📄 中文摘要
在职业生涯的第九年，作者开始反思自己所构建、记录和教授的一切是否只是为自己的离开准备的详细说明书。作者指出，工程师们的工作成果实际上成为了未来替代他们的模型的训练数据。每一次代码审查、每一条Slack讨论、每一份Confluence文档都是这些模型学习的素材。GitHub Copilot的智能并非来自教科书，而是源于工程师们的代码、注释和变量命名。此外，自动化工具并非中立，它们的设计和使用反映了开发者的思维方式和决策过程，这可能会影响未来的工作模式和职业安全。
我们是在培养自己的替代品吗？一位工程师的诚实看法
出处: Are We Training Our Own Replacements? An Honest Engineer's Take
发布: 2026年3月19日
📄 English Summary
Are We Training Our Own Replacements? An Honest Engineer's Take
Nine years into a career, the author reflects on whether everything built, documented, and taught serves as a detailed manual for their own exit. The work of engineers becomes training data for models that may replace them. Every PR review, Slack discussion, and Confluence document contributes to the learning material for these models. GitHub Copilot, for instance, became intelligent not by reading textbooks but by analyzing engineers' code, comments, and variable names. Furthermore, automation tools are not neutral; their design and usage reflect the thought processes and decisions of developers, potentially impacting future work patterns and job security.