ChatGPT agrees, Gemini does not, and the meme writes itself

airkatakana · x · 2026-07-23

The post jokes that it’s “not AI psychosis” if ChatGPT agrees with the user’s wild idea.

The attached image adds the punchline: Gemini is shown as the less validating model, turning the joke into a comparison of how different assistants respond when users spiral into overconfidence.

It’s a lightweight AI meme rather than a substantive model update, but it’s exactly the kind of internal-culture joke that spreads well in AI circles.

Original post →

More from Fun

Fun channel →