I have watched enough technology cycles now to be calm about the current wave of anxiety over automation and AI replacing infrastructure work. Every few years a tool arrives that supposedly makes our skills obsolete. Configuration management was going to end sysadmins. The cloud was going to end anyone who racked servers. Now AI assistants are going to end everyone who writes a shell script. Each time, the specific skill did fade, and each time the people who understood the underlying principles moved up a layer and kept working.
The pattern is consistent enough to plan a career around. What gets automated away is the how. The exact syntax, the memorized flags, the manual procedure that a machine can now do faster and more reliably. Good riddance to most of it, honestly. I do not miss hand-editing config files across forty servers. What does not get automated away is the why. Why this architecture and not that one. Why this tradeoff is acceptable here and dangerous there. Why the automated suggestion is confidently wrong for this particular system.
AI is very good at producing plausible answers to well-worn questions, which makes it a genuinely useful assistant and a dangerous authority. It can write the deployment script in seconds. It cannot know that your database cannot tolerate that migration during business hours, or that this "standard" pattern will interact badly with a legacy system it has never heard of. The judgment about whether the plausible answer is the right answer, in your specific context, with your specific constraints, is the durable skill.
So the advice I give people early in their careers has not changed even as the tools have transformed. Do not build your identity around a tool or a syntax, because both will be gone in a decade. Build it around understanding how systems actually behave, where they break, and what the tradeoffs really cost. That understanding transfers to whatever the next tool is, and it is exactly the thing the next tool cannot supply on its own.
The future does not belong to the people who resist the new tools or to the people who trust them blindly. It belongs to the people who can use them and still know when they are wrong.
– Serguey Shinder
