Well, if they're bought in bulk, there's gonna be some duplicates. They probably don't know what they're buying until they start scanning.
But still, the number of duplicates will be inversely proportational to the rarity of the book. Lots of copies of James Patterson and Twilight, probably few or zero rare 16th-century manuscripts.
I would assume the very first step would be cataloging the author, title, and revision of the book so hundreds of thousands of hours of time are not wasted scanning duplicates.
They probably still throw away the duplicates even if they don't scan them, because the wholesale value of a used paperback of the Da Vinci Code is like half a cent.