Yeah, avoidance of "harmful" or "offensive" responses is an awful metric for the quality of an LLM. I understand big corporate feel they need to cover their ass

Yeah, avoidance of "harmful" or "offensive" responses is an awful metric for the quality of an LLM. I understand big corporate feel they need to cover their asses, but I much prefer software that operates on the principle that we're all consenting adults.