Unverified50% confidenceOpinionExact time
The high accuracy of large models on everyday tasks provides fertile ground for automation bias—the higher a system's accuracy rate, the more easily its errors are overlooked by humans.
1
Sources
50%
Confidence
Long-term
Relevance
9/23/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedLarge models have higher overall accuracy, but their errors are harder to detect because their output is more fluent and persuasive.75% similarUnverifiedError dependence is stronger among high-accuracy frontier model reviewers, even when those reviewers come from different vendors.74% similarUnverified对于日常简单问题，调低推理强度可获得即时反馈；对于复杂分析任务，调高推理强度可获得更严谨可靠的输出73% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/945262API
curl https://kongchang.com/api/v1/knowledge/claims/945262MCP
get_claim(id=945262)