KI-Modelle mit den niedrigsten Halluzinationsraten: Zhipu AI- und Gemini-Modelle haben die niedrigsten Raten

Yeah but Gemini is fucking awful at anything I tell it to do. I’m not exaggerating when I say that it almost never does anything right when i tell it to do anything. ChatGPT on the other hand, is almost always right. It’s literally a night and day difference. If I use Gemini I literally always end up yelling at my screen saying “How can this thing be so bad?” I want to go to the Google offices and hurt somebody, so I almost never use it anymore. I have no idea how anybody at all uses it.

I guess that means that Gemini and Zihpu AI get to share the first-place coffin, but only if they can somehow manage to succeed at the final event in the Upper-Class Twit of the Year competition. But that’s a high bar to clear apparently, so I’m not expecting much. But I hope they both manage to win *that* prize because boy howdy, they deserve it!