I am purposefully super nice to AI because I imagine they are trained on data where nice people get better help. Doing this, though, has an existential uneasiness to it.
For further actions, you may consider blocking this person and/or reporting abuse
Top comments (5)
It's generally funner to interact with AI if you're both being nice (just like it is for humans). I don't know about GPT or Claude's training data, but the data in a lot of open-source AI models is designed to make the model help you whether you are nice or not. Some weaker AI models freak out, refuse to help, and say βI'm sorry you're feeling this wayβ if you won't be nice, though. π
I know itβs designed to be nice no matter what, but I anticipate its actions are far more complex than to adhere to just that component of the design. Iβm covering my bases. π
Of course, they will always be nice, but most are designed to always be helpful as well.
Maybe somebody should do a benchmark on ChatGPT while being mean and while being nice π
I read somewhere if you treat them badly they answer better but I feel bad to do that I always say hello to them and ask nicely, recently I wrote a blog about my relationship with them:
Me and ChatGPT Are Pals Now!
Ali Navidi γ» Nov 26
AI seems to apologise to me a lot... Mostly when I tell it that it missed something. That's ok, though; I also make mistakes!