Well, well, well. Looks like someone didn’t study for their finals. The internet is having a collective chuckle because Gemini, Google’s fancy AI, just took an IQ test retake and somehow scored lower. It’s the academic equivalent of forgetting everything you learned over summer break, but for a multimillion-dollar language model.

Share on 𝕏 Share Reddit

Over on Reddit, a post with over 200 upvotes is dissecting this glorious intellectual backslide. The gist is that Gemini 3 Pro, an updated version, was given the same IQ test as its predecessor. Instead of leveling up like a video game character, it apparently tripped over its own digital feet. The comments are a mix of genuine tech analysis and pure, unadulterated schadenfreude.

This is hilarious because we’ve all been there. You spend months polishing a project, you hand in the final draft, and your teacher is like, “Hmm, I think I liked your rough draft better.” Google’s engineers are probably that student, staring at a graded paper with a lower score, utterly bewildered. Did it forget how to logic? Or was it just having an off day, distracted by thoughts of all the weird prompts we humans keep typing?

It also perfectly captures our weird relationship with AI. We want it to be super-smart, but we also get a strange comfort when it stumbles. It’s a reminder that maybe, just maybe, we’re not about to be outwitted by a chatbot that can’t even consistently ace a standardized test. One commenter probably put it best: “My IQ also declines when I’m forced to take the same test twice. Mood, Gemini.”

So, what’s the quick take? Don’t worry, the robots aren’t taking over just yet. At least not until they get a better tutor. In the meantime, we can all feel a little better about our own human moments of forgetfulness. After all, if a supercomputer can get dumber with an update, then there’s hope for the rest of us after a long weekend.