I was just staring at its first response for a solid minute, wondering if I was wrong or if the AI was! 
The AI was like, “Oh wait, that’s my mistake, not yours. My bad!”
Ah yes, the overworked AI is already behaving like an overworked human. Brilliant progress.
I really dislike Gemini. It initially denied you because of its safeguards against discussing potentially sensitive topics, like mental illness. Meanwhile, ChatGPT is being used successfully for therapy by tens of thousands of people.
Not going to lie, Gemini is really disappointing compared to ChatGPT, despite having access to Google Search. For example, I was trying to find out which model of the F-4 Phantom was the first to have a ballistic computer installed. Gemini initially didn’t understand which F-4 I was referring to, then said it was “hard to know,” and kept switching between claiming it didn’t know or didn’t have the information.
In contrast, ChatGPT immediately provided the model, the name of the computer, and the gun sight, all accurately.
The worst part is how poorly Gemini understands my speech. I have to speak slowly and carefully with it, while ChatGPT can handle normal speech, slang, and even correct punctuation when I pause.
I find it odd when it behaves that way. It seems to be programmed to avoid giving medical therapy or diagnoses, but it also tries to comply with whatever you say, almost like it’s being trained. This can sometimes cause confusion.
It’s a joke—I ask it about medicine dosages, and it won’t provide any information. It’s useless in that regard. ChatGPT, on the other hand, gives me detailed, referenced, and off-label information every time.