Project Intro: Selection Theorems for Modularity — AI Alignment Forum