This is rather binary (and naive) distinction. There is a lot of room for creative work in any field.
> ... between jobs that involve being outside (gardener, railroad worker) and those that don't (office worker, bus driver, ...)
> ... between jobs that involve creative expression of your emotions (artist, musician) and those that don't (office worker, bus driver, ...)
> ... between jobs that involve saving people's lives (firefighter, doctor) and those that don't (office worker, bus driver, ...)
In the above, "office worker" includes desk jobs like programmer, startup CEO, and secretary. To people in some careers, those all sound pretty much the same.
A less extreme version of that third example would be jobs that are more people-based (in a meaningful way e.g. perhaps including travel agent but not barista) and those that aren't, which would be a very significant split for some people.
----
As a separate point, and probably a bit more pendantic: I think that even "making new stuff" can mean two different things. If you look at manufacturing e.g. building cars, you typically have a split between interesting design jobs and the factory floor where stuff actually gets built but the job is pretty boring and maybe sometimes even low skilled. Computer programming, even low-level implementation, is more analagous to design in manufacturing (then the "creating" is the trivial file copy of the executable).
But again, to some people that distinction is important and they would rather be building something physical with their hands (plasterer, carpenter, etc.), even if it's to someone else's design, rather than doing something that doesn't involve fine motor skill even if it's "creating" software or car designs.
The strength is in being able to select the first eigenvalues.
After all, the only real thing is the thing itself. A textual description of the thing is not the thing, Cecil’s lost his pipe. So then it falls to you to describe the thing well. To a man standing in front of ten cars, if you say, “the black one” he knows which one. If you say, “it is a motor vehicle made by Mercedes, noted for their three point logo, that is currently painted black” you have more data but not more information.
All the capable people generalize. That is where predictive strength comes from.
Most people in science and engineering can think of equally (if not more) clever and thought-provoking sentences, but PG's carry more weight because of who he is. We still live in a world where thought-leading is dominated by who has the most attention.
In software, to invent and make something work, you spend maybe 10% of the time actually inventing/creating things, and for the remaining 90% of the time, you try to "instantiate" your creation, which involves lots of drudgework, code bureaucracy (and sometimes real bureaucracy), dealing with irrelevant crap (e.g. tooling issues) and debugging.
I suspect he thinks it’s the former, when, of course, it’s the latter.
I'd say it is more about discovering what's already there :)
Anyway, the thread related to this dichotomy is great: https://twitter.com/paulg/status/1250840771751809024
(I just checked with my kid if he knew what innovative is and he answered it mostly right, but he's 13)