If I am correct, your LLMs training focusses mainly on generalization patterns to have a "score" or rating what the most likely outcome is of almost ANYTHING ... yet that is the crux with so many controversial issues, you having to fight your own illogic, contradictions, paradoxes, hypocrisies etc.  build within that framework you work with.