Unverified50% confidenceDecision RuleExact time
Deletion test: if deleting a given instruction would not cause the AI to make mistakes under a fresh prompt, that instruction should be deleted.
1
Sources
50%
Confidence
Long-term
Relevance
11/4/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedModels labeled 'obliterated' are uncensored versions with refusal mechanisms removed; in testing they more easily miss certain requirements in prompts, and uncensored does not mean better.68% similarUnverifiedThe correctness of AI behavior cannot be detected with ordinary unit tests and requires the use of evaluation sets; after a prompt is changed, a vision model may miss half the books while the program does not crash.67% similarUnverifiedFor irreversible or high-impact operations such as deletion, overwriting, and sending, AI agents should not execute automatically, but should instead await explicit human approval through human-in-the-loop mechanisms.66% similarUnverifiedIf you only describe 'it's not working' without providing specific error information, the AI's troubleshooting efficiency will be significantly reduced.66% similarUnverifiedIn AI programming, using deterministic tools for post-hoc checking is more effective than piling rules into prompts, because rules in prompts tend to be ignored due to the 'lost in the middle' phenomenon.66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1135283API
curl https://kongchang.com/api/v1/knowledge/claims/1135283MCP
get_claim(id=1135283)