I think (because this is how it was for me), they're useful to a point where problems with the music is more objective and agreeable by everyone. Once stuff gets subjective (which is a lot earlier than people would think), the feedback might continue contributing to shaping how the person writes and produces, but isn't at all necessary for that person to keep improving.
Like you said people who have the drive for this stuff will keep going, and I particularly relate to not coming to WIP forums anymore and just bouncing tracks at a few peers instead. When I do it, it's more one-sided. I'm not asking what's wrong with the track or what I should fix, but rather I'm gauging the reaction of a person who isn't already familiar with it just as a fresh set of ears. Just if they might say something like "this ____ part feels like _____ ". I might've clearly perceived that and been fine with that in my taste, but it comes down to how listeners will react to it, so I'll try to compromise a little. It's like when a game dev polls about a control scheme or something. The dev might be perfectly fine with it, and it's not just that they created it, because it really does just work for them, but they just want to make sure other people can enjoy it too, so they send out a survey.
Even my most skilled peers, I'll send stuff to them and on rare occasions they're like "wow this sounds really good", but I'll come back to the stuff a year later and I can still clearly perceive that I've been improving and I can see flaws in the sound I used to have. And those flaws are subjective to my own taste, because it's been shown other people with experienced tastes still enjoy them.
My improvement, personally, is pretty self-driven at this point. I don't think a lot of friends really share much of my influences at this point, and I chase production techniques I don't see my friends doing. I think at a certain point you can absolutely trust yourself to be your own critic (without causing self-esteem issues or creative paralysis).