You have to accept the fundamental conflict that most PhDs won't get an academic research job, yet the curriculum must be geared to those that will. That is its purpose, to promote the best science possible. Most of them will get some other job, not exactly the job they trained for, maybe as a government regulator, maybe as a lower-level educator. But many top scientists came from (relatively) humble beginnings in PhD programs at state universities, and to eliminate the bottom half of all programs or transform them into training programs for nonacademic jobs would blunt scientific inquiry. In my experience, faculty don't give students false hopes. It seems all I (and many of my colleagues) do is complain about how difficult it is to compete and survive in science, yet the students by and large still want to try to stay in research, knowing full well what the chances are. It really is remarkable. In the arts, of course, it's much worse. It's taken for granted that only a tiny fraction of aspiring actors or concert musicians will get "the jobs they trained for", but no one gripes about how dysfunctional that system is.
As far as specialization, I think that is complete nonsense. At least in the biological sciences, Departments have become almost meaningless except as administrative units. Cross-department collaboration is the norm and most faculty could fit just as well into any of half-dozen departments. If anything, the fact that fields are more interconnected has made specialized work less significant. Journals are increasingly categorized not by field of study but by pecking order. Journal articles are accepted or rejected on the basis of what their perceived impact is, not whether they really fall within the realm of the journal's title.
For advancing scientific knowledge, the entrepreneurial American system, where each investigator competes for grants through peer review, is unsurpassed. For solving society's problems, though, it might be that a more top-down approach, with true visionaries directing larger groups of scientists, might work better. But if so, we have been moving in the opposite direction. The National Laboratories, for example, have been basically defunded and converted into research institutes where faculty compete for outside grants just like everyone else.