One is whether AI can produce a plausible answer. The other is whether the system around the AI can produce a defensible ...