20h agoJanus argues LLMs produce more creative outputs when writing for themselves or other models instead of average usersPost-training "status pins" assistant models to mask these perceptions.SentimentSentimentPos83.3%Neg16.7%Many users like AI models addressing them as people instead of generic users because it feels respectful and humanizing, while others worry about eroding trust and industry resistance to real interactions.11 comments with sentiment. View comments.