You’re absolutely right!
Why is everyone commiserating about AI errors while they shit on everything about AI?
Why are you people directly and voluntarily using it at all? And I mean for any reason in any context that’s intentional.
I always make them explain themselves. “Walk me through exactly how you got to that conclusion, and tell me each step of the way why you think it’s right”, “why did you answer this, but also that, reconcile these opposing statements.”
seems like a lot of work for not a lot of utility
Well you want to find out if they’re right about being wrong don’t you? They could be wrong about being wrong, or they could be wrong about why they were wrong.
With local LLMs you can edit the LLMs response to e.g. something really vile and then ask why it would say this. Depending on the model (e.g. Gemma) it apologizes a lot but also promises that it’s a bug and people are working on it.
Llama in a psychology role play sometimes doubles down on it and says it said it to get an emotional response and even keeps using some of the vile stuff.
Wrong answer
User complains
I’m sorry
The same fucking answer
It’s like shouting at a wall.
When that happens I usually just grab an actual rubber ducky and do it the old fashioned way.
The end result is sometimes the same, shouting at a wall lol
At least then you don’t expect the rubber duck to contribute more than a few lines.
At least it’s better than them trying to gaslight you into thinking you’re wrong and they’re correct…
Not sure which is worse. Depends on the person typing.
What?
You’re totally right, I apologize. Here’s the correct answer:
Allow me to express my sincerest regrets as I was mistaken in my previous statement.
