📄 中文摘要
在开发一个韩国四柱命理应用的过程中，作者意识到，关键不在于让人工智能做什么，而在于不让它做什么。尽管应用的收入仍为零，但作者记录了调试过程。在尝试让大型语言模型（LLM）分析某个特定出生日期的命理时，虽然模型的回答看似流畅且专业，但基础计算却是错误的。这揭示了LLM在处理复杂的命理计算时的局限性，强调了人类在这一领域的重要性和必要性。
为什么不应该让人工智能进行算命 — 以及如何正确地进行算命
📄 English Summary
Why You Shouldn't Let AI Do Your Fortune Telling — And How to Do It Right
The experience of building a Korean four-pillar fortune-telling app highlighted that the focus should not be on what to ask AI to do, but rather on what not to let it do. Despite the app generating zero revenue, the author documented the debugging process. When attempting to have a large language model (LLM) analyze the fortune for a specific birth date, the response appeared polished and professional. However, the underlying calculations were incorrect, revealing the limitations of LLMs in handling complex fortune-telling computations and underscoring the importance of human involvement in this domain.