4h agoSystems engineer Yacine argues users should avoid reading conversational LLM outputs unless they are immediately verifiableHe warns that persuasive anthropomorphic phrasing risks misinformative influenceSentimentSentimentPos21.4%Neg78.6%Positive users enthusiastically agree non-verifiable LLM outputs are mind poison, while negative users highlight risks of manipulation, unhealthy AI attachments, and eroded human competence from overreliance.14 comments with sentiment. View comments.