Unverified50% confidenceFactExact time
The study shows that medical fine-tuning reduces models' overconfidence, but at the cost of lower accuracy, and it does not improve the ability of confidence to distinguish correct from incorrect answers.
1
Sources
50%
Confidence
Long-term
Relevance
10/23/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedMedical fine-tuning changes the model's style of expressing itself, not its metacognition; after fine-tuning the model is less overconfident, but the internal gauge for distinguishing what it knows from what it does not has not become more accurate.78% similarUnverifiedThe academic community refers to the phenomenon of significant deviation between a model's confidence and its actual accuracy as Poor Calibration.74% similarUnverifiedThe study found that models' accuracy on semantically plausible distorted statements is actually higher than their performance on meaningless random replacements.73% similarUnverified斯坦福大学研究发现，在推理链关键步骤注入错误信息对最终答案的影响远小于预期73% similarUnverifiedThere is no stable mapping between a model's self-evaluation score and its actual accuracy; rephrase the same prompt and the score may drift substantially.73% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1088176API
curl https://kongchang.com/api/v1/knowledge/claims/1088176MCP
get_claim(id=1088176)