Unverified50% confidenceFactExact time
商用人脸识别系统在不同种族和性别群体间存在显著准确率差异，深肤色人群和女性的误识率明显偏高，根源在于训练数据构成不均衡
1
Sources
50%
Confidence
Long-term
Relevance
8/21/2026
First Seen
Sources
Related Claims
UnverifiedIf training corpora contain gender, racial, or cultural bias, models will inherit these biases.70% similarVerifiedThere is a gap between benchmark scores and real user experience, and some vendors engage in targeted optimization for specific benchmarks (score gaming) as well as training data contamination that inflates scores.67% similarUnverified来自不同厂商、经历不同训练路径的模型持有不同的先验偏见，从而在交叉审查中形成真正的视角差异64% similarUnverifiedHuman annotators exhibit a cognitive bias toward more information being better when evaluating answers, causing reward models to implicitly link length with quality, which is one of the causes of model verbosity.63% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/782858API
curl https://kongchang.com/api/v1/knowledge/claims/782858MCP
get_claim(id=782858)