- "Many worlds" concept was promoted by Hugh Everett originally (1957) and it usually means, different observers would see the same space-time event differently.
- Parallel universe concept usually considers the existence of another, less or more similar - just causally separated - universe at the different place of space-time.
- Multiverse in string theory interpretation usually means, at some (if not all) places of our space-time another space-time events may occur in hidden dimensions set alternatively.
If we define our universe like place, where known laws of existing physics remains valid, we can say, our Universe is defined as a place, where some sufficiently general theory of everything remains valid. If string theory is such a theory, for example, we can postulate our Universe as a place, where string theory remains valid and the parallel universe will be place, where string theory isn't valid. Such definition would mean, string theory isn't TOE anymore, or it even contradicts itself, as it predict something, which violates the string theory. Which is probably not, what the string theorists (prof. Brian Greene or Michio Kaku in particular) expects/promote.
OK, we can define a parallel universe as a place, where all existing physical laws remains valid, it just exhibits another configuration or different physical constants. For example, it's the place, where existing particles are creating different structures, then those, which we know from our Universe. For example, in tropic areas electrons and protons are forming rain forests, whereas in Arctic areas they're forming ice only - is that what the multiverse means? Supposedly not. So, we can consider, in multiverse only gravitons (axions or neutrino or black holes..) can exists, all remaining particles or structures are of different nature, stability, etc. As such multiverse could serve the interior of neutron star, after then. Which is still not, what we are looking for.
Another definition of multiverse may be based on relativity phenomena. Each of observer would see a bit different reality, while these realities are still tied together by common radiative time arrow. For example, observer in relative motion would see the same muon less stable or even decomposed, then observer at rest. It seems, such situation still doesn't fulfill the definition of multiverse, as it falls belong realm of predictions of classical relativity.
The uniqueness of radiative time arrow would be violated at the case, when light signals propagates through density gradient of vacuum, so that the gravitational lensing takes place. It means, a single space-time events has a multiple consequences at different places. Is this, what the multiverse means, already? We should realize, the violation of radiative time arrow would mean the violation of Lorentz symmetry as well, which is prohibited in strictly rigorous string theory, because this theory is using special relativity in its derivations and as such it uses Lorentz symmetry as one of its fundamental postulates.
At the case of rotating black hole, the location of event horizon would depend on polarization of light - vacuum exhibits a birefringence and related phenomena, vacuum analogies of Faraday and Kerr effects. It's the another, even deeper manifestation of quantum uncertainty at macroscopic scale. Everett's interpretation of quantum entanglement uses the many worlds concept - is such manifestation of quantum uncertainty what, the multiverse means in string theory? It seems not, as string theory itself is quantum mechanics based already - the prediction of quantum uncertainty in quantum mechanic way would become a trivial conclusion here.
In AWT the existence of multiverse depends solely on observational perspective and the scope of observations. From AWT perspective, which is most general possible one, no real place for multiverse concept can exists, because everything fulfills the very same theory here. If it doesn't, then the AWT cannot predict it in causal way. From the above follows, observational perspective is always defined by observer. Because people are the only known creatures, who can observe the things like electrons, photons and black holes, the observational scope is defined by abilities of people to observe these artifacts. It has no meaning to ask, whether the multiverse can appear differently, because we couldn't see them at all. Which effectively means, no multiverse can exist here, or it would become a part of our Universe.
Because the causal perspective of people is defined by transversal energy wave spreading inside of random system of Aether particles, we would see whole Universe exactly the same. If some newly revealed physical laws would enable us to see a larger portion from our Universe, then the newly revealed portion of Universe would share the same laws with the portion of Universe observed by now. For example, we can consider, if the observable portion of Universe appears like interior of black hole, then the outer Hyper-Universe would appear exactly the same, like the observable generation of Universe - or we could never interact with it, to predict or to see it the less.
Such conclusion fits the Copernican principle in particular, by which every observable portion of Universe doesn't differ significantly from another portions of Universe and our location in Universe is not significant in any way. This theorem corresponds the anthropic principle, in which every product of long term evolution would see its environment as a best place for its life - just because it's a product of long term adaptation to this environment. From this perspective anthropic principle becomes a sort of self-referencing tautology. Which means, you cannot predict nothing very much usefull with using it, but you cannot disprove it so easily as well.
Some positivistic people aren't convenient with antropic view of reality, though - as they believe, the random model of Universe isn't sufficient for explanation of "unique character" of observable reality. They didn't realize, inside of dense random particle environment only transversal energy can propagate at distance, because longitudinal waves compensate mutually even at short distance. To be able to see our Universe sufficiently large, we should become adopted perfectly to observation of Universe in causal way, i.e. with transversal waves, which puts rather strict limit to the possible appearance of universal randomness. And the deeper in space-time perspective we are supposed to see, the stronger such limit should follow.