在实践中学习
5 分•作者: csacademy•3 个月前
只要我们还没有通用人工智能（AGI）或超级智能，我们就仍然需要优秀的软件工程师。即使我们达到了AGI，编码问题也得到了“解决”，我们真的应该就那样信任生成的代码吗？
将我们的认知思维和理解外包，既不是我们人类进步的方式，也不是我们保持在AI之上的方式。如果我们放弃这些，我们该如何限制AI产生的垃圾信息？如果到了那个地步，我们又该如何拔掉插头？
即使大多数学生和工程师似乎都在让他们的编码和算法推理能力退化，理解计算机科学概念和发展良好的软件工程实践仍然非常受欢迎。
是时候以一种有趣且引人入胜的方式重新培养这些能力了。如果AI在这方面有什么好处，那就是它能帮助我们学习，而不是替我们思考。
这就是为什么我们构建了一个Claude技能，通过让Claude指导你从头开始实现核心组件来教授计算机科学概念。
它是开源的，并且还处于早期阶段，因此非常欢迎社区贡献，目标是推广计算机科学和算法思维。
主技能仓库：https://github.com/csacademy-ma/learn-by-building
技能生成的示例练习：https://github.com/csacademy-ma/learn-by-building-samples
查看原文
As long as we don't have AGI or superintelligence, we still need good software engineers. And even if we do reach AGI and coding is "solved," are we really supposed to just trust the generated code?<p>Outsourcing our cognitive thinking and understanding isn't the way to advance as humans, nor is it the way to stay ahead of AI. If we give that up, how do we limit AI slop? And how would we even pull the plug, if it came to that?<p>Understanding computer science concepts and developing good software engineering practices is still very much in demand, even as most students and engineers seem to be letting their coding and algorithmic reasoning skills atrophy.<p>Time to build those back up, in a fun and engaging way. If AI is good for one thing here, it's helping us learn, not thinking for us.<p>That's why we built a Claude skill that teaches computer science concepts by having Claude guide you through implementing their core components from scratch.
It's open source and still early, so it's very much open to community contributions, with the goal of promoting computer science and algorithmic thinking.<p>Main skill repo: https://github.com/csacademy-ma/learn-by-building<p>Example exercises generated by the skill: https://github.com/csacademy-ma/learn-by-building-samples