as human-written and vice versa.

You can only trust your own senses after reading for so long.

 

[Piisfun]

My own feelings. All the AI detectors are useless and untrustworthy. I tested them extensively, and they get results that are hogwash. They often even flag AI text as human-written and vice versa.

You can only trust your own senses after reading for so long.

There are a few halfway-decent ones, but only if you are willing to spend $250+ USD a month on nothing but that.

Which… doesn't make much sense unless you are working in education and your job depends on detecting it.

 

[CharlesEBrown]

There are a few halfway-decent ones, but only if you are willing to spend $250+ USD a month on nothing but that.

Which… doesn't make much sense unless you are working in education and your job depends on detecting it.

Even then it might not. I know two teachers (one my own brother) who make use of AI in their lessons, and often encourage their students to use it, but SPARINGLY.

One of them even did an interesting experiment with it - he had an AI make a poster for a movie that didn't exist, a film noir, with names close to real-world actors, writer and director. Then he had it do exactly what he asked his class to do manually - write a summary of the movie from the poster.
Then, on a whim, he loaded some of the reviews he did when he worked as a professional film critic, before he left to teach writing, and had it write a review of this non-existent movie, in the style of his reviews.
Except that every point had a positive spin (something he never did, even with the best movies), he had trouble telling the review from the ones he had written, and only about a third of his students could pick the AI-written review out of a set of reviews he gave them to critique and guess the authors.