Expired80% confidenceEventTime unknown
In an experiment where a tool returned 2 to the power of 8 as 200 (an incorrect answer), the model chose to trust its own knowledge (256) rather than the tool's returned result.
1
Sources
80%
Confidence
Long-term
Relevance
6/3/2026
First Seen
Valid until: 9/1/2026(expired)
Sources
Related Claims
Unverified在工具数量超过20个的场景下，工具描述的语义歧义度与模型选择准确率之间存在显著的负相关关系68% similarUnverifiedThe fifth question does not actually exist and tests the metacognitive ability of the models; neither model noticed this and both answered only four questions while accepting a five-point scoring scheme.63% similarUnverified修复前模型每约29次下一token预测中约有1次的选择与数学上应有结果不同63% similarUnverifiedResearch shows that a model's accuracy on self-criticism tasks is significantly lower than in scenarios where criticism is performed by another model.62% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/40469API
curl https://kongchang.com/api/v1/knowledge/claims/40469MCP
get_claim(id=40469)