Earlier quoted context omitted.
Creating a new universe inside our own would not prove that universes other than our own and the newly created one exist. Sure, it proves that it is possible, but that would not be surprising. The argument is that we have no physical argument that can either prove or disprove the existence of extremely distant other universes, and simply creating one would not move that debate in any direction at all. This goes even…
"a theory that does not make testable predictions." When people use this phrase, I think they are often referring to a theory that "does not make testable predictions that are different from the mainstream accepted theory ". But I wonder to what extent new science requires looking at familiar things differently first ?
However, string theory does make significant new predictions, though a lot of them are untestable at the moment. The problem with this is that if you accept string theory you have to accept, for example, that there are 10 spatial dimensions, though we can't notice them in any way. You may have to accept that there exist many (or an infinity) of other universes that are too far for us to be able even theoretically to notice.
The meat of the issue is this: physics is ultimately concerned with describing what the real world looks like and how it works. When evaluating a physical theory, we shouldn't care just about its self-consistency, but also about whether it is likely to be a good model of the physical world. And to be able to tell, we need to test all predictions of this theory, especially the counter-intuitive ones, and check whether they actually describe the real world, not simply whether they are theoretically possible. Sure, if one or two new predictions remain elusive, but we've checked all of the others, we may have good reason to believe the theory (e.g. we had good reason to believe general relativity even though it predicted black holes even before we were able to observe the first black hole, because it made so many other predictions that we did confirm, and that no other theory could describe).
A good example of a theory that is still floating around, that doesn't make new predictions compared to QM, but that is more desirable in other ways is the DeBroglie-Bohm pilot-wave theory of QM. It's not yet complete (it can't account for all aspects of QM), so there is still a good chance it is simply wrong. But, IF it will be able to account for all of QM, even without making any new predictions of its own, it might become a much more popular new theory, on account of simply being more intuitive (no more particle-wave duality, no more randomness, and particles would have properties even before you measured them; note however that it does have non-locality, so it's not all roses).