[nluken]

The giveaway is almost always an over-dependence on "Not 'x' but 'y'" structure. Even when the author changes the wording so that the phrase doesn't read exactly like that, they tend to leave the structure intact, and the bots really like to lead with the inverse of what the author wants to say to create contrast.

A human author might have used this technique once to really emphasize a strong point, but today's LLMs use it so often that it loses its emphasis, and instead becomes a distinct stylistic fingerprint.

[myself248]

As a human who's written like that for decades, this witch-hunt makes it really hard to participate online anymore. This was a standard expository device taught in school, some of us latched onto it, and now it's impossible to argue that we're human because someone's conviction that we're AI is just an opinion and there is no evidence that will convince them, nor do they deserve the effort it would take even if it were possible.

What's the path forward?

[nluken]

Keep using that structure. The bots use it so much, and in ways that human authors wouldn't, that it becomes self-parodying. As long as you're not using the same device over and over again in the same piece you'll be fine.

[marstall]

yeah it's a crazy trend. suspicion that everything is AI is a poison to forums. have no idea how this moves forward. secretly hoping it destroys the entire internet, and rapidly. so we can go back to conspiring in smoke-filled rooms.