Hacker News Reader

Comment by wisemanwillhear | Hacker News Reader

With AI, I often like to act like a 3rd party who doesn't have skin in the game and ask the AI to give the strongest criticisms of both sides. Acting like I hold the opposite position as I truly hold can help sometimes as well. Pretending to change my mind is another trick. The idea is to keep the AI from guessing where I stand.

View on HN

> Acting like I hold the opposite position as I truly hold can help sometimes as well.

I find this helps a lot. So does taking a step back from my actual question. Like if there's a mysterious sound coming from my car and I think it might be the coolant pump, I just describe the sound, I don't mention the pump. If the AI then independently mentions the pump, there's a good chance I'm on the right track.

Being familiar with the scientific method, and techniques for blinding studies, helps a lot, because this is a lot like trying to not influence study participants.

A lot of getting good mileage out of LLMs is promoting them to behave like they are blind and can only base their outputs on what is in front of them. Maintain an emic stance.

I have tried it a lot aswell. A single mistake and it guesses the side and changes the tone.

I will generally ask for the "devil's advocate" view and then have it challenge my views and opinions and iterate through that.

It generally does a pretty good job as long as you understand the tooling and are making conscious efforts to go against the "yes man" default.

Sounds like rubber-ducking with extra steps, tbh.

Reply on news.ycombinator.com