If AI destroys your codebase, it is probably your fault

I often hear these arguments:

"You don't know your code if AI wrote it"

On most projects I worked on, I had not written the majority of the code. Navigating an unknown codebase is a developer's daily job. Nothing new.

Failing to reread the code or understand the project in front of you is not the LLM's fault.

"Vibe coding creates technical debt"

Catastrophic codebases existed long before AI. The problem was never the tool.

Failing to challenge choices, failing to understand them, and letting things slide is not the LLM's fault.

"You can't justify your library choices"

Before, some people took the first Stack Overflow result or a colleague's suggestion.

That situation was not different. The curious still want to understand. Others will accelerate their ignorance.

The real question

The real question is not AI or not.

Have we already put a hammer on trial?

AI amplifies what exists. In a rigorous frame it accelerates. In a missing frame it accelerates debt.

What AI changes fundamentally: the frame becomes even more indispensable than before. Tests, standards, reviews, code and vulnerability analysis.

That is why I standardise my developments as much as possible. Not to protect myself from AI. To get the best out of it.

I consider that from day one we are in production.