> Has that understanding changed?

All you can measure is correlation. The rest are theories, and always have been.

"Heritability" is a misnomer, since research into genetics relies on twin studies, so it is impossible to delineate what is 'inherited' from influences during prenatal development. Since technology to "look into the womb" has vastly improved in the past decades, there is more and more research into prenatal effects, which shifts the potential narrative (working theory!) from "it's in the genes" to "it is prenatal trauma due to adverse environmental circumstances".

We don't really know if there ever was or is a "genetic predisposition". Remember, in contrast to "diagnoses" in medicine, psychological classifications such as "schizophrenia" do not describe etiology or biology, they merely describe observable symptoms, with a lot of overlap and redefinitions in between the different categories. More and more voices in the psychotherapeutic community argue that these classifications do more harm than good and should be replaced by a multidimensional system. At least in Europe, my understanding is that they serve mostly health insurance billing purposes, not patient-oriented treatment purposes.

The danger in the "genetic predisposition" line of arguments is that it may sound like something that cannot be "healed", only managed, which we know now thanks to epigenetic research and [brain] plasticity is not the case. Modern therapies can achieve more than merely manage symptoms, and what was previously believed to be "untreatable" is now known to be fixable. Which, as an aside, is one of the reasons why "narcisstic personality disorder" has been dropped, so "it doesn't exist any more".