one thing that seems like an obvious application for machine learning is databases: you "train" a model by just using the database and it automatically creates and manages indexes based on what's in it, how it's accessed, and how much space you've allocated for indexing who's working on that?