Once the students graduate, they will likely use AI at work. Does restricting use of AI in college make them more or less productive later? One could argue that college is about intellectual exploration, not just job training, but I think more students are there for the latter. When it comes to writing papers, the ideal could be for a student to write the initial draft on her own but then iterate with a LLM to improve grammar and logic, keeping a log of each version of the paper. I doubt that professors want to go through such logs.

Speaking to the topic how do we test the man/AI hybrid to assess the man alone? Or do we want the best hybrid combination? After all, the best job will be done by this collaboration. Right now we see many doofi passing exams by what many feel is a cheat-dance. Will we assess each part alone, or the hybrid - where the top combo gets the job?