You've raised a very good question and you may (or not) be surprised that I've been pondering it for about 15 years. The other comments on the thread can be answered first.
Your question is good and easier to answer. At "SBL" or any academic conference you must assume that the audience canNOT understand what you read in Greek. Pronunciation is irrelevant. If you read a sentence outloud in Greek, you better have it printed in a handout. The processing apparatus in the head of most listeners cannot process whole sentences that are read at the speed of normal speech. About halfway through the first sentence the input collapses into a jumble and listeners are picking out pieces of recognition. Now listeners may claim that they couldn't follow because of a dialectical difference in pronunciation, but even if you read in pure "Astartean US Seminary" they will not fully follow. However, if it is a quote from the NT they may follow along supported by previous memory. If the quote is from outside the NT (and not the first lines of the Illiad), chances are slim that anyone would follow completely. Finally, where pronunciation does make a difference is in quoting single words. If I am talking about ke/dhe ratios in Greek prose (και/δε) it takes a few moments for the 'coin to drop' (do people still know this idiom?). By the end of the presentation there is no problem, though afterwards I might get a comment like 'you use modern pronunciation'. Which I don't, of course.
Yes, exactly. Sidney Allen wrote the book, and Stephen Daitz has done the most extensive recording of our best shot at 5th century Attic Greek. I have full respect for teachers who would attempt to speak in such a system. It may be compared to Chaucerian teachers who would try to speak Chaucerian English. However, this is NOT the same thing as "Erasmian" and most classicists still use a compromised Erasmian, typically puffing on pi, taw, kappa, going fricative on P(h)i, T(h)eta, K(h)i, inconsistently pronouncing length, sometimes inverting ω and ο, η and ε, and without dropping 'baru' accents or distinguishing perispomenos. I haven't done a survey, but my feeling is that as classics programs shrink in the US, the percentage of teachers using Allen-Daitz will rise.
Back to James--
The question for classics can be two-fold. What does one speak with? And what does one read with? If I were a classicist and spent all of my time in the 4th century and earlier I would consider adopting Allen-Daitz. As a publisher, I would certainly replace any "Erasmian" recordings with Allen-Daitz, (though see following about adding Koine). If I were a mixed Koine/NT/Attic teacher, I would consider Josephus as my guide. (I would ignore Josephus' confession that he did not have a native Greek accent, I don't and won't either.) He was obviously fairly well trained. We can assume that he spoke Greek with a 1st century pronunciation system and would presumably have done most of his reading from previous eras with that system. The DeadSea Greek texts probably document something close to Josephus'.) However, he may have had teachers who would have taught a 'literary reading style', an archaic reading for poetry. In any case, as an academic I would want to know how to read in an archaic style, even if I spoke with a contemporary Koine style. And sometimes when spelling English to oneself it is helpful to artificially pronounce 'ca-no-oo' "know". If classicists follow Allen-Daitz, however, they need to know that the system is kakophonous to native Greeks and they probably need to learn to follow Greek when read in Koine and/or modern accents. Our world is shrinking and isolation is a diminishing luxury.
Demographics may also play a role here. How many 'classics-only' programs are there, in comparison to NT/Koine programs or mixed programs with heavy Koine (LXX/NT/Church fathers/Yosepos/Epiktetos/Ploutarchos)? My impression is that a lot more people work with Koine than classics-only. Time will tell, though I would recommend a Koine (kueneh) for NT and mixed. The issue revolves around where a person wants to end up when training has matured and solidified? Astartean? Most shudder at that (though Markos is a stauch defender of the Ben Franklin tradition ['the French should speak like me']). Allen-Daitz? Koine? Modern? 6-vowel Koine/Byz? (This is a third-century CE Koine, where Hta is Iota)
From a modern perspective one may also re-orient along the above lines. If one speaks modern Greek, they could learn to add two vowels (η and υ/οι) when reading ancient/Koine Greek. This is easily adaptable and provides a framework that can 'carry' the language: ΥΜΕΙΣ and ΗΜΕΙΣ can be heard. Things still sound 'like Greek', though a little stilted. It is an easy rapprochement between native Greeks and ancient Greek. For what its worth in the opposite direction, after adopting a KOINE speech, it is easy to follow a 'modern' reading of NT.