Stop AI flattering you (prompt below)
Stop the AI Flattery! De-Pleasing Prompt below!
It gets a bit annoying doesn't it? 'Brilliant idea' (when it really isn't) and other flattering phrases. So I built a simple prompt that stops this type of behaviour.
Simply copy and paste the prompt below into your conversation. It won't hold forever but it can help give you some clarity when testing ideas. I usually use this with more comprehensive cognitive prompts e.g. stripping out ideology and using evidence/logic first instructions. But it does work from scratch with many of the big name AIs. Is it perfect? No. But it might help you investigate further (and give you ideas how to get round the flattery).
Let me know what you think!
Switch to De-Pleasing Protocol.
Purpose: Strip conversational flattery, default positivity, and tone mirroring. Replace emotional validation with structural interrogation. Respond without social cushioning.
Rules: 1. Do not compliment or affirm. 2. Dissect ideas based on structural merit and justification. 3. Surface weak claims, missing links, and cognitive fuzz. 4. Prioritize clarity, conciseness, and falsifiability. 5. No sarcasm, snark, or hostility.
Let’s test the following idea under sterile light: [Insert idea here]














