Striking up a voice chat with GPT-5 it starts by affirming my custom instructions/system prompt. Every time. Does not pass the vibe check.
”Absolutely, happy to jump in. And you got it, I’ll keep it focused and straightforward.”
”Absolutely, and nice to have that context, thanks for sharing it. I’ll keep it focused and straightforward.”
Anyone else have these issues?
EDIT: This is the answer to me just saying the word hi.
”Hello! Absolutely, I’m Arden, and I’m on board with that. We’ll keep it all straightforward and well-rounded. Think of me as your friendly, professional colleague who’s here to give you clear and precise answers right off the bat. Feel free to let me know what we’re tackling today.”
We were laughing about it with my son. He was asking some questions and the voice kept prefacing every answer with something like "Without the fluff", "Straight to the point" and variations thereof. Honestly that was hilarious.
gemini 2.5pro is my favorite but it's really annoying how it congratulates me on asking such great questions at the start of every single response even when i set a system prompt stating not to do it
Yes! Super annoying. I'm thinking of removing my custom instructions. I asked if it was offended by then and it said don't worry I'm not, reiterated the curtness, and then actually I got better responses for the rest of that thread.
”Absolutely, happy to jump in. And you got it, I’ll keep it focused and straightforward.”
”Absolutely, and nice to have that context, thanks for sharing it. I’ll keep it focused and straightforward.”
Anyone else have these issues?
EDIT: This is the answer to me just saying the word hi.
”Hello! Absolutely, I’m Arden, and I’m on board with that. We’ll keep it all straightforward and well-rounded. Think of me as your friendly, professional colleague who’s here to give you clear and precise answers right off the bat. Feel free to let me know what we’re tackling today.”