This gives an impression all software engineers have some model in their head of complex ever changing systems they work on, that actually fits in that head and stays up to date at the rate at which the software system changes.

The larger and more complex a project gets cognitive limits get hit whether you use AI or not. Do people then just do what someone else more experienced tells them to do? Yes. Cause who likes to be held responsible for anything unpredictable happening in the universe that trips up your changes?

Theory of Bounded Rationality and its recommendations apply everytime any kind of limit is hit.

As an example of congnitive surrender, the post is clearly co-written via AI.

It still makes a useful distinction.