It's more a matter of people misusing the terms than any sort of "scale" shifting. Officially, the term "low-level language" still only refers to programming languages whose structure corresponds directly to that of the intended an instruction set architecture, i.e. machine code and assembly languages. Everything else is a "high-level language", according to the academic definition. That includes C, C++, FORTRAN, COBOL, Java, Python, Typescript and Swift.
Of course, we all know that there's a pretty marked difference in programming in a language like C vs a language like Python, for example. So often people use "higher-level" or "lower-level" to express this comparison. I wouldn't even argue that those are bad terms to express that difference, but over time people have conflated "C is a lower-level language than Python" with "C is a low level language", which brings us to where we are now where "high-level language" and "low-level language" have, as you said, fuzzy definitions in colloquial usage.
The scale is useless because the center has moved over time?
It used to be less fuzzy (assembled -- compiled) and now has more components (dynamic typing, garbage collection), but I don't see how anyone could deduce that the scale is nolonger a meaningful high-level (:^o) classification of a language.
Of course, we all know that there's a pretty marked difference in programming in a language like C vs a language like Python, for example. So often people use "higher-level" or "lower-level" to express this comparison. I wouldn't even argue that those are bad terms to express that difference, but over time people have conflated "C is a lower-level language than Python" with "C is a low level language", which brings us to where we are now where "high-level language" and "low-level language" have, as you said, fuzzy definitions in colloquial usage.