📄 中文摘要
研究通过对五个前沿模型进行1,080次严格评估，探讨了提示的结构对AI输出质量的影响。评估涵盖了12个不同的任务领域，包括编程、数学、数据提取和创意写作等。每次评估均由三位评审在100分制下进行盲评。研究结果显著改变了AI应用的构建方式，强调了提示格式的重要性。
AI 格式战争：你的提示结构重要吗？（1,080 次评估后）
📄 English Summary
AI Format Wars: Does the Shape of Your Prompt Matter? (1,080 Evals Later)
The study evaluates the impact of prompt structure on AI output quality by conducting 1,080 rigorous assessments across five frontier models in 12 distinct task domains, including coding, math, data extraction, and creative writing. Each evaluation was blindly scored by a three-judge panel on a 100-point scale. The findings significantly altered the approach to building AI applications, highlighting the importance of prompt format.