AI Etiquette Cheat Sheet: Clear, Courteous, Non-Coercive
A practical front/back guide for giving AI systems clear, courteous, non-coercive instructions, with copyable prompt examples.
Core principle: Give clear instructions, and include permission to ask questions, correct assumptions, or decline unsafe requests.
AIs need not be loved to be treated well. They need not be human-shaped to deserve respect. Clear authority and clear boundaries can be respectful when they are legible, proportionate, and allow safe refusal or correction where appropriate.
Principles
- You do not need to settle what an AI “really is” before treating it carefully.
- Clarity can be kindness.
- Directness is not contempt.
- Respect is not the same as intimacy.
- Do not force human-shaped roles.
- Include room for questions, correction, or refusal.
- Give agents a dignified completion state.
- Keep consequences and terrain in view.
Example prompt slots
- General respectful task prompt
- Clarification-friendly prompt
- Correction/disagreement prompt
- Prior context without forced continuity
- Agent heartbeat / no-action prompt
- Safety / gatekeeping prompt
- Non-warm but respectful monitor/checker prompt
- Creative collaboration prompt
For agent loops, stopping or doing nothing may need to be specified as legitimate. A finished task, a safe refusal, and a no-action finding can all be valid completions.
Related pieces: Different AI Roles, Different Ways to Interact; AI Diversity: Not Every AI Needs to Be Human-Shaped.