On the other hand, yes. On the other hand, especially in non-elite universities, the education should be structured so that the most natural and easiest way to succeed at the studies is to learn something and study, rather than cheat.

The problem, which I experienced in the past: If you put no pressure on students to regularly do exercises themselves, most of them will not do them, so most of them will do very badly in the exam. Now you can try to fail 90% of the students, and some lecturers do, but this does not only result in political pressure on the lecturer, but will also lead to other undesired consequences like many students repeating the semester or very few students in higher semesters (also meaning: less money for your faculty).

@JFabianMeier Certainly. I'm all in favor of putting the pressure on them during the whole semester. However you'd better first tell me how exactly I am supposed to do it in the uncontrolled situation when, say, they are doing their homework without supervision. If I had the powers of an army drill sergeant, that would solve the problem, but I don't. And giving long speeches about "academic honesty" or telling them that "one can cheat only oneself on the long run" is a sheer waste of time. The only way I know is a cold shower when they get failed on the first exam that is given early.

@JFabianMeier That brings some (but never all!) of them to their senses. But you are absolutely right and if you have some better ideas that do work for you, I'm all ears :-) As to "many students repeating the semester", it is a desired outcome because many of them do need to take the course twice even if they honestly try (poor background and lack of basic skills is another problem, but it deserves a separate thread)

I like this approach, because it gives people an opportunity to learn how to use ChatGPT as a tool, while still requiring the student to actually learn the skill. More than likely, machine learning will be tools that people use at their jobs, which requires 2 things: the person understands the base-level skills (learned in an educational setting) and the person understands how to use the tool. Understanding the skills and the tool will make that person more effective. Pandora's Box is open. We cannot close it. Just work with it. The rest is up to the student.

@GregBurghardt Yeah, IMHO, ChatGPT is no worse and no better than an advanced graphing calculator in a calculus class (and those have been there for decades). Helpful in good hands, destructive in bad (for the user). Everybody should learn how to solve a routine problem by hand on nice data and do it 3 times in his/her life, but starting with the fourth, the routine should be delegated to the machine and the human task should be upgraded from executing the algorithm to understanding what and when can go wrong with it. But without the first 3 times this upgrade is just unfeasible.

@JFabianMeier If your concern is that they will not be prepared for the in-person no-internet exam, you should be giving them a course syllabus at the start so that they know that will be coming up, and emphasize when handing it out that online tools (call out ChatGPT specifically if that's your concern) will not be allowed.

@Zibbobz you should be giving them a course syllabus at the start so that they know that will be coming up At the USA universities it is the standard (almost legal) requirement. Also it saves you a lot of time and energy: when they start whining, one can just say "read syllabus, lines 16 to 24".

@Zibbobz Why does it help you that the students know that? If they don't do the exercises, the results will be bad. And they don't believe you that the exercises are necessary to do the exam. Their experience from high school is usually that teachers always warn you, but then the exam is easy.

@JFabianMeier Well along with what fedja said just above (it is a standard requirement for many universities) and gives them reasonable warning, you also have to remember that they are not high school students - they are college students and part of learning in college is learning that when they are told about the consequences of their actions (or inaction), that it will have actual consequences. Whether or not they learn from it is still up to them - but it is to their service and your own that you give them ample warning upfront.

You're right in principle, but it just doesn't work that way. It is too easy for a student to say (in good faith) "I'll use a ChatGPT answer now and learn at the end of the term" and then put off learning until it is too late. Mandatory tasks are supposed to motivate people to learn during the semester, as people are bad at time management and may realise too late that they have set the wrong priorities. I don't even want to weed out students, I want to motivate them so that as many as possible have a good chance of passing the exam.

@JFabianMeier: IMHO it’s enough motivation if the course work gets graded and makes up some part of the final grade. I’ve had courses where it was theoretically possible to pass just by doing great on the final exam, but very few people would risk it intentionally.

@JFabianMeier And? I don't understand your argument. "If they don't do the exercises, the results will be bad." is as it should be. You learn pretty fast whether or not you need to do the exercises. That's why homework is worth like 5% of the mark so it's not nothing, but it doesn't count so much that it's worth cheating on them but makes little difference if you do. "Their experience from high school is usually that teachers always warn you, but then the exam is easy." So have an in-class quiz once month in to give them a taste of reality. If they still don't learn, that's on them.

@DKNguyen If large parts of the students do badly in the exam, it is not only "on them". As I explained, failing 90% of the students usually causes other problems, like political pressure, bad reputation for the lecturer, many students repeating the semester, too few students in the next semester courses etc.

@Michael Using course work for the grade becomes hard if people can just copy&paste from the internet.

@JFabianMeier But you're specifically talking about 90% of the students failing specifically because they all choose not to do the exercises? Come on. If class with that so many students that don't even glance at the homework can even really exist there's nothing you can do and they do deserve to fail. Are you willing to inflate grades in that scenario to avoid said problems? You seem to want to operate under the assumption that every student is negligent, even with exercises worth a small amount of the mark, even with a quiz early on to give them a taste of what exams may be like.

@JFabianMeier You don't care if they are copy/pasting homework solutions. That's one of the reason it's worth so little (usually 5%) but not worth nothing so that it does not explicitly feels like a waste of time. The other reason it's not worth a lot is to disincentive cheating and incentivize taking a shot at it. It's not worth enough that cheating on it, nor doing poorly on it, would really make a difference. You can only lead a horse to the water but you can't get it to drink.

@JFabianMeier What does that even mean to have a requirement to solve 50%? Required for what? What happens if you did 50% but it was not correct? I don't understand. At my university there was no requirement whatsoever. Homework was just worth 5% and that was what. You completed or copied however much you wanted to, and submitted it for marking and that counted towards the 5%. Everything seemed fine.

@DKNguyen What does that even mean to have a requirement to solve 50%? Required for what? That I can explain. When I was a student (in Russia), this (actually rather 80%) was the requirement to be allowed to take the exam (until you've done it, you wouldn't be even admitted into the exam room; there was only one exam at the end of the course, and it was oral)

@fedja I see. That was not the case at my university. If you so chose, you could just not do any homework at all and only write the exams (and attend the labs if there were any). On some occasions when things lined up really badly between classes students would selectively not do certain assignments or finish problems that were taking an inordinate amount of time. I don't think I ever had a course with just one exam though I may have just forgotten the one. They always had at least a midterm and a final.

@fedja But what constitutes "solving" 50%? Does that mean getting 50% correct? Or just attempting a solution, any solution to finality regardless of correctness? Also, are these classes curved? Virtually all my classes were curved.

@DKNguyen With me all homework was given (and checked) on the recitation sessions. The typical task (when studying the integration techniques, say) was to compute 100 definite and indefinite integrals. We were given the list and could then submit them in any order any time during the rest of the semester. If something was wrong, it had to be redone. No formal score was assigned but you had no chance to get "zachjot" (pass to an exam) until you solved at least 80 correctly (80 it was for my particular instructor, but that was a typical percentage)

@fedja What were your class sizes?

Ah, that's interesting. A typical student group was about 25-35 people and each group had an individual recitation section schedule and instructor. The lectures were given to 3-4 groups simultaneously and the exams were conducted for the whole body of students (10-12 groups) on one day in several big rooms with all professors and instructors from the corresponding subdivisions and some of their graduate students involved as examiners. It was a full day ordeal for the faculty. :-)

@fedja There's the difference. My engineering, math, and physics classes were mostly 100+ students per professor...in first year and second anyways where OP's question is most pertinent. Your descriptions are more akin to my foreign language classes with only 20-30 students.

@DKNguyen So you didn't have recitations/labs in small groups and one professor had to assign and grade all 100 homeworks? In my case one professor delivered lectures to 3-4 groups at once (the same 100 number), but lectures only.

@fedja The professor had teaching assistants that graded the homework, and the labs were smaller and also run by teaching assistants. But your recitations sound to me sound like my seminars. These were kind of like labs for courses that did not require equipment so we just sat there solving problems packages. However, courses that had seminars were quite rare. They only really existed for particularly brutal first and second year courses. And certainly nothing oral in any case.

@DKNguyen Well, we were in pure math. The only labs were the computer rooms when we practiced programming. Those were even smaller because there was just one big machine with a few terminals (it was in 1980s and Russia was quite behind the US at that time in terms of computer development) and half of the time the operating system crashed in the middle of the lab time, so it was quite funny and dramatic experience :lol: Engineering or physics might be different but I don't know much about it.

@fedja Also exams were conducted for everyone at the same time. I do not recall ever having a math or physics class with a seminar. All the seminars I remember were engineering. I guess pure math labs probably wouldn't involve equipment other than computers. My labs were for things like chemistry, physics, and electronics so equipment other than computers was involved.

Ahh yes, let's use any excuse to regress to the least effective way of assessing whether someone actually understands the material. Proctored exams have been proven time and time again to tell you more about how good a student is at exams than whether they actually know the material. Sure, if someone does well, they probably know the material, but someone who does badly can't be said to necessarily not understand the material. They just might not cope under the synthetic stress that's caused by this style of assessment.

@ScottishTapWater Looks like you totally misunderstood me. I'm not offering to competely abolish all other kinds of assessments or to make the exams stressful. As a matter of fact, I never give more than 5 problems on my hour tests and never give tedious exercises, so if the student understands what should be done, he/she would finish each problem in 5 minutes with computations involving only rationals with small numerators and denominators, so if a student cannot pass such an exam, he/she is certainly just does not understand the material. As to "proved", I wonder by whom and how.

But we are neither policemen, nor babysitters => you kind of are, if you're failing the student in class? If no grades were involved then yes, it wouldn't really matter, but they are involved so you're kind of a policeman

Nope, I ain't. When it comes to grading, I'm an just examiner. Like when you test for a driving (or a pilot: that test is more serious) license, the examiner merely checks your skills and follows standard guidelines to issue his pass/fail verdict. He is completely indifferent to whether you pass or fail and has nothing against you or in your favor. If he is in a good mood, he can drop you a hint or let some minor error slip. That is the approach I take when assigning grades.

At all other times I am an instructor. If the student shows good attitudes and genuine interest in learning, I become passionate about him/her but it he/she tries just to pass through by cheating, whining, and other means like that, I don't care in the slightest if he/she fails (like my flying instructor once said about one of his former students "I was happy to part with him, and if he ever manages to get his license, God save the people he takes on his plane!".