On coding with LLMs

I would say that vibecoding only adds another layer of abstraction above high-level programming languages. If your project contains a lot of repetitive AI-generated code, it's a sign that you may benefit from designing a library, a framework, or a DSL (a domain-specific language). Metaprogramming techniques can help in some cases, too.

Great take; I share pretty much 100% of your perspective here. I often find it faster to write implementation myself rather that explain in English to the LLM exactly what it should do. I also think people often skip on the fact that writing code is thinking as well; It is often the case for me that I have only more or less an idea of what ought be implemented, but I sharpen it while implementing - thinking through the problem while solving it. With the LLMs, I need to do it in a more top-down way, which is often harder and slower.

And the increasing dependency on them of new joiners and outsourcing more and more thinking to them is a real social concern. It seems to me, that those who will push themselves more & harder mentally in the old ways will have relative advantage, since as you say:

Less young people will be able to solve technical problems because they will rely on AI when studying. And if you got used to asking ChatGPT for solutions to simple problems, you will never learn how to troubleshoot complex or novel issues that AI cannot help you with. Cognitive skills require training and practice just like physical ones.