The article is quite provocative, making it easier to see the areas where generative AI can be beneficial.

Regarding boilerplate code, it does depend somewhat on the programming language, but in my opinion, it’s precisely the responsibility of senior developers to minimize boilerplate as much as possible. This aligns well with the principle of efficiency, often mistaken for laziness.

Consider this perspective:

Even if senior developers say that 80% of their code is AI-generated, what matters most is the 20% they write themselves.

The issue here likely isn’t laziness but rather the need to write code that junior developers can still understand. AI-generated code is often good enough, provided it’s properly supervised, to fit into this niche.

This does not support the hypothesis of "brain-dead" senior developers at all.

A programmer should be smart enough to design frameworks that enable less sophisticated tools (like AI) to fill in the gaps. This concept isn’t new; it requires effort to think through not only the implementation but also how to present these frameworks to other programmers in a way that minimizes their cognitive load.

I see nothing wrong with senior developers designing frameworks that allow AI to handle routine tasks. In some cases, those slots might not even involve traditional algorithms, as is the case with declarative, constraint, or logic programming.