Unverified50% confidenceSolutionExact time
Declaring "I don't know" via prompts alone is insufficient to solve the knowledge leakage problem; external controls or post-processing are needed for detection and correction.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/30/2026
First Seen
Valid until: 1/28/2027
Sources
Related Entities
Related Claims
Unverified承认不知道并展示如何推理，远胜于强行编造标准答案，面试官更看重结构化解决问题能力而非记忆力78% similarUnverifiedIf a system scores primarily on whether an answer is given rather than whether an answer should have been given, then responding 'I don't know' is counted as a failure.76% similarUnverifiedIn prompt engineering, boundary constraints, requiring source citations, and specifying "say you don't know if you don't know" can reduce the probability of hallucination, but cannot be eradicated at the mechanism level.76% similarUnverifiedFind silent failures by evaluation rather than by dashboards: inspecting outputs is necessary but far from sufficient; what really catches the problem is asserting on the shape of the run itself, and that can only be done where you already know beforehand what the run should look like.76% similarUnverified只有当问题足够明确可清楚表述时才值得开一张工单，若问题本身尚未想清楚则应留在迷雾中暂不处理75% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1115567API
curl https://kongchang.com/api/v1/knowledge/claims/1115567MCP
get_claim(id=1115567)