In recent years, Automatic Item Generation (AIG) has increasingly shifted from theoretical research to operational implementation, a shift raising some unforeseen practical challenges. Specifically, generating high-quality answer choices presents several challenges such as ensuring that answer choices blend in nicely together for all possible item stems and reducing the labor required to write computer specifications for each answer choice. In this paper, using the context of K-12 mathematics items, I give an overview of the unique challenges associated with developing answer choices for automatically-generated items and then illustrate a methodological innovation known as the distractor suite that can improve the quality of automatically-generated answer choices while simultaneously reducing the labor to write and review answer choice specifications.