Why ChatGPT Won't Teach You Linux (And What Will)

There's a provocative claim making rounds in tech education circles: if you're learning Linux primarily through ChatGPT, you're not going to make it. That's the position taken by Mischa van den Burg, a DevOps engineer with over a decade of Linux experience, during a recent livestream aimed at beginners.

The statement landed in chat like a grenade. But van den Burg wasn't trolling—he was articulating something many experienced engineers feel but rarely say out loud: AI tools might be making us worse at the fundamental skill of learning how to learn.

The Anti-AI Manifesto (Sort Of)

Van den Burg's argument isn't a blanket rejection of AI. It's more specific: "If you are always defaulting to AI and ChatGPT, I think you're not going to be a well-rounded engineer if all your only source of information is ChatGPT."

His alternative? A hierarchy of learning resources that feels almost quaint in 2024: man pages first, then the help command for shell built-ins, then—and this is where he gets evangelical—physical books. Specifically, the Unix and Linux System Administration Handbook, which he calls "the best Linux book on the planet."

"Any self-respecting engineer, any self-respecting Linux engineer needs to have this book," van den Burg told his audience of 86 viewers. "If you are taking a course from someone who claims to know Linux and he does not own this book, then that's an indicator that he might be full of shit."

It's a bold gatekeeping move, and it prompted immediate pushback. One viewer asked about "poor people who don't own that book." Van den Burg's response was characteristically blunt: "If you stop buying coffee for a week, you can afford this book."

The exchange reveals a tension at the heart of modern tech education: How much friction is necessary for learning? Van den Burg's position is that some struggle—the struggle of reading documentation, of piecing together information from primary sources—is the point, not an obstacle to be removed.

What Linux Actually Is (And Why It Matters)

The livestream itself was structured around answering fundamental questions, starting with "What is Linux?" and "What is a Linux distribution?" Van den Burg walked viewers through the distinction between the Linux kernel—the core operating system code originally written by Linus Torvalds in the '90s—and Linux distributions, which package that kernel with the commands and tools users actually interact with.

Using the Unix and Linux System Administration Handbook as his reference, van den Burg demonstrated his research process in real-time. "A Linux distribution comprises the Linux kernel, which is the core of the operating system, and packages that make up all the commands you can run on the system," he read from the text.

Then he showed what this means in practice, running commands like echo and ls and tracing them back to their origins in the GNU core utilities. It was pedagogy as archaeology—showing learners not just what things are, but how to figure out what things are.

"I am not a walking encyclopedia," van den Burg emphasized. "Neither do I have any ambition to be a know-it-all. Rather, I would much rather be someone who can find the information when needed also without AI and ChatGPT."

The DevOps Competency Question

One viewer asked a question that gets to the practical heart of this debate: "For DevOps career, which parts of the OS should one understand for an easier DevOps career?"

Van den Burg's answer was refreshingly bounded. "You don't have to become a Linux kernel maintainer in order to land a DevOps job," he said. "To be a DevOps engineer you need to have a functional competency in Linux."

That functional competency, he clarified, means understanding Linux at the distribution level, potentially diving into cgroups and namespaces if you're working with containers, but not necessarily compiling your own kernel. It's a pragmatic middle ground—deep enough to be effective, not so deep you're drowning in theory.

This raises an interesting question about the current state of DevOps education. Van den Burg criticized many bootcamps and competitors for treating "Linux as one of the check boxes that you should check off," when he believes "Linux is actually the most important topic to master in any tech career."

His free Linux course is eight hours long—a fraction of the 20+ hours on Linux in his paid DevOps program, which totals 75 hours of video material. The scale suggests he's serious about the foundation-building. Whether that's the right amount of time, or whether other paths might work equally well, is the kind of question the tech education market will eventually answer.

The Book Versus The Bot

The core tension here isn't really about ChatGPT versus books. It's about what kind of struggle produces competence. Van den Burg's model assumes that wrestling with documentation, cross-referencing sources, and building your own mental models creates deeper understanding than having answers served up instantly.

There's research suggesting he might be right. Cognitive science has long documented the "desirable difficulties" effect—the finding that learning that feels harder often sticks better. But there's also research showing that well-designed AI tutors can accelerate learning for some students in some contexts.

The real question might be: desirable difficulty for whom, and for what? Someone trying to quickly solve a specific problem at work might legitimately choose ChatGPT. Someone trying to build foundational expertise might need the long, slower path through man pages and textbooks.

Van den Burg's advice to "invest in yourself" by buying books and "quality learning materials" positions education as something you pay for—with money, with time, with effort. It's a model that rewards patience and punishes shortcuts. Whether that's inspiring or gatekeepy probably depends on which side of the gate you're standing on.

But his underlying point about research skills might transcend the specific tools. In a field that changes as rapidly as technology, knowing how to find and evaluate information might matter more than any particular fact you learn. ChatGPT can give you an answer. The Unix and Linux System Administration Handbook and the man pages can teach you how to find answers. The difference between those two things might be the difference between getting a job and building a career.

---

Marcus Chen-Ramirez is a senior technology correspondent for Buzzrag.