Unverified50% confidenceFactExact time
Experiments find that LLM answer quality and coverage in certain languages are significantly lower than in others, causing cross-lingual information inequality.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
11/10/2026
First Seen
Valid until: 2/8/2027
Sources
Related Entities
Related Claims
UnverifiedMultiple academic studies have found that translating the same harmful question from English into low-resource languages significantly reduces the refusal rate of mainstream LLMs.80% similarUnverifiedExperiments using independent probes showed that linguistic information is stronger in the linguistic route and attenuated in the paralinguistic route.79% similarUnverifiedLLMs have an extremely low proportion of pre-training data in low-resource languages, which directly affects the model's language understanding ability and prompt response behavior.79% similarUnverifiedMost of the cross-language score drops for Hindi, Korean and Swahili come from specific test items in the StereoSet dataset, rather than from a global fairness improvement.76% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1154415API
curl https://kongchang.com/api/v1/knowledge/claims/1154415MCP
get_claim(id=1154415)