Unverified60% confidenceFactExact time
A 2019 study found systematic gender-association biases in word vector embeddings, for example, "nurse" was highly associated with women while "engineer" was highly associated with men.
2
Sources
60%
Confidence
Long-term
Relevance
8/7/2026
First Seen
Sources
谷歌AI搜索被曝性别偏见：相同问题不同性别答案截然不同
Related Claims
UnverifiedThe gendered 'expert pool' phenomenon is often called the 'Matthew effect of authoritative citation' in communication studies.70% similarUnverifiedThe research team systematically validated the context confusion phenomenon in three domains: gender equality, privacy, and personal safety.69% similarUnverifiedIf training corpora contain gender, racial, or cultural bias, models will inherit these biases.69% similarUnverifiedIf bias is encoded in a model's embedding space in the form of vector associations such as "female–home, male–workplace," it may still seep into actual outputs through pathways such as contextual inference, even if the output layer has been trained not to express it directly.67% similarUnverifiedThe gender bias in word embedding models originates from training corpora (such as the Google News corpus) reflecting pre-existing human social attitudes, rather than being artificially introduced.67% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/702886API
curl https://kongchang.com/api/v1/knowledge/claims/702886MCP
get_claim(id=702886)