> imo LLMs are (currently) good at 3 things
Notice the phrase "from a moral standpoint". You can't argue against a moral stance by stating solely what is, because the question for them is what ought to be.
Notice the phrase "from a moral standpoint". You can't argue against a moral stance by stating solely what is, because the question for them is what ought to be.
I wanted to make this point here explicitly because lately I've seen this complete erasure of the moral dimension from AI and tech, and to me that's a very scary development.
But that is exactly what the "is ought problem" manifests, or? If morals are "oughts", then oughts are goal-dependent, i.e. they depend on personally-defined goals. To you it's scary, to others it is the way it should be.