There is a key argument that supports the claim that we need to recreate natural philosophy. It goes like this. Physics only every accepts unified theories, even though endlessly many disunied rivals can always be concocted to fit the phenomena even better. These disunified rivals never get considered for a moment, precisely because they are disunified. (A disunified theory is one that does not ascribe the same laws to all the phenomena to which it applies. It attributes different laws to different phenomena, the greater the number of different laws, the greater the disunity of the theory. It might be called a "patchwork quilt" theory.)
This persistent acceptance of unified theories only, when endlessly many empirically more successful disunified rivals are available, means that physics makes a big, highly problematic assumption about the nature of the universe, whether this is acknowledged or not: the universe is such that all disunified theories are false. It is such, in other words, that some kind of unified pattern of physical law runs through all phenomena. The universe is physically comprehensible.
This assumption needs to be made explicit within physics precisely because it is substantial, influential, purely conjectural, and very likely to be false in the specific form accepted by physics at any stage of its development. It needs to be made explicit so that it can be critically assessed, so that alternatives can be developed and assessed, in the hope that improved versions of the assumption can be developed - versions that do better justice to the way the universe actially is.
A glance at the history of physics reveals that different versions of the assumption have been adopted at different times, all earlier versions now judged to be false. In the 17th and 18th centuries, it was thought that the unvierse is made up of minute corpuscles that interact only by contact - all phenomena everywhere, at all times, being the outcome of corpuscles interacting and being in relative motion. In the 19th century it was thought that the universe is made up of point-atoms interacting by means of rigid forces at a distance. In the late 19th and early 20th century, it was thought the universe is made up of a unified field spread thoughout space, particles being particularly intense regions of the field. This gave way to the idea that the universe is made of quantum particles or a quantum field, which in turn gave way to the idea that it is made up of minute quantum strings in ten or eleven dimensions of space-time.
All these ideas are metaphysical; that is, they are untestable. Nevertheless, they are integral to physics, to science. And it is all-important that they are acknowledged explicitly within physics - precisely because they are metaphysical, conjectural and all-too likely to be false - so that they can be critically assessed, developed and, we may hope, improved. Untestable metaphyiscs, or philosophy, needs in short to become an integral part of physics, and thus an integral part of natural science.
But how are we to improve these untestable, metaphysical conjectures of physics? In In Praise of Natural Philosophy I have put forward a new conception of science which holds that we need to represent physics as making a hierarchy of metaphysical assumptions about the comprehensibility and knowability of the universe. These assumptions become less and less substantial, and so more likely to be true, and more nearly such that they have to be true for science to be possible at all, as one goes up the hierarchy. In this way a framework of relatively unproblematic assumptions (and associated methods) high up in the hierarchy is created within which very much more substantial, specific and problematic assumptions (and associated methods), low down in the hierarchy, can be developed and assessed. These latter assumptions are selected on the basis of which seem to best support empirically successful research programmes within physics, and which accord best with assumptions higher up in the hiearchy. I call this hierarchical view aim-oriented empiricism.
Aim-oriented empiricism brings together physics, metaphysics, and views about what the aims and methods of physics ought to be - the philosophy of physics in other words. Aim-oriented empiricism turns physics into natural philosophy! See especially my In Praise of Natural Philosophy (McGill-Queen's University Press, March 2017), chapter 5.