Anyone who follows AI ethics and safety debates can recognize the pattern: worries about bias, about “hallucinations,” about ...