07/07/2026
So AI is too much of a hype-man? I get that.
Actually, kudos to you for noticing. An LLM's default is to agree with you but there are ways around that if you know how.
The reason it hypes you is simple. You ask "is this good?" and it goes hunting for every reason to tell you 'yes'. The question was framed so agreement was the only answer. Change the frame, you change what comes back.
Stop asking it to grade your idea and instead ask it to break it.
Try this- Hand it a role that has a reason to disagree with you: "Read my pitch as a skeptic and tell me what you'd say." Now it's not your hype-man, it's the hardest person in the room, and that's who you actually needed.
My favorite one is: "Before you agree with me, give me the strongest case against this with specifics." This makes it argue the other side first, so you know it's not just giving yes-man energy.
I keep saying this but AI is a tool. It'll respond to the inputs it's given, we just have to think differently about the way we're using it.
Trust me, you're not alone. What's the last thing AI cosigned that you knew good and well it shouldn't have? 😏