Unverified50% confidenceCautionExact time
LLM-based AI personas may inherit biases from training data, including linguistic bias, socioeconomic bias, and temporal bias, leading to systematic misjudgment of minority groups or non-mainstream viewpoints.
1
Sources
50%
Confidence
Long-term
Relevance
9/12/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedIf training corpora contain gender, racial, or cultural bias, models will inherit these biases.77% similarUnverified训练数据构成决定模型世界观边界的现象在数字人文领域被称为'语料库偏差'（Corpus Bias）76% similarUnverifiedLLM evaluation is more objective than human evaluation, as it can score candidates consistently within a unified evaluation framework, avoiding the unfairness caused by cognitive biases such as fatigue, mood, and the primacy effect in human interviewers.73% similarUnverifiedexposure bias的根源在于训练与推理的分布不匹配，训练时使用teacher forcing策略而推理时模型依赖自身生成的token73% similarUnverifiedSingle judge models have known biases including position bias, length bias, and self-enhancement bias.72% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/914929API
curl https://kongchang.com/api/v1/knowledge/claims/914929MCP
get_claim(id=914929)