See, the thing is, we have to stop treating competent bilingual English-users are "mere learners." We can't treat their language use or their metalingusitic talk about their language as deficient. Even if that is how they want to see themselves. (There, I said it! I'm an academic asserting that I actually know better than the people whose behavior I am studying! I have crossed over to the dark side!) Modesty about your ability in an L2 is one thing. I am exceedingly modest about my L2 abilities (not without cause, in my opinion). But English is a special case. It is a lingua franca. It is a working language for millions of non-native speakers. It is a language that L2 English teachers function in at a high level.
I believe my Chinese colleagues who teach English at the tertiary level to be highly competent users of English. (I believe many upper-level L2 college students and certainly grad students are highly competent users of English as well.) Certainly I believe my professors at UBC who speak English as a second language are highly competent users of English. Is there a difference between their English usage/knowledge and mine? Sure, in some abstract way. I am 'closer' to English on a personal level than they are in some ways, though I would argue that many of them are closer to it in a professional way.
Moving acceptability judgments from the domain of theoretical linguistics frees it from accusations that it is not actually measuring what it claims to be measuring -- linguistic/grammatical knowledge at a deep/unconscious mental level.
Moving acceptability judgments from the domain of second language acquisition frees it from the undue burden of having to provide an accurate map of a learner's 'interlinguistic competence' (to use Birdsong's term).
Putting them squarely in the domain of sociolinguistics, language attitudes, and world Englishes, etc. allows AJTs to be considered as another kind of behavioral data, elicited from people who use English in a specific way and a specific context, that can be analysed in a social practice framework.
This AJT data can then be considered in light of the sociocultural realities of English around the world -- insead of in terms of an abstract generative linguistics based on native-speaker intuitions (which is not that relevant to applied linguistics, much of the time), or a gauge of how well someone is learning the 'target language' which should conform to native speaker standards.