How is the notion that "the vast majority of curriculum (quantum, EM, mechanics) are things THAT HAVE BARELY CHANGED IN THE PAST 50 YEARS" relevant to this discussion?
Because the author draws a lot of parallels between computer science majors and physics majors, citing that physics majors are better prepared for the type of work ML requires. I am a physics major turned data scientist, and my argument is that I would have better prepared having been a CS major, given what the majority of my work requires. While in a CS major, a lot of data structures and algorithms haven't changed in 50 years either, you're much more likely to take electives with marketable skills or with up-to-date technologies (distributed systems, operating systems, OO/fp, databases, concurrency) that would've helped me in my day to day more than a math course or too did during my physics degree.