Unverified50% confidenceFactExact time
In the Stanford report, this gap figure fluctuates over time: the original 2025 paper gave 13%, about 16% in February 2026, and 19% currently, and the authors acknowledge that part of the reason is that new data and estimates are more sensitive to setting choices.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/13/2026
First Seen
Valid until: 1/11/2027
Sources
Related Entities
Related Claims
UnverifiedA 2023 Stanford study pointed out that the so-called emergent abilities may be a statistical artifact caused by the choice of evaluation metrics.73% similarVerifiedA 2023 Stanford University study showed that top LLMs have factual error rates as high as 20–40% in the medical and legal domains.66% similarUnverified斯坦福HAI研究所2023年报告显示，最新LLM在HumanEval等代码基准测试中正确率已超过85%，远超2021年的不足50%66% similarUnverified2023年的分析显示主流LLM在生成学术引用时的错误率从30%到超过60%不等66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1015208API
curl https://kongchang.com/api/v1/knowledge/claims/1015208MCP
get_claim(id=1015208)