ARTICLE · 1350946
软件工程师当下该做什么
软件工程师当下该做什么我听不少人说，软件工程师要持续学习代码、技术栈，才能不被淘汰。我并不认同这个观点。
I’ve heard some people say that you should keep learning code and tech stacks to stay relevant as a SWE. I disagree.
每一次新模型发布，都在持续削弱这类技能的价值。我越来越觉得，自己更像一名管理AI的产品人，对深度技术知识的需求在不断降低。我曾经以为的技术护城河，如今都已经干涸。比起写一份技术方案，现在用用户故事去引导大模型，反而更加高效。我的工作所需技能，正在越来越去技术化，转向人的能力。
Each new model release is making those skills less and less valuable. Every day I feel more like a product person who manages AI, and the need for deep technical knowledge decreases. The moats I thought existed have all dried up, it’s now better to prompt a model with a user story than a technical plan. The skills I need for my job are becoming less and less technical, and more human.
有时候我会怅然：AI正在解决我过去深耕的问题，比如搭建SaaS软件。亲手打磨一款应用、看着它落地成型，曾带给我巨大的成就感。看着自己习得的技术近乎贬值，难免让人泄气。过去开发SaaS应用是一件难事，这也是我热爱它的原因。但现在它不再是难题了，所以我要去往还有高难度问题的领域，我要把格局放得更大。
Sometimes I get bummed out that AI is solving the problems I used to, like creating SaaS software. I get a lot of joy out of working hard on an app and seeing it come to life. Seeing the skills I learned become almost worthless is demotivating. Building a SaaS app used to be a hard problem, that’s why I loved it. It’s not a hard problem anymore, so I need to move to where there’s hard problems again. I need to think bigger.
我很喜欢《无穷的开始》这本书，其中一个观点让我深受启发：人类创新永无止境，“只要不违背自然法则，在掌握对应的知识后，一切皆可实现”。当我们创造出新事物，又会催生上千种新的可能性。计算机的诞生淘汰了很多岗位，但最终，它创造出了更多新工作。
One of my favorite books is the Beginning of Infinity, one of my takeaways is that there is no end to human innovation: “everything that is not forbidden by the laws of nature is achievable, given the right knowledge.” When we create a new thing, 1000 other new things become possible. The invention of the computer made a lot of jobs obsolete, but it eventually created so many more.
AI是一场伟大的创新，让我们能做到更多事，这也要求我们更新思考方式。旧的认知里，一款SaaS产品已经算体量很大的成果，但现在不是了。如今，复杂企业级软件的开发难度，甚至低于五年前做一个简单网站。现在的“大”，是站在更高维度思考：国家级、全球化、乃至更宏大的尺度。如果给你一支100人的工程师团队，你能打造什么？1000人呢？如今我们用极低的成本，就能拥有同等的生产力，我们的思考不能局限在应用层面。
AI is an incredible innovation that allows us to do more, which requires new thinking. My old ways of thinking tell me a SaaS app is a big thing, but it’s not. Complex enterprise level software is now easier to make than a simple website was 5 years ago. Big now is thinking on a much larger scale, national, global, cosmic. If you had a team of 100 software engineers, what could you build? What about 1000? We now have that for pennies on the dollar, we have to think bigger than apps.
作为工程师，我们也要拓宽自身能力边界。不能只局限在技术能力，要理解完整的商业链路：产品、营销、销售。
We also need to think bigger about our range as engineers. We need to expand beyond technical skills, to understand the whole business stack: product, marketing, and sales.
我当下保持竞争力的思路，是去学习AI无法代劳的能力：产品定位、创造力、用户体验、心理学、营销等等。这些属于人的课题，AI无法独立解决，因为它们没有标准答案，也没有可供训练的测试集。AI擅长模式匹配，但它永远无法理解生而为人的感受。这正是我们的价值所在，而我正全力深耕这个方向。
My approach to staying relevant now is to learn things AI can’t help with: product positioning, creativity, delighting humans, psychology, marketing, etc. These are human problems AI can never solve on its own, because there are no simple answers or tests to train on. AI is great at pattern matching, but it will never understand what it’s like to be a human. That is where our value is, and I’m doubling down.
I’ve heard some people say that you should keep learning code and tech stacks to stay relevant as a SWE. I disagree.
每一次新模型发布，都在持续削弱这类技能的价值。我越来越觉得，自己更像一名管理AI的产品人，对深度技术知识的需求在不断降低。我曾经以为的技术护城河，如今都已经干涸。比起写一份技术方案，现在用用户故事去引导大模型，反而更加高效。我的工作所需技能，正在越来越去技术化，转向人的能力。
Each new model release is making those skills less and less valuable. Every day I feel more like a product person who manages AI, and the need for deep technical knowledge decreases. The moats I thought existed have all dried up, it’s now better to prompt a model with a user story than a technical plan. The skills I need for my job are becoming less and less technical, and more human.
有时候我会怅然：AI正在解决我过去深耕的问题，比如搭建SaaS软件。亲手打磨一款应用、看着它落地成型，曾带给我巨大的成就感。看着自己习得的技术近乎贬值，难免让人泄气。过去开发SaaS应用是一件难事，这也是我热爱它的原因。但现在它不再是难题了，所以我要去往还有高难度问题的领域，我要把格局放得更大。
Sometimes I get bummed out that AI is solving the problems I used to, like creating SaaS software. I get a lot of joy out of working hard on an app and seeing it come to life. Seeing the skills I learned become almost worthless is demotivating. Building a SaaS app used to be a hard problem, that’s why I loved it. It’s not a hard problem anymore, so I need to move to where there’s hard problems again. I need to think bigger.
我很喜欢《无穷的开始》这本书，其中一个观点让我深受启发：人类创新永无止境，“只要不违背自然法则，在掌握对应的知识后，一切皆可实现”。当我们创造出新事物，又会催生上千种新的可能性。计算机的诞生淘汰了很多岗位，但最终，它创造出了更多新工作。
One of my favorite books is the Beginning of Infinity, one of my takeaways is that there is no end to human innovation: “everything that is not forbidden by the laws of nature is achievable, given the right knowledge.” When we create a new thing, 1000 other new things become possible. The invention of the computer made a lot of jobs obsolete, but it eventually created so many more.
AI是一场伟大的创新，让我们能做到更多事，这也要求我们更新思考方式。旧的认知里，一款SaaS产品已经算体量很大的成果，但现在不是了。如今，复杂企业级软件的开发难度，甚至低于五年前做一个简单网站。现在的“大”，是站在更高维度思考：国家级、全球化、乃至更宏大的尺度。如果给你一支100人的工程师团队，你能打造什么？1000人呢？如今我们用极低的成本，就能拥有同等的生产力，我们的思考不能局限在应用层面。
AI is an incredible innovation that allows us to do more, which requires new thinking. My old ways of thinking tell me a SaaS app is a big thing, but it’s not. Complex enterprise level software is now easier to make than a simple website was 5 years ago. Big now is thinking on a much larger scale, national, global, cosmic. If you had a team of 100 software engineers, what could you build? What about 1000? We now have that for pennies on the dollar, we have to think bigger than apps.
作为工程师，我们也要拓宽自身能力边界。不能只局限在技术能力，要理解完整的商业链路：产品、营销、销售。
We also need to think bigger about our range as engineers. We need to expand beyond technical skills, to understand the whole business stack: product, marketing, and sales.
我当下保持竞争力的思路，是去学习AI无法代劳的能力：产品定位、创造力、用户体验、心理学、营销等等。这些属于人的课题，AI无法独立解决，因为它们没有标准答案，也没有可供训练的测试集。AI擅长模式匹配，但它永远无法理解生而为人的感受。这正是我们的价值所在，而我正全力深耕这个方向。
My approach to staying relevant now is to learn things AI can’t help with: product positioning, creativity, delighting humans, psychology, marketing, etc. These are human problems AI can never solve on its own, because there are no simple answers or tests to train on. AI is great at pattern matching, but it will never understand what it’s like to be a human. That is where our value is, and I’m doubling down.
