Unverified50% confidenceFactExact time
早期研究发现同一问题用不同表述方式提问，模型性能差异可高达数十个百分点
1
Sources
50%
Confidence
Long-term
Relevance
7/14/2026
First Seen
Sources
Prompt中一个词如何左右AI输出：可能性之井原理解析
Related Claims
Partially VerifiedResearch shows that the output quality of the same model can vary by several times under different prompts.83% similarUnverified学术研究已证明，同样的问题以不同方式表达，模型的回答准确率可能相差20%以上82% similarUnverified不同模型在基准测试上的性能差异往往远小于同一模型在不同任务类型、不同输入格式下的行为差异76% similarUnverified该研究发现同一模型不同部署时间点在代码生成和逻辑推理任务上通过率可能出现10%至20%的显著波动，且方向不总是单向改善75% similarUnverifiedResearch shows that the performance gap between different prompt designs can exceed 10 percentage points even for the same model.72% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/509945API
curl https://kongchang.com/api/v1/knowledge/claims/509945MCP
get_claim(id=509945)