English

A few days ago, with the help of AI, I went through multiple rounds of discussion with Alibaba Cloud customer service around a concrete technical problem.

What makes this worth recording is not whether AI was right or the engineer was right. What matters is that the structure of technical problem-solving is clearly changing.

In this process, AI was genuinely useful:

• it helped me understand the structure of the problem faster,
• expanded the range of possible solutions,
• improved how I organized questions and compared options,
• and increased the efficiency of communication.

But in the end, the more reliable closure still came from a more experienced human engineer. The Alibaba engineer’s method and conclusion turned out to be correct.

This does not mean AI failed. It means we are living through a transitional pattern:

AI can already participate deeply in technical problem-solving, but in complex systems, real-world environments, and experience-based judgment, human engineers still hold stronger final authority.

That is a very typical scene of the 0.5 era.

Today, AI expands the solution space. In the future, it will increasingly move into device control, system scheduling, fault diagnosis, runtime optimization, and automated execution.

At that point, AI will no longer be only a tool for answering questions. It will become part of the driving layer behind machines and systems.

So this experience confirmed something important for me: AI is not yet the final judge in every technical case, but its rise as a major future driver of systems and machines already feels irreversible.

---

中文

前几天，我在人工智能的帮助下，和阿里云客服围绕一个具体技术问题进行了多轮讨论。

这件事真正值得记下来的，不是“AI 说得对，还是工程师说得对”，而是我已经明显感觉到：技术问题的求解结构正在变化。

在这次过程中，AI 并不是摆设。它确实起到了作用：

• 它帮助我更快理解问题的结构
• 帮我展开了多种可能路径
• 让我能更快整理问题、组织提问、比较方案
• 也提高了我和客服沟通时的效率

但走到最后，真正更可靠的收束仍然来自一位更有经验的人类工程师。最终证明，阿里工程师的方法和结论是正确的。

这并不意味着 AI 失败了。恰恰相反，它说明我们正处在一个非常典型的过渡阶段：

AI 已经可以深度参与技术问题求解，但在复杂系统、真实环境和经验判断面前，人类工程师仍然掌握着更强的最终裁决能力。

这正是我理解的“0.5 时代”现场。

在这个阶段里，AI 负责的是把“可想空间”迅速铺开：

• 帮人拆解问题
• 帮人枚举方案
• 帮人补充信息
• 帮人提升沟通与试错效率

而更成熟的人类工程师，负责的是把“可行空间”真正定下来：

• 哪条路在真实系统里能成立
• 哪些风险不是表面上能看出来的
• 哪些方法看起来合理，其实在生产环境里会出问题

所以这次经历真正让我确认的，不是“AI 还不行”，而是另一件更重要的事：

AI 的发展势不可挡，未来一定会成为主要设备、系统和机器的重要驱动者。

今天它更多像一个协作助手、分析助手、沟通助手； 但以后，它会越来越多地进入设备控制、系统调度、故障诊断、运行优化和自动执行层。

到那时，AI 就不再只是回答问题的工具，而会逐渐成为很多机器和系统背后的“驱动层”。

如果说今天的人机协作，仍然是“AI 参与，人类裁决”； 那么未来更可能会变成：

AI 成为主要驱动层，而人类负责更高层的校正、监督与价值判断。

这不是一个已经完成的时代， 但它已经非常清楚地开始了。