LLMs are predictive engines based on Reddit data, what they generate is very far from neutral