r/MachineLearning - [D] Societal problems/topics that can be reasonably tackled during a Computer Vision/ML PhD while generating interesting research from a theory standpoint.