Large language AI models do not behave with consistency, which is harmless in some business contexts but dangerous in others. Adapting AI for determinism—the classic promise of software—is a major ...