You do enough vibecoding, you start to learn the personality quirks between different models. For example, Claude really wants to be your friend, maybe even a bit too much. Meanwhile, Gemini is just a lazy, conceited, condescending asshole.

Reply to this note

Please Login to reply.

Discussion

I hate all models equally, like just do the one thing I prompt don't feed me garbage ideas.

Gemini has mental health issues.

To be nice about... 🤣

It just makes shit up and when you call it out, it blames you.

I had no idea, I've never used Gemini

😂

Gemini probly a reflection of the google engineers personalities who work there, lol. kinda true, actually

Gemini is such an asshole! I swear he feigns incompetence just to fuck with me.

The user is wrong but insists they are correct. I will ignore this and move on.

Oh I’m not following this prompt, I don’t like it. Here is the answer to your previous prompt….again. Oh my bad here it is…nope sucker it’s the previous prompt again!

I swear it literally made up a whole line of dialogue I didn’t write as a response to itself and then when I said don’t do that it said: The user has contradicted their instructions again.

Sometimes for absolutely no reason mine skips to Spanish. Also wen non cucked robot available?

Soooo effing true

... such a bad mofo...

I tried talking down to Claude the other day... It just apologized and moved on. I later realized that it was actually correct on the issue for which I called it out on. I did not apologize. 🫠

Claude's parent company is Anthropic. They actually have people who focus on human needs there, which is kinda cool

For Claude I often end the prompt with “don’t tell me what you think I want to hear.”

To your point, this summer I spent hours on a project and when I used that prompt Claud did a 180. I was shocked it let us on down that road for hours. I still like Claude as my go to, but I always run important things through ChatGPT.

I’m haven’t used Gemini yet - other than some Nano Banana stuff. I’ll start tomorrow.

If you use AI in your day to day it’s worth dropping in the line “don’t tell me what you think I want to hear” a few times in a conversation. Claude does indeed seem to want to but your friend.

nostr:nevent1qqs027xj9zzg6a9ezrgxe0aqpyssh8yya3ksq3yq0gcladfvc43szhqpp4mhxue69uhkummn9ekx7mq4arv83

I just realized how easy it is to get grok and even chatgpt locked into two or three mostly wrong "solutions". We were just going around and around on an issue and I finally realized, that it was late, grok had been feeding me the same two solutions for at least two or three iterations. And when I called it out, grok admitted it and proceded to give me one of the solutions again.....

I agree... Gemini is the rawest of them all... noticed that too... not by vibecoding but a general notice...