> we're now entering some kind of abstract space that cares not for the stated intentions of humans

But that's the thing about systems: they may involve humans but they don't necessarily reflect the intentions of the individual humans involved. Even when a system is created with a stated intent (i.e. for a stated purpose) that doesn't mean it will actually behave in a way that aligns with this intent. Logically you then shouldn't take the human intent into consideration when analyzing a system's actual effects and outcomes (except to determine whether it aligns with those but that's secondary).

IOW the purpose of a system (i.e. "what it exists for") can be different from the purpose for which it was created (i.e. "what it is meant to do"). I guess "purpose" in this case is an overloaded term because the former more uses a meaning that more closely aligns with "function" (like the function of a predator in an ecosystem may be controlling prey population but that doesn't suggest intent nor design) and the latter uses a meaning that more closely aligns with "intent" (like during wildfires controlled burns are performed with the intent of stopping the spread of the wildfire).

But I'd say it's a stretch to apply this to statements like "the purpose of organisms is to increase entropy" because that strongly implies intent rather than function (because the latter could also be simply expressed as "organisms create entropy").

POSIWID is usually the end result of asking some basic questions about "A System":

  -Has the System taken (on) a mind of its own?

  -What does It want?

  -How do we know what It wants?
For organisms, sometimes just asking it directly can give more useful answers (or surprises,YMMV)