I don't think it's "better" or "worse" to "say" anything, LLMs don't have 'speech' or 'moral opinions' they're just text-generators. As I said above, it's not a question of the existence of moral truths, it's a question of getting the LLM to execute syllogisms consistently in a context that people are very touchy about (as you so helpfully demonstrate over and over again). If you're expecting LLMs to be universal truth generators I think you will be disappointed for a very long time. If you want LLMs to demonstrate perfect moral clarity likewise. Seems like getting mad at a hammer for hitting your thumb, personally I'd prefer to just hold my hand farther away from the nail and sidestep the problem altogether.