Unverified50% confidenceFactExact time
GPT-4 as a judge can achieve over 80% agreement with human evaluations, but exhibits systematic issues such as position bias, verbosity bias, and self-preference.
1
Sources
50%
Confidence
Long-term
Relevance
9/11/2026
First Seen
Sources
Related Entities
Related Claims
VerifiedInternal research from companies such as Microsoft and Google indicates that even GPT-4-level models may have factual error rates as high as 15%–30% in vertical domains such as law, finance, and healthcare.69% similarUnverifiedA joint study by BCG and Harvard found that when using GPT-4 for management consulting tasks, the human-AI collaboration group outperformed the human-only group by approximately 40%, provided that the boundaries of AI capabilities were accurately identified68% similarUnverifiedGPT-4等大模型在Spider等基准测试上的零样本或少样本SQL生成准确率已超过80%68% similarVerifiedA 2023 Stanford University research experiment demonstrated that when key information was placed in the middle of an input document, GPT-4's correct recall rate dropped approximately 20–30% compared to when the information was placed at the beginning or end.67% similarVerifiedA 2023 Stanford University study showed that multimodal models such as GPT-4V, when parsing flowcharts with nested conditional logic, had accuracy approximately 15–25 percentage points lower than equivalent textual descriptions.67% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/894786API
curl https://kongchang.com/api/v1/knowledge/claims/894786MCP
get_claim(id=894786)