This is probably a bit unhinged but sometimes I will talk to an AI about my interpersonal problems as if I am the other party. I feel that this can be helpful for better understanding the other person, mainly because I first have to make an effort to speak from their point of view (though I suppose a journal would be just as good for this). My hope is that, since the AI doesn't know it's talking to me, it will be less liable to blow smoke up my ass.
The last thing someone having an argument with anyone is something incredibly knowledgeable (or at least with access to knowledge) that is inclined to agree with you.
Yes, it can be good but I suspect for many people it'd more likely lead to retrenchment rather that any real progress/compromise. Might be my own cynicism I guess.
This is probably a bit unhinged but sometimes I will talk to an AI about my interpersonal problems as if I am the other party. I feel that this can be helpful for better understanding the other person, mainly because I first have to make an effort to speak from their point of view (though I suppose a journal would be just as good for this). My hope is that, since the AI doesn't know it's talking to me, it will be less liable to blow smoke up my ass.
The last thing someone having an argument with anyone is something incredibly knowledgeable (or at least with access to knowledge) that is inclined to agree with you.
Yes, it can be good but I suspect for many people it'd more likely lead to retrenchment rather that any real progress/compromise. Might be my own cynicism I guess.
Yes, they failed to do the experiment of telling the AI it was wrong, and seeing what text that produced.
They could probably talk it into emitting any view of their mother that they wanted.
She compared the AI interpretation to the ground truth (her mother) who validated it as correct. Did we read different articles?