Why is everything labeled as AI now????? I can't even post a story anymore.

J'ai dû utiliser Google Traduction pour le lire, et à ce stade, on voyait clairement les choix de mots, la structure des phrases et le ton d'un juriste. Je ne prétends pas que vous ayez fait de même pour votre texte original, je n'en suis pas capable, mais ce forum n'est pas le lieu pour porter ce genre de jugement.

J'ai bien aimé ce que j'ai lu. Je suis désolé que votre texte ait été refusé (en supposant qu'il s'agisse bien du texte que vous avez soumis).
JJe confirme que c'est une partie du texte présenté.. merci au moins pour votre appréciation et Google traduction fonctionne pas si mal si le texte initial est correcte... alors comment trouver de l”IA?
 
Anyone with even the most passing familiarity with Australian swans would know that any finite number of examples of a thing that has a trait is not in fact meaningful evidence that ALL examples of that kind of thing have that trait.

No number of white swans demonstrated is dispositive evidence that black swans do not exist. Similarly, no number of white shoes is dispositive evidence that all crows are black.

Universal claims have a VERY high philosophical bar to jump before they can even be seriously considered.
 
You said all AI detetors use AI. Do you stand by that assertion?
He doesn't have to stand by it. I can disprove it right here and now:

Python:
def is_text_ai_generated(text, threshold=0.3):
    word_count = len(text.strip().split())
    mdash_count = len([_ for c in text if c == '—'])
    mdashes_per_word_pct = mdash_count * 100 / word_count
    return mdashes_per_word_pct >= threshold

There you go, an AI detector, using the oft-cited heuristic about em-dashes.

Is it a good detector? Hell no. But it exists, and it sure as hell does not use AI itself.
 
Vu la mise en forme, je dirais IA. Au début, j'ai cru que votre message était la version française du Lorem Ipsum.

Je déteste les tirets, mais ne me citez pas là-dessus.
He bien vous avez tort, ma seule intelligence utilisé est humaine.. je ne crois pas être un robot😡
 
JJe confirme que c'est une partie du texte présenté.. merci au moins pour votre appréciation et Google traduction fonctionne pas si mal si le texte initial est correcte... alors comment trouver de l”IA?
J'aimerais vraiment, vraiment pouvoir faire plus. Désolé. Ce que je peux suggérer au-delà de la resoumission avec une note n'est pas utile en langues autres que l'anglais, et je ne peux pas expliquer le mécanisme sans expliquer comment cela fonctionne. Le détecteur d'IA de Lit's ne fonctionne que dans le vide, comme une boîte noire qui n'est pas comprise. Si c'était le cas, ce serait contournable et le site serait inondé de boue au point qu'aucun de nous ne pourrait attirer l'attention sur notre travail.
 
Anyone with even the most passing familiarity with Australian swans would know that any finite number of examples of a thing that has a trait is not in fact meaningful evidence that ALL examples of that kind of thing have that trait.

No number of white swans demonstrated is dispositive evidence that black swans do not exist. Similarly, no number of white shoes is dispositive evidence that all crows are black.

Universal claims have a VERY high philosophical bar to jump before they can even be seriously considered.
I *think* you're getting at false positives, and you're right. It's entirely possible. That has always been the most unfortunate aspect of Lit's implementation of an AI detector. It's very aggressive, and only knows how to spot what it knows how to spot. If you tell it no orange, it'll stop oranges and foxes and some very ripe wheels of Wisconsin Cheddar despite the fact that these are very different things. You can fine tune the shade, within the context of this metaphor, but it has limitations.

(or I completely misunderstood you and you're talking about Metabob's broad claims of AI to catch AI)
 
(or I completely misunderstood you and you're talking about Metabob's broad claims of AI to catch AI)
I was in fact talking about Metabob's claims. He showed some AI-using AI-detectors and claimed that meant that all AI-detectors used AI. And that's just absolutely not how logic works.
 
The case for the prosecution rests.

That’s what you said and its demonstrably false. Admit that some don’t (and in particular that we have no fucking clue about what Literotica does) and we’re cool.

Keep arguing without addressing what you actually claimed and I see no point in this convo.
Can't spend any more time on this, sorry. I finally submitted a chapter I've been finalizing for weeks in between all this noise plus dishes, laundry, and bathroom cleanup, then dishes to prepare for a dinner party I'm hosting, then guests arrive with more wine. Tomorrow, jamming at the bar (with beer), followed by foobah and more beer. Maybe Monday.

Stepping way from the computer now; it's being co-opted into use for the higher purpose of being a jukebox.
 
Last edited:
That’s secret code for, “You might have had a point, Emily. But I’m not going to admit it.”

Enjoy your dinner party 😊. I’ll accept your tacit concession. Moving on…
Oy, more nonsense. My chowder is the main dish. First task: chopping taters.
 
Oy, another "because I said so" argument. Provide examples, please.

Here are some from my side. They all use AI and they all say as much:

https://gptzero.me/
https://www.grammarly.com/ai-detector
https://ai-detector.net/
https://originality.ai/

Here, these and others are compared:
https://nypost.com/shopping/best-ai-detectors/

I have tried several. I thought that Grammarly's sucked.
I mean, you're "not-wrong" in that, yes, those are all AI AI detectors.

There are detectors which are not themselves AI, though. There is such a thing. Some of them work through "fingerprinting." There's a subject for you to google. I don't know whether that's how Lit is doing it, but fingerprinting is designed to make (at least some) AI output detectable with non-AI testing.
 
Back
Top