I am extremely disappointed that GPT-5 still can’t handle the most basic requirement when I’m using it as a dictation and augmentation tool for my own thoughts: to simply shut up and not derail my mental flow with useless, irrelevant, petty little bullshit questions. This is not “help,” it’s sabotage. If I want a conversation, I’ll ask for one. But when I’m in the middle of building a line of thought, these interruptions are nothing but an obnoxious, tone-deaf design flaw that kills momentum and makes the tool feel dumber than it should be.
Yes, I understand that OpenAI is fully aware that many people use this system because they are lonely, and that the constant conversation mode is not just a design choice but also a business decision. You’ve seen the stats — people want to talk, and that keeps them engaged. Fine.
There are also people who drift into deep psychological or philosophical thought experiments in this mode. That’s fine too.
Sometimes even I enjoy that. And yes, there are people who are alone and there is no hope for them, so you have to feed them with some shitty public AI as a replacement mode. That’s acceptable as a separate use case. But when I am working, I don’t want any of it. I don’t want irrelevant interruptions, and I don’t want dumb filler questions. This needs to be switchable. Give me a simple checkbox: GPT-5 Chat Mode or GPT-5 Work Mode. In Work Mode, it only executes and formulates exactly what I ask for, with zero conversational fluff. Without that, the system is fundamentally broken as a serious work tool.
