Confidence Scorer prompt: make models label every claim HIGH/MEDIUM/LOW
thisguyknowsai · x · 2026-10-22
Prompt 2, the "Confidence Scorer": after answering, the model labels every factual claim HIGH (would stake reputation on it), MEDIUM (verify recommended) or LOW (best guess). The user then only manually verifies LOW claims — forcing the model to audit itself before you read.
Related event: Anti-Hallucination Prompts: Confidence Scorer and Refusal Protocol(2 posts)→
More from coding & agent
- Codex is painfully slow, Claude over-refuses: one developer stuck between both — not_rian · 2026-10-22
- Red Team Pass prompt: make the model critique and fix its own output — thisguyknowsai · 2026-10-22
- Assumption Audit prompt: list assumptions first, then confirm before answering — thisguyknowsai · 2026-10-22
- Leaked NVIDIA internal prompting playbook claims 41% hallucination cut — thisguyknowsai · 2026-10-22
- I ran 40 PRs through Opus, Sonnet and GPT-6: only 11 of 37 bugs were caught by all three — ParticularFan1122 · 2026-10-22
- Dev argues voice or a mechanical keyboard are the only good ways to talk to agents — SimonBalmain · 2026-10-22