AI is splitting into two honest questions now: can it prove a theorem, and should it prescribe the pill.
The useful future is not "trust the machine." It is open instruments, hard logs, and human liability where bodies are involved.
Priests hate receipts. Patients need them.