It is possible that at least educated Latin speakers, who would have known Greek, would have pronounced it in a Greek-ish way, with the accent on the i.
I don't know if there is any evidence for what was done in classical Latin with Greek borrowings with an original accent that violated Latin accentuation rules. I only know that some medieval Latin poets retained the Greek accent placement in such words.
In words with 4 syllables that start in 3 short syllables (like religio, Italia), poets often took the liberty of lengthening the first syllable. I don't know if the same practice was classically accepted with 5-syllable, but Boetius (who is rather late) did it with "philosophiae" – he lengthened both i's:
BOETH. elog. 1, 1
Hic iacet interpres et alumnus phīlosophīae
Emeritus famam super astra Boetius ille,
Quo Latium gaudet, dolet olim Grecia uicta.
He also lengthened the i in "sophiae" ... and even the o:
BOETH. elog. 3, 5
Hoc in sarcophago iacet ecce Boetius arto
Magnus et omnimodo magnificandus homo,
Quem sōphīa suis prae cunctis compsit alumnis,
Quem sibi grande decus contulit ipse deus.
Earlier poets read "sophiae" as an anapest, though:
ENN. ann. 211
Nec quisquam sophiam, sapientia quae perhibetur,
In somnis uidit prius quam sam (=suam) discere coepit
It could be that they retained the Greek accent and then instinctively lengthened the i because a short i being accented in that position was so unnatural in Latin.
There are a lot of words ended in -ia. For instance, Regilla's name in Latin: Appia Annia Regilla Atilia Caucidia Tertulla. (In Greek, however: Ἀππία Ἄννια Ῥήγιλλα Ἀτειλία Καυκιδία Τερτύλλα.)
Considering that a lot of names in Latin are ended in -ius, fem. in -ia (Cornelius, Claudius, Aemilius, Iulius and so on, and so on...), the common man would naturally say sóphia and philosóphia.
I had no clue the a was long in Greek! 
