Unverified50% confidenceBenchmarkExact time
Human validation experiments conducted by the research team showed a strong positive correlation between LLM-generated scores and human scores (Spearman's rho = .693, p < .001).
1
Sources
50%
Confidence
Long-term
Relevance
9/11/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedResearch shows that the agreement rate between LLM judges and human scores is approximately 80%, approaching but not yet surpassing the level of human inter-annotator agreement (approximately 81%).74% similarUnverified研究表明人类评估者在判断LLM输出的事实准确性时，正确率往往只有50%-70%72% similarUnverifiedMultiple studies from 2024 indicate that the agreement rate between LLM judges and human experts is typically between 70% and 85%.70% similarUnverifiedA meta-analysis by Schmidt & Hunter (1998) found that general cognitive ability tests have a predictive validity of r=0.51, while work sample tests reach r=0.5468% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/907807API
curl https://kongchang.com/api/v1/knowledge/claims/907807MCP
get_claim(id=907807)