AI trust is often driven less by evidence than by emotion, context, and cognitive bias, which is why conversational systems can be persuasive even when their limits are unclear.