Really experiencing the “alignment” problem with AI right now. Even when the task is data summarization in an objective way (x grew by Y%) different people expect the information to be presented in different (and slightly incompatible) ways