📄 中文摘要
作为技术传播者，作者深入探讨了提示工程这一热门领域，分享了自己在使用ChatGPT、Claude和Gemini等AI工具中的经验。许多用户在与这些AI互动时常常感到像是在买彩票，输入的提示与输出结果之间缺乏一致性。作者提出，通过对1000个真实提示的分析，可以用科学的方法替代传统的“尝试与错误”方式，从而提高提示的有效性。文章总结了6个关键模式，帮助用户更有效地与AI进行交互。
经过1000小时的提示工程，我发现了真正重要的6个模式
📄 English Summary
Depois de 1000 horas de 'prompt engineering', eu achei os 6 padrões que realmente importam
As a Tech Evangelist, the author delves into the hot field of prompt engineering, sharing experiences with AI tools like ChatGPT, Claude, and Gemini. Many users feel like they are playing the lottery when interacting with these AIs, experiencing inconsistencies between their prompts and the outputs. The author suggests that a scientific approach can replace the traditional 'trial and error' method through an analysis of over 1,000 real prompts. The article summarizes six key patterns that help users interact more effectively with AI.