"Doesn't exist, or doesn't give a fuck about suffering" is the answer that matches the data
I agree with you. (Though I might rephrase the second as 'doesn't care about suffering the way we do'. Either way, your point is valid.)
My point wasn't to say 'doesn't exist' is wrong, but that there is more than one possibility. If you or anyone has taken the time to evaluate the possibilities and come to the conclusion that 'doesn't exist' is the more likely / simple / predictive model, then I commend you. That is what rationality is about.
All I ask is the same courtesy as I might be exploring a different set of models than you are.
Could part of the issue with hallucination in LLMs be that they are often trained on Internet conversations where no one is likely to chime in with "I don't know"?
I agree with you. (Though I might rephrase the second as 'doesn't care about suffering the way we do'. Either way, your point is valid.)
My point wasn't to say 'doesn't exist' is wrong, but that there is more than one possibility. If you or anyone has taken the time to evaluate the possibilities and come to the conclusion that 'doesn't exist' is the more likely / simple / predictive model, then I commend you. That is what rationality is about.
All I ask is the same courtesy as I might be exploring a different set of models than you are.