how do you know whether a llm chatbot is giving good advice or being overly sympathetic to your framing of conflict and might be equally justifying to the other person if they told their side, such that both ppl feel more self righteous n stubborn?