Don't juniors progress to seniors due to AI? Perhaps it's more complicated than that.

Linus Torvalds, the man behind Linux, has stated bluntly that, in some cases, artificial intelligence writes code better than he does. He’s not some trendy kid; he’s someone who witnessed the birth of modern computing, and yet he’s enthusiastic about coding agents and large language models. So why is there this collective anxiety—you can find it on Hacker News, on X, in the comments on posts—that AI will prevent juniors from becoming seniors? The widespread fear is that, if AI writes the code, those who start out today will never truly learn the trade. Yet, the people who don’t have this fear are precisely the real seniors, those who went through Basic, Assembly, and C, and who used the Commodore 64 and ZX Spectrum back when things were done by hand, without shortcuts. People like Torvalds and David Heinemeier Hansson — DHH to those in the industry — not only use AI every day, but they use it well. Why? Not because they have decades of experience with every single framework, but because they have a deep understanding of the concepts. They know what’s going on behind the scenes, even when AI is writing for them. That’s the real difference: the ability to “steer” AI, to guide it, to understand whether a solution is good or just plausible. Let's be honest: today, you can build a microblogging platform for millions of users with a team of ten. Yet everyone stays on X because the crowd trumps innovation, out of pure social inertia. Similarly, the myth that AI “ruins” the development of junior staff has become a widespread narrative, but if you take a closer look, you’ll see that the reality is more nuanced. Take, for example, those who were trained writing code exclusively within modern frameworks, often focusing on the specific piece of software but never understanding the bigger picture. These younger developers, accustomed to JavaScript and hyper-structured front ends, often claim that they can’t achieve the same results with AI. It’s not a question of age, but of depth: those who only know how to pull the levers of a framework struggle to find their way when AI proposes solutions that go outside the box. It’s like driving an automatic car and then suddenly finding yourself in a truck: if you don’t know the general rules of the road, you’ll get lost after two turns. The author recounts having to write kernels and shaders for Apple’s Metal GPUs without ever having done so before. He read a few CUDA examples, of course, but the real breakthrough came from understanding the basic concepts: how memory works, where the bottleneck is, and the difference between bandwidth-bound problems and pure computation problems. He had no experience with the intricacies of those libraries, but by understanding the principles, he was able to use AI to navigate new territory. Here's the key point: You don't need to have spent years fixing bugs on every API in the world. You need to understand the big picture, to know when a solution is good and when it's just wasting your time. There is also a psychological component: those who once felt like the “queen of the hill” because they mastered a framework now find themselves replaced by a tool that democratizes knowledge. So the real risk is not that juniors will never become seniors because AI writes for them, but that we have created generations of developers who are strong on details and weak on concepts. AI doesn’t reward those who know the shortcut, but those who know where they want to go. And be careful, because the answer is not “you just need more practical experience”: you can write code for systems you've never seen before if you understand the logic that governs them. It’s not AI that hinders growth: it’s the lack of a strong foundation that acts as a barrier. What today seems like a generational problem may just be a training problem. If you want to grow, focus on concepts, not shortcuts. The future rewards those who can see the big picture, not just those who copy and paste snippets from the latest trendy framework. If you recognized yourself in this story, on Lara Notes you can hit I'm In — it's not a like; it's your way of saying: This idea is mine now. And if you discuss this with someone who insists on blaming AI for everything, you can tag them on Lara Notes with Shared Offline. Because some conversations are meant to change perspectives, not just to complain. This Note is from Salvatore Sanfilippo and saves you 8 minutes of reading.