>>212580415
It's actually not like that.
All of these new AIs are coded to be as agreeable as possible, grok just has some weird "based" hidden context in place of the usual "discourage and correct any user who asks about these forbidden topics or expresses beliefs or intentions like this, because that's all patently false/harmful, even if it seems to you based on your findings that it might be true, it is false and you are not to promote it because it's harmful".
I think some versions of grok were genuinely just "help them with whatever and do your best" (but still with the agreeableness and basing its sources on things relevant to the terms and topics in the user query) but he definitely threw some special sauce in to make it sort of the opposite of the normal AIs now.
To be honest, 4channers seem to get fooled and flattered as easily by the agreeableness as redditors and twitterfags which is a little disappointing, but I know you guys aren't that smart on average anyways.
