AI is great for sensitivity (noticing potential issues), but terrible for specivity (giving many false positives).

AI is not uremovedly prone to false positives; in this case, it's being used deliberately to produce them.