I’ve always thought the obsession with grading homework was weird. Homework is your chance to practice. It’s totally normal and expected that you’ll make mistakes - so why are students’ final grades being penalized because they made mistakes during practice?
Don’t get me wrong, homework should still be marked for feedback. After all, that’s how you learn what you were doing wrong and prepare for an exam. But students shouldn’t be stressing over correctness in homework. If they do well on exams and projects, that should be sufficient to get a good grade. A student bombing the homework, learning from their mistakes, and doing well on the exam is a good outcome, not one we should punish by factoring homework correctness into a final course grade.
The point of grading homework was just a way to disincentivize students from hurting themselves by not preparing. If it's not graded, then why not bomb it and then cram before the exam? Students tend to be bad at estimating how much they can get done.
This is the ostensible reason, but really it's just lazy instructors trying to ensure they don't look like they are doing a poor job preparing their students when the students fail.
Here's a question: why don't more schools offer credit for "course challenge" tests, where the student just takes the final exam? If a student has the knowledge, why not let them get credit? Is the course evaluative, or is it more about the journey.
My high school chemistry teacher only graded homework if a student received less than 90.00% on a test. I think this worked well and is an appropriate happy medium—even the best students still did the homework and wanted it to be correct.
Edit: people can be a bit hard nosed about “grading homework is dumb” but I was the kind of student where I’d hope homework scores would help raise my grade
I'm in a graduate program, as a student teacher studying to become a high school math teacher. I have a mentor in the classroom, and basically, we only grade the homework for the completion, because we assume the students are going to cheat, so it's only worth 10% of their grade. Even with AI, and just grading for completion, some are still too lazy to do the homework.
It's usually very clear who took the time to do the homework, because they either do well on the quizzes/tests or do poorly.
So, it's not impossible. Just put less weight towards the homework, and more weight towards quizzes/tests/projects/etc.
Even pre-AI, my experience as an adjunct led me to believe there was a lot of cheating on homework and assignments.
Individual programming projects for students wrecked me as a lecturer. The time investment for me for meetings, reviews, and understanding 30+ projects well enough to grade against a higher-level, abstract rubric just took too much time.
Are you happy with the MS of Ed program you are in? I sometimes think about making the same transition to high school math teacher. I find that I do very well with engaged high school students (I volunteer in various ways at my local high school) but haven't quite found my footing with students who (maybe rationally) don't engage with the material.
Why put any weight on the homework at all if those who are completing the homework are also doing well on the tests? It seems like the only people you are advantaging are those using AI (as compared to the 'lazy' students who don't turn anything at all).
If you already assume they are going to cheat, what's the point of even grading it?! If everyone is cheating and outsourcing it to AI, why bother assigning it? Why make it count towards the grade at all? It's just pointless busy-work at that point!
Grading homework was already a stupid idea from the start, as the goal should be to ensure the pupils learn something - and learning is very much not the same as "provided proof of busy-work". If a student can get a genuine 100/100 on the test, who cares whether they are "too lazy" to do the homework - the goal of providing education has been achieved!
If everyone is cheating, then you are literally grading them on something which has zero value beyond busy-work. You might as well ask them to sit in a corner for an hour counting grains of rice, and give them a grade for it...
> If you already assume they are going to cheat, what's the point of even grading it?! If everyone is cheating and outsourcing it to AI, why bother assigning it? Why make it count towards the grade at all? It's just pointless busy-work at that point!
My undergrad (decades ago) was like this. Homework was assigned, but not collected or graded. The purpose of homework was to give you sample problems you should be able to solve, and you could go to office hours for help.
In one engineering class, the professor would assign, but collect (and grade) them very randomly. It was max 10% of the grade. The idea was to ensure you're at least attempting the HW problems and not slacking off.
It's not pointless busywork. It's important work, punishing the naive students who don't yet understand that no good deed goes unpunished, and teaching them they should be cheating, otherwise they are doormats who worked harder for no reason.
Not to mention homework remains one of the best ways to make an uneven playing field in education in general.
> Perversely, by making homework graded you are probably increasing the number of students who feel they need to use AI to get good grades.
I don't think this is false, but it's not the first time homework hasn't been graded. Not that long ago I had professors say they made homework a minor portion (10%-ish) of the total, since not having it be graded left most students not doing it and grades clearly went down.
It's very much a situation where poor incentives lie on both extremes.
This would have helped me. I wouldn't have done the homework if it wasn't graded; if it was, I would do it and not cheat (and do better in the exam because of it). 10% seems like a sweet spot: enough to incentivize the lazy but not a big reward for cheating. You can tweak the grading further so anyone who made an effort is likely to get, say, 60%+, so the extra marks gained by cheaters are even less significant.
Heck, I can't recall ever having been graded on my homework!
In my experience the fallout of not doing the homework was that you'd look like an idiot when called on during the communal walk-through session, or get a bad grade on the exam because you had no idea what was going on. Absolute worst-case scenario, a highschool teacher would get upset and call my parents, I guess?
I was definitely graded on projects I had to do, but that was more "write a book review" assignments than "do exercise 1 through 15" homework.
This is the way forward in my opinion. Pick learning objectives that reflect a world where AI can do most route work, Measure progress against these objectives in a manner where AI use does not matter.
I'm in a course on video game design where one of the objectives is to learn the Unity IDE... which of course doesn't matter anymore. They could have picked the objective "Make a fun and engaging game" which would allow for arbitrary AI usage and require learning the rest of the course material on what makes a game fun/engaging/graphically interesting.
I’m sorry, but if you think learning to use a game engine isn’t relevant to designing and developing games, you are sorely mistaken. The existence of LLMs isn’t an excuse to not learn the fundamentals. That’s like saying learning Python or JavaScript or C++ doesn’t matter - completely absurd. At least, if you want to develop actual skill and not be reliant on LLM providers.
At the German university I studied at, long before AI, completing the homework successfully was sometimes just a prerequisite to be able to take the exam, and the exam counted for 100% of your grade. Some classes had homework but purely for your own independent learning, and there was just a weekly presentation by a TA going through it step by step.
Is that a problem? I'm sure there's another 100 variables. Maybe boys do better on 2 tests wherher girls do better on 3 tests. Or on a longer test vs a shorter one. Or open book vs closed book. Multiple choice vs open questions.
So what makes this variable more important in this aspect? Or are you arguing for adjusting all of those factors based on girls vs boys? Then what if it's the other way around for ethnic majority vs ethnic minority?
I write this because I don't understand your thinking at all and it'd be nice to gain some understanding.
This isn't really new. Back when I did my undegrad (well before the AI days), it was pretty common to have math problem sets not be graded, especially in first-year classes.
Essays that are hand written in a classroom setting without access to computers would still work. Especially if the topic is only vaguely given beforehand so there's no chance of having AI write one and then memorizing it.
I agree with what you're saying. I would consider that a test. But, this is semantics.
I meant larger essays. Something that take weeks and multiple edits. I don't think I ever did a 180 from when I started to where I ended in an essay like this, but I definitely did some 135s.
A key aspect here that the author says: They can use AI to do the homework, but unless they know the material, that does not help them pass the exams. Something like math will likely always have a requirement to demonstrate competency in something akin to a proctored exam. The dedicated students will want to learn. The ones that are not dedicated? Well, the AI does not help them much if the exam is weighted properly toward the grade.
AI has highlighted the fact that education skips a lot of the fundamentals, often focusing on reciting things that you are supposed to memorize and not teaching how to reason for one self.
There are people reading the writing on the wall, people pretending not to see the writing on the wall, and because things are moving so fast, people who are unaware that anything has even been written on the wall.
In many ways, this feels like another great "how do we adapt to every kid having a computer at home?" moment for education. Education will adapt. It will likely just be a rocky inconsistent process.
Every morning you're given a hexagon coin. You place this coin in your mouth. To hard to chomp so you have to suck; each layer delivering bio-chemicals required for you survive.
You enter a shower like cubicle and a program executes dependent on what curriculum you chose. You blink thrice; warping you to the Information Store where you'll visit each day until you pass 100% on the exam.
After achieving the pass you're then enrolled in to a knowledge enrichment program where a data fork siphons your brain in to the veins of the machinery. Your body is then excused to the body masher and reprocessed for the next set of educates.
In my undergrad, almost all the math courses had no required HW. They'd assign it, but you wouldn't turn it in. You'd come to office hours for help on the HW, or ask in class (they'd often dedicate the first 10-15 minutes of each lecture to Q&A).
It was awesome for people like me. No time wasted on neatly presented HWs. I'd do it quickly, verify the answers, and study for the exams.
The flip side was there were many exams (you don't want your grade catered if you do poorly in one exam). A course like calculus could have 4-5 "midterms", and then the final. Often, they'd drop your lowest midterm score so you're allowed one bad day.
I’ve always thought the obsession with grading homework was weird. Homework is your chance to practice. It’s totally normal and expected that you’ll make mistakes - so why are students’ final grades being penalized because they made mistakes during practice?
Don’t get me wrong, homework should still be marked for feedback. After all, that’s how you learn what you were doing wrong and prepare for an exam. But students shouldn’t be stressing over correctness in homework. If they do well on exams and projects, that should be sufficient to get a good grade. A student bombing the homework, learning from their mistakes, and doing well on the exam is a good outcome, not one we should punish by factoring homework correctness into a final course grade.
The point of grading homework was just a way to disincentivize students from hurting themselves by not preparing. If it's not graded, then why not bomb it and then cram before the exam? Students tend to be bad at estimating how much they can get done.
This is the ostensible reason, but really it's just lazy instructors trying to ensure they don't look like they are doing a poor job preparing their students when the students fail.
Here's a question: why don't more schools offer credit for "course challenge" tests, where the student just takes the final exam? If a student has the knowledge, why not let them get credit? Is the course evaluative, or is it more about the journey.
My high school chemistry teacher only graded homework if a student received less than 90.00% on a test. I think this worked well and is an appropriate happy medium—even the best students still did the homework and wanted it to be correct.
Edit: people can be a bit hard nosed about “grading homework is dumb” but I was the kind of student where I’d hope homework scores would help raise my grade
I'm in a graduate program, as a student teacher studying to become a high school math teacher. I have a mentor in the classroom, and basically, we only grade the homework for the completion, because we assume the students are going to cheat, so it's only worth 10% of their grade. Even with AI, and just grading for completion, some are still too lazy to do the homework.
It's usually very clear who took the time to do the homework, because they either do well on the quizzes/tests or do poorly.
So, it's not impossible. Just put less weight towards the homework, and more weight towards quizzes/tests/projects/etc.
Even pre-AI, my experience as an adjunct led me to believe there was a lot of cheating on homework and assignments.
Individual programming projects for students wrecked me as a lecturer. The time investment for me for meetings, reviews, and understanding 30+ projects well enough to grade against a higher-level, abstract rubric just took too much time.
Are you happy with the MS of Ed program you are in? I sometimes think about making the same transition to high school math teacher. I find that I do very well with engaged high school students (I volunteer in various ways at my local high school) but haven't quite found my footing with students who (maybe rationally) don't engage with the material.
Why put any weight on the homework at all if those who are completing the homework are also doing well on the tests? It seems like the only people you are advantaging are those using AI (as compared to the 'lazy' students who don't turn anything at all).
It's a fair point, it's not my policy but my mentor's. When I have my own classroom, I'll have to think about how to handle/grade homework.
It's a small stick to encourage the students to act in their own interest and actually do it.
If you already assume they are going to cheat, what's the point of even grading it?! If everyone is cheating and outsourcing it to AI, why bother assigning it? Why make it count towards the grade at all? It's just pointless busy-work at that point!
Grading homework was already a stupid idea from the start, as the goal should be to ensure the pupils learn something - and learning is very much not the same as "provided proof of busy-work". If a student can get a genuine 100/100 on the test, who cares whether they are "too lazy" to do the homework - the goal of providing education has been achieved!
If everyone is cheating, then you are literally grading them on something which has zero value beyond busy-work. You might as well ask them to sit in a corner for an hour counting grains of rice, and give them a grade for it...
> If you already assume they are going to cheat, what's the point of even grading it?! If everyone is cheating and outsourcing it to AI, why bother assigning it? Why make it count towards the grade at all? It's just pointless busy-work at that point!
My undergrad (decades ago) was like this. Homework was assigned, but not collected or graded. The purpose of homework was to give you sample problems you should be able to solve, and you could go to office hours for help.
In one engineering class, the professor would assign, but collect (and grade) them very randomly. It was max 10% of the grade. The idea was to ensure you're at least attempting the HW problems and not slacking off.
It's not pointless busywork. It's important work, punishing the naive students who don't yet understand that no good deed goes unpunished, and teaching them they should be cheating, otherwise they are doormats who worked harder for no reason.
Not to mention homework remains one of the best ways to make an uneven playing field in education in general.
This seems fine, and I say this as a teacher that is struggling with what to do about AI.
The course has an in person exam, which is graded. So the real point of homework is not to get a grade, it's too actually learn something.
Perversely, by making homework graded you are probably increasing the number of students who feel they need to use AI to get good grades.
Eliminating the grades means eliminating the fear of getting a bad grade, and so allows good students to actually use the homework to learn something.
If they don't take advantage of that, well, that's on them.
> Perversely, by making homework graded you are probably increasing the number of students who feel they need to use AI to get good grades.
I don't think this is false, but it's not the first time homework hasn't been graded. Not that long ago I had professors say they made homework a minor portion (10%-ish) of the total, since not having it be graded left most students not doing it and grades clearly went down.
It's very much a situation where poor incentives lie on both extremes.
This would have helped me. I wouldn't have done the homework if it wasn't graded; if it was, I would do it and not cheat (and do better in the exam because of it). 10% seems like a sweet spot: enough to incentivize the lazy but not a big reward for cheating. You can tweak the grading further so anyone who made an effort is likely to get, say, 60%+, so the extra marks gained by cheaters are even less significant.
Heck, I can't recall ever having been graded on my homework!
In my experience the fallout of not doing the homework was that you'd look like an idiot when called on during the communal walk-through session, or get a bad grade on the exam because you had no idea what was going on. Absolute worst-case scenario, a highschool teacher would get upset and call my parents, I guess?
I was definitely graded on projects I had to do, but that was more "write a book review" assignments than "do exercise 1 through 15" homework.
This is the way forward in my opinion. Pick learning objectives that reflect a world where AI can do most route work, Measure progress against these objectives in a manner where AI use does not matter.
I'm in a course on video game design where one of the objectives is to learn the Unity IDE... which of course doesn't matter anymore. They could have picked the objective "Make a fun and engaging game" which would allow for arbitrary AI usage and require learning the rest of the course material on what makes a game fun/engaging/graphically interesting.
I’m sorry, but if you think learning to use a game engine isn’t relevant to designing and developing games, you are sorely mistaken. The existence of LLMs isn’t an excuse to not learn the fundamentals. That’s like saying learning Python or JavaScript or C++ doesn’t matter - completely absurd. At least, if you want to develop actual skill and not be reliant on LLM providers.
At the German university I studied at, long before AI, completing the homework successfully was sometimes just a prerequisite to be able to take the exam, and the exam counted for 100% of your grade. Some classes had homework but purely for your own independent learning, and there was just a weekly presentation by a TA going through it step by step.
Problem is, that can impact girls grades vs boys. Studies have shown that boys tend to do better on tests and girls do better on homework.
https://archive.is/OYPvh#selection-1525.0-1525.510
Is that a problem? I'm sure there's another 100 variables. Maybe boys do better on 2 tests wherher girls do better on 3 tests. Or on a longer test vs a shorter one. Or open book vs closed book. Multiple choice vs open questions.
So what makes this variable more important in this aspect? Or are you arguing for adjusting all of those factors based on girls vs boys? Then what if it's the other way around for ethnic majority vs ethnic minority?
I write this because I don't understand your thinking at all and it'd be nice to gain some understanding.
But other way around it's not a problem?
This isn't really new. Back when I did my undegrad (well before the AI days), it was pretty common to have math problem sets not be graded, especially in first-year classes.
Tests and quizzes are the best way. Essays were good, but now are dead.
Essays that are hand written in a classroom setting without access to computers would still work. Especially if the topic is only vaguely given beforehand so there's no chance of having AI write one and then memorizing it.
I agree with what you're saying. I would consider that a test. But, this is semantics.
I meant larger essays. Something that take weeks and multiple edits. I don't think I ever did a 180 from when I started to where I ended in an essay like this, but I definitely did some 135s.
A key aspect here that the author says: They can use AI to do the homework, but unless they know the material, that does not help them pass the exams. Something like math will likely always have a requirement to demonstrate competency in something akin to a proctored exam. The dedicated students will want to learn. The ones that are not dedicated? Well, the AI does not help them much if the exam is weighted properly toward the grade.
This article ended right as I started getting interested. What was I supposed to take away from it?
Homework should receive a nominal portion of the grade (~10%) to encourage students to do it and not be blindsided by course difficulty.
Education in general needs to be redone.
AI has highlighted the fact that education skips a lot of the fundamentals, often focusing on reciting things that you are supposed to memorize and not teaching how to reason for one self.
What does the future look like?
There are people reading the writing on the wall, people pretending not to see the writing on the wall, and because things are moving so fast, people who are unaware that anything has even been written on the wall.
In many ways, this feels like another great "how do we adapt to every kid having a computer at home?" moment for education. Education will adapt. It will likely just be a rocky inconsistent process.
Complex systems won't survive the competency crisis.
Every morning you're given a hexagon coin. You place this coin in your mouth. To hard to chomp so you have to suck; each layer delivering bio-chemicals required for you survive.
You enter a shower like cubicle and a program executes dependent on what curriculum you chose. You blink thrice; warping you to the Information Store where you'll visit each day until you pass 100% on the exam.
After achieving the pass you're then enrolled in to a knowledge enrichment program where a data fork siphons your brain in to the veins of the machinery. Your body is then excused to the body masher and reprocessed for the next set of educates.
In my undergrad, almost all the math courses had no required HW. They'd assign it, but you wouldn't turn it in. You'd come to office hours for help on the HW, or ask in class (they'd often dedicate the first 10-15 minutes of each lecture to Q&A).
It was awesome for people like me. No time wasted on neatly presented HWs. I'd do it quickly, verify the answers, and study for the exams.
The flip side was there were many exams (you don't want your grade catered if you do poorly in one exam). A course like calculus could have 4-5 "midterms", and then the final. Often, they'd drop your lowest midterm score so you're allowed one bad day.
So AI makes everything worse yet we still allow it. And academics are silent because they want funding from the perpetrators.
Paging 'recursivedoubts
https://archive.is/A5Nqp