Off topic because I am skeptical of the claims in the first place. As a general principle: the licensing exam is not the benchmark AI should be evaluated against to then go about calling itself a doctor.

I'm from a different country but these exams are the minimum standard to demonstrate a doctor is safe prior to interacting with patients. To be really explicit, the core competency being assessed is identifying potentially serious situations and answering the same way every time: "I WOULD CALL FOR HELP" +/- principles of basic care.

The benchmark for doctors actually making decisions about patient care are the assessments to become fully qualified consultants in a each specialty.

Again, to be really explicit don't confuse a test for "doctor won't immediately kill someone and commence reasonable first steps" with an actual "doctor with years of experience and subspecialty training who regularly makes decisions about patient care".