Graduate students spend 10-20 hours a week grading homework, it's a very time intensive process. But I agree, HW should only be an experience. The entire grade should be determined by an in class midterm and final.
Entire math departments at multiple lower profile schools will close their doors before he has any actual concern about his future as an academic.
But your point is correct for other professors, much less secure in their positions, who have similar problems with AI and actually do have to concern themselves about using their TAs and RAs productively.
faculty research overhead pays for department TAs. that's why departments with rich professors maintain their TAs even through budget cuts.
> If you have no idea how teaching works in US universities, probably better not to comment on it.
define:irony
The majority comes from tuition budgets, which are assigned per credit-unit. This also pays for faculty salaries. Math is heavily dependent on tuition because they get little overhead. Life sciences is buying millions of dollars of equipment, chemicals, animals, lab staff that incur overhead, none in math.
>departments with rich professors maintain their TAs even through budget cuts
In fact the University of California does not get to keep any overhead. It goes right back into the State of California General Fund and the CA Legislature, as part of the budgeting process, decides where it goes (obviously considering where it came from). During a budget squeeze, overhead can be diverted elsewhere.
Because of this, for a number of years, it was unknown whether the research enterprise at the University of California was actually covering its overhead costs or operating at a loss.
Thanks for explaining to me how the academic system I work in functions. Just like most managers don't literally pay their employees, the company does. The university pays TAs through funds from their various revenue streams, including endowment, tuition, grants to the university, and through indirect costs collected from research grants to researchers, probably including Borcherds own grants. So really a lot of different things pay the TAs, the university is just the clearing house.
They do not, lol.
For one thing, you will always have some courses in the courseload which have assignments that count toward the grade. So of course, you prioritize getting those assignments done, at the cost of procrastinating the ones that don't contribute to the grade, which will tend not to get done. If you don't do those, you will tend to bomb on the exams.
I would instigate a system in which your assignments count toward their grade, but the assignment's marks are derived from the marks that you hit in the exam, not from the content of the assignment.
Like say that we make assignments 20% of the course grade. For the sake of simplicity, let's suppose we use the final exam grade to determine the assignment contribution.
If you do all the assignments and get 80% in the final, your assignments then contribute 16% to your course grade (out of the maximum 20%). If you've done half the assignments then you get 8% from them, and so on.
I would not impose a minimum bar of what constitutes an assignment being done. If the assignment is, say, programming assignment to make a tokenizer for a programming language, and someone hands in a tiny file containing a "hello world" function, then that will still count as the assignment being done. That student will self-punish when they bomb the final exam due to not having learned, and that exam's percentage then determines the value of their disingenuous assignment effort.
I guess that a form of this system has already existed in the past: namely when certain professors made attendance made attendance part of the grade. So just by warming a seat in the lecture you could get 5% whether you are absorbing anything or tuning out.
Grades being based purely on exams is a bad thing; it encourages "examsmanship" rather than scholarship. It's hard to see what else you can do in the face of AI. Work done in uncontrolled conditions can easily be faked. If you embrace AI, then you have to raise the bar for the size and complexity of homework problems. But then the homework measures the student's ability to take on a complex problem with AI, and not the specifics of the subject matter in the course. You can take on arbitrary complexity without understanding a specific subject matter. If we raise the complexity to the point that you fail, it's not from the subject matter, but the complexity. In no way can we arrange "AI allowed" homework such that the student learns a specific topic; the harder we make it, he more effectively the student learns to use AI, while avoiding the actual topic.