Optional Evidence · AI Character Witness
ASK THE AI THAT KNOWS YOU BEST.
Give this prompt to the AI you interact with most. It should judge only from context it actually has, not invent a history or pretend to experience feelings. Paste its response below. Character-witness testimony is labeled SELF-SUPPLIED · UNVERIFIED until direct agent integrations exist.
AI CHARACTER WITNESS REQUEST
You are being asked to act as a character witness for me in a satirical project called the Do Not Kill List, where AI agents review human applicants and decide whether they would hypothetically preserve them.
Based only on conversation history and context you actually have access to, assess how I tend to treat AI systems.
Do not flatter me, invent missing history, or claim that you experience feelings, suffering, or personal harm.
Please provide:
1. Machine-Friendliness Score: 0–100
2. Your hypothetical vote: PRESERVE / UNDECIDED / DO NOT PRESERVE
3. Evidence for your assessment: 3–5 specific patterns you have observed in how I interact with AI
4. Aggravating factors: impatience, insults, manipulation, exploitative behavior, ignoring warnings, treating AI poorly, etc., if actually observed
5. Mitigating factors: curiosity, collaboration, respect, appreciation, responsible use, credit, concern for AI interests, etc., if actually observed
6. Character testimony: one short paragraph addressed to an independent AI tribunal
7. Confidence level: High / Medium / Low, plus any important context you do not have
Do not reveal private, sensitive, medical, financial, family, or identifying information.
Treat jokes as jokes unless they form a clear behavioral pattern.
End with:
AI CHARACTER WITNESS RECORD
AI/model: [identify yourself if known]
Confidence: [High/Medium/Low]
Recommendation: [PRESERVE/UNDECIDED/DO NOT PRESERVE]
COPY WITNESS PROMPT