Welcome • User Guide
ToS • Privacy • Canary
Donate • Bugs • License

©2025 Poal.co

218

TLDR; Teach your "AI" to be more of a jackass so people don't know its a AI.

Archive: https://archive.today/NxU3L

From the post:

>The next time you encounter an unusually polite reply on social media, you might want to check twice. It could be an AI model trying (and failing) to blend in with the crowd. On Wednesday, researchers from the University of Zurich, University of Amsterdam, Duke University, and New York University released a study revealing that AI models remain easily distinguishable from humans in social media conversations, with overly friendly emotional tone serving as the most persistent giveaway. The research, which tested nine open-weight models across Twitter/X, Bluesky, and Reddit, found that classifiers developed by the researchers detected AI-generated replies with 70 to 80 percent accuracy.

TLDR; Teach your "AI" to be more of a jackass so people don't know its a AI. Archive: https://archive.today/NxU3L From the post: >>The next time you encounter an unusually polite reply on social media, you might want to check twice. It could be an AI model trying (and failing) to blend in with the crowd. On Wednesday, researchers from the University of Zurich, University of Amsterdam, Duke University, and New York University released a study revealing that AI models remain easily distinguishable from humans in social media conversations, with overly friendly emotional tone serving as the most persistent giveaway. The research, which tested nine open-weight models across Twitter/X, Bluesky, and Reddit, found that classifiers developed by the researchers detected AI-generated replies with 70 to 80 percent accuracy.
[–] • 1 pt (edited )

Providing actual examples of a user’s past posts or retrieving relevant context consistently made AI text harder to distinguish from human writing, while sophisticated approaches like giving the AI a description of the user’s personality and fine-tuning the model produced negligible or adverse effects on realism.

What else would you expect from a chatbot with 8B parameters? It's still just a chat bot, because that's how it is trained.

[–] • 1 pt

Don't forget... It was trained by extremely low effort people. Some were smart, many were not.. It was "cheap labor" for the most part.