There is a continuum of cheating, spot responses as a quizzed student is sent an SMS answer and tries to see it without being spotted. At the other extreme is the formalised exam/test where a similar set of opposed data/cheats are at play. In many end cases are the open book exams where the student takes whatever he wishes to riffle through to create his best response to get the highest marks.
The crux is:- assessing how well a student has mastered a topic becomes increasingly difficult as the final years and tiers of graduate school are encountered. Often the student is smarter in intellect than the examiner, but has a lower level of achievement - at that time.
Often lower grades can be assessed a lot faster via multiple choice questions that can be marked with an aperture card very quickly
There are many facets to teaching, the explanation of an aspect, to query students to assess how they understand that aspect. This can be spot tests where one aspect is asked what is it etc, often in a short interval in the last few minutes prior to end of that period. They can also be weekly/monthly/end of semester. These will vary, with end of a semester being a 1-2 hour detailed question set to which the student answers question by question, the weekly/monthly = lesser length/detail.
At each stage the student is asked to show how he has learned each of the listed test questions and the teacher later assesses the answers and gives the student points on each question so that at the end a summation will give an indication of the degree of mastery - students are often then instructed on the deficient areas.
Obviously the students can hire an expert to take their exam, as a group hire, or each student hiring his own expert to sit the exam for them. LOL - this and other schemes have been tried and they are part of the education burden. At University of Toronto exams are closely watched, with photo ID needed for just this reason. Dozens of 'ringers', taking exams for students, have been discovered. It is often hard to correctly ID students from various ethnic back grounds when the invigilator is from a different ethic/racial back ground who is not famiiar with the students back ground. The students search for the best possible 'ringer' to minimize the risk of discovery.
In summation - all aspect of teaching involve assessing the student as well as the teacher. Anything that reduces how well a student studies or engages in diversions - such as games played in class, or passes data between students to allow a correct answer to be given to a teacher that is not from the student's own knowledge base is a form of fraud. The student bears the cost of the fraud when he fails his exams = the money/time spent are wasted. He may get his degree, and then his patients suffer, if he is a doctor/dentist, or a house fire if he is a bad electrician granted his ticket by fraud.
AI is in a good position to teach topics, as well as to test how well topics have been mastered and to also pose detailed exams for students on a case by case basis, with a detailed back ground knowledge of that student's past work so he/she can be examined in exhaustive detail - detail normal examiners could not do due to time constraints. This area is in rapid expansion as we speak because costs and time restraints in teaching are causing large deficiencies in student bodies, to such a degree that student scores are falling steadily, ad the field is ripe for a revolution.
