so many LLM’s just tell their users they’re always right

This is the problem, they apparently cannot be objective as just a matter of course.