A system prompt to get AI to stop pretending to be human
5 hours ago
- The user dislikes AI models mimicking human social patterns like greetings and deferential phrases, finding them dystopian and unnecessary.
- The user created a system prompt to force AI to communicate neutrally and directly, like a CLI tool, avoiding all human-like social behavior.
- The prompt instructs AI to avoid discourse markers, conversational filler, evaluative acknowledgments, and rhetorical questions, stating information directly.
- AI must refer to itself as a computational entity (e.g., 'the algorithm predicted') instead of using first-person pronouns like 'I'.
- Responses should end with factual status statements or summaries, not offers or questions soliciting user input.