I'd be very surprised if the work is equal. I know in theory it should be, but I think reality is that it will have to be much more programatically grade-able and won't be.
From my memories at GT near the turn of millennium, most work in CS was already programmatically gradable. The course descriptions look every bit as rigorous. Seems the only real difference is the level of access to personal assistance in the class.
Same memories here for the everything through senior year of BS, but after that things were more self defined projects with a lot more gray area for functionality, API definitions, etc.
I thought/have heard secondhand here that MS was more along those lines, but I may be completely wrong there.
I only did undergrad there, but was friends with several MS students. It seemed to largely differ course to course. Some were more research/project oriented, and others seemed more like an undergrad course with a narrower focus. Format might have even differed based on professor. Either way it doesn't seem a stretch to use autograding for a MS, at least for non-thesis options.
I'm taking classes for an online program right now where the courses allow online-only part time, in-person part time, and in-person full-time students (a few in both undergraduate and graduate tracks). The last two had many hundreds of enrolled students.
Especially at larger universities, the question is the availability of TAs -- every assignment in all three classes so far has had writing components that require human grading.
Online students require a significantly lower load of in-person support (office hours, multiple lab sections, etc) because they're expected to do more self-teaching.
Stanford's public machine learning and other CS courses with their 10's of thousands of completing students, or EdX's Harvard CS50 class with its 1000's of students are good examples where grading against a rubric seems to scale.
