Brian Eno has spent decades pushing the boundaries of music and technology, but when it comes to artificial intelligence, his biggest concern isn’t the tech — it’s who controls it.
I agree. I never doubted my career choices more than when LLMs became mainstream and I saw the harm they do in classrooms. People need to drop this techno-solutionism and stop pretending that certain technologies can’t be inherently harmful.
I don’t doubt it can be harmful, but I can’t help wondering if the problem in education is the LLMs themselves or the educational system being broken in general. I never had LLMs when I was in college/university to begin with, but also, the only time I can remember even being tempted in the general direction of cheating was an online statistics class (one of the few times I did fully online class) that I had trouble focusing on or understanding much at all (at the time, I didn’t understand that I probably struggled more with some classes than others because of ADHD/executive functioning focus problems). And I still didn’t ideate about actual cheating itself, I just had some test where I tried to guess answers with intuition as an experiment; unsurprisingly, that went badly and I had to study harder going forward in that class to make it up.
Maybe I was just too goody two shoes to ever consider it seriously, I don’t know. But it wasn’t really something I even considered as an option. Notably, I also genuinely enjoyed learning if it was a subject that interested me and I usually more liked classes that had projects I could do, rather than rote memorization or long research papers.
I don’t have data on it off-hand, so maybe I’m talking out of my backside, but it seems to me that if people are focused on assignments as meeting metrics and expectations rather than the learning itself, they’re more likely to look for ways to game the system, whether for an edge, to get approval, to avoid rejection, etc. So although I can easily believe AI is making it worse in the short-term, I have to wonder why people would go for it in the first place and what can be done at the root at cheating motivations.
My main gripe isn’t so much with cheating, but with it being so easy to avoid learning anything fundamental with LLMs and then getting stuck when things get more advanced. This is very common with programming, where intro level students are now able to pass easily by relying on tools like co-pilot, but get absolutely destroyed once they reach more advanced courses.
With LLMs kids don’t feel like learning, studying or developing critical thinking skills in fundamentals classes because they are constantly spoonfed ready answers, and so they are woefully unprepared later on. As with most “successful” inventions in the smartphone age, it turns humans into passive observers and consumers rather than engaged actors with skills for investigation. I am genuinely worried for what sort of professionals and scientists we are forming today.
Oh I see, I think I read cheating with the other poster and went off of that. That is a fair point and I dislike how LLMs are getting pushed as solution rather than tool. To make a rough comparison, a hammer is a tool and makes some tasks easier / more doable, but you still have to physically use the hammer with human dexterity to pound the nail in. Whereas with LLMs, you can ask it for answers or to write things for you and it will, even if nonsense; and there’s a big problem within that of “not knowing what you don’t know” with LLMs that if you have the skills/knowledge to know it’s feeding you BS, you probably don’t need it, but if you don’t, you’re more apt to think you do need it… but also lack the skills/knowledge to debug / fact check / etc. what it’s giving you. So the people who would get the most immediate use out of it are also putting a lot of trust in something that is nowhere near a reliable tutor or subject matter expert.
I don’t differentiate cheating from avoiding learning. Students get caught and fess up all the time, so it’s possibly even more pervasive than I believe.
I teach things like global political economy and ethnic studies. If a sizable portion of students are using generative AI to fudge their way through these topics, then we couldn’t be more fucked. We can only get more fascist from here.
And AI just compounds on other issues, like our political climate where people basically piss their pants because heaven forbid someone ask you to read 30 pages about enslavement. Not only do they not want to read but they are fairly often very racist and anti-intellectual. How do you address that when everything you can ask them to do to improve their engagement can be fudged?
100% agree that education focuses way too much on test scores. I went through alot of college realizing that “I dont have to actually learn this stuff, I just have to know what questions will be asked on the test and have good answers for those”. When you approach it that way as a student, you dont actually go through the process of really learning the material and cultivating your mind. After spending alot of time out of college and reading alot more, I realize that true education is having the space to acquire knowledge and wrestle around with it until you really understand it, and socializing with others about the material. You kind of do this when studying for tests, which is why I guess tests have been around for so long. In my time though there was the internet, which was just kind of like an extended library that was easier to search, and I realize that it saved me the hassle of needing to socialize with people about the material, which I think is pretty important when learning. LLM’s are probably going to result in worse outcomes, though maybe test scores will remain the same.
I had a friend in college who I thought was very intelligent, because he would basically take every assignment or exam and distill it down to what needed to be done to fit in with the grading system. I realize now just how harmful that approach is to truly educating yourself, yet it seemed right at the time because that is how the education system is designed (and it relates to capitalism because people want to use education to get a job and to get a good job you need good grades etc.)
It’s also pretty telling whenever I ask someone if they ever studied for a test and forgot all the material on the subject right after. That is contradictory to the whole purpose of education.
It’s also pretty telling whenever I ask someone if they ever studied for a test and forgot all the material on the subject right after. That is contradictory to the whole purpose of education.
Yeah. I immediately think of gamified language learning apps here, cause it’s something I’ve been into in recent years. I get what I can out of them, but a lot of them have the same formula of basic answer correct/incorrect and either keep trying until they’re all correct to complete the lesson, or get high enough % correct to complete the lesson and move on. Whether you are actually comprehending it and integrating it as knowledge into an understanding of the particular language is a whole other question. And personally, I often feel like I’m understanding just enough to get by. I don’t feel like those apps tend to spend even close to long enough on any given concept and generally cram too much into one lesson. But also, the impersonal nature of it (not unlike the impersonal nature of large classes where the teacher doesn’t have much time for personal attention, but even worse than that, cause there’s zero personal attention) means there’s no way to “check in” and see how a person understands what they’re learning. And without that, you’re just sort of hoping that they’re getting something from it.
I remember some time back, someone I think in early childhood education talking about methods they used for understanding what the child was learning, so they could adjust if need be. I don’t remember the details now, but that kind of thing seems very important for education at all ages. And it’s something more organically present in tutoring from a human teacher, but doesn’t automatically come with the mass education setups.
In my experience this was more common 10+ years ago than it is today. Memorizing things temporarily can only get you so far. The problem is that grade inflation means you need to make an A or A+ or you just don’t get it at all. Getting a B means you are very mediocre and getting a C means you need to retake the course or explore a different field imo. If you are only really able to articulate that you understand 80% then you only know enough to be wrong in new creative ways.
I’m worried enough about good students falling into a state department trap, but these mediocre students will destroy us all one day.
I agree. I never doubted my career choices more than when LLMs became mainstream and I saw the harm they do in classrooms. People need to drop this techno-solutionism and stop pretending that certain technologies can’t be inherently harmful.
I don’t doubt it can be harmful, but I can’t help wondering if the problem in education is the LLMs themselves or the educational system being broken in general. I never had LLMs when I was in college/university to begin with, but also, the only time I can remember even being tempted in the general direction of cheating was an online statistics class (one of the few times I did fully online class) that I had trouble focusing on or understanding much at all (at the time, I didn’t understand that I probably struggled more with some classes than others because of ADHD/executive functioning focus problems). And I still didn’t ideate about actual cheating itself, I just had some test where I tried to guess answers with intuition as an experiment; unsurprisingly, that went badly and I had to study harder going forward in that class to make it up.
Maybe I was just too goody two shoes to ever consider it seriously, I don’t know. But it wasn’t really something I even considered as an option. Notably, I also genuinely enjoyed learning if it was a subject that interested me and I usually more liked classes that had projects I could do, rather than rote memorization or long research papers.
I don’t have data on it off-hand, so maybe I’m talking out of my backside, but it seems to me that if people are focused on assignments as meeting metrics and expectations rather than the learning itself, they’re more likely to look for ways to game the system, whether for an edge, to get approval, to avoid rejection, etc. So although I can easily believe AI is making it worse in the short-term, I have to wonder why people would go for it in the first place and what can be done at the root at cheating motivations.
My main gripe isn’t so much with cheating, but with it being so easy to avoid learning anything fundamental with LLMs and then getting stuck when things get more advanced. This is very common with programming, where intro level students are now able to pass easily by relying on tools like co-pilot, but get absolutely destroyed once they reach more advanced courses.
With LLMs kids don’t feel like learning, studying or developing critical thinking skills in fundamentals classes because they are constantly spoonfed ready answers, and so they are woefully unprepared later on. As with most “successful” inventions in the smartphone age, it turns humans into passive observers and consumers rather than engaged actors with skills for investigation. I am genuinely worried for what sort of professionals and scientists we are forming today.
Oh I see, I think I read cheating with the other poster and went off of that. That is a fair point and I dislike how LLMs are getting pushed as solution rather than tool. To make a rough comparison, a hammer is a tool and makes some tasks easier / more doable, but you still have to physically use the hammer with human dexterity to pound the nail in. Whereas with LLMs, you can ask it for answers or to write things for you and it will, even if nonsense; and there’s a big problem within that of “not knowing what you don’t know” with LLMs that if you have the skills/knowledge to know it’s feeding you BS, you probably don’t need it, but if you don’t, you’re more apt to think you do need it… but also lack the skills/knowledge to debug / fact check / etc. what it’s giving you. So the people who would get the most immediate use out of it are also putting a lot of trust in something that is nowhere near a reliable tutor or subject matter expert.
I don’t differentiate cheating from avoiding learning. Students get caught and fess up all the time, so it’s possibly even more pervasive than I believe.
I teach things like global political economy and ethnic studies. If a sizable portion of students are using generative AI to fudge their way through these topics, then we couldn’t be more fucked. We can only get more fascist from here.
And AI just compounds on other issues, like our political climate where people basically piss their pants because heaven forbid someone ask you to read 30 pages about enslavement. Not only do they not want to read but they are fairly often very racist and anti-intellectual. How do you address that when everything you can ask them to do to improve their engagement can be fudged?
100% agree that education focuses way too much on test scores. I went through alot of college realizing that “I dont have to actually learn this stuff, I just have to know what questions will be asked on the test and have good answers for those”. When you approach it that way as a student, you dont actually go through the process of really learning the material and cultivating your mind. After spending alot of time out of college and reading alot more, I realize that true education is having the space to acquire knowledge and wrestle around with it until you really understand it, and socializing with others about the material. You kind of do this when studying for tests, which is why I guess tests have been around for so long. In my time though there was the internet, which was just kind of like an extended library that was easier to search, and I realize that it saved me the hassle of needing to socialize with people about the material, which I think is pretty important when learning. LLM’s are probably going to result in worse outcomes, though maybe test scores will remain the same.
I had a friend in college who I thought was very intelligent, because he would basically take every assignment or exam and distill it down to what needed to be done to fit in with the grading system. I realize now just how harmful that approach is to truly educating yourself, yet it seemed right at the time because that is how the education system is designed (and it relates to capitalism because people want to use education to get a job and to get a good job you need good grades etc.)
It’s also pretty telling whenever I ask someone if they ever studied for a test and forgot all the material on the subject right after. That is contradictory to the whole purpose of education.
Yeah. I immediately think of gamified language learning apps here, cause it’s something I’ve been into in recent years. I get what I can out of them, but a lot of them have the same formula of basic answer correct/incorrect and either keep trying until they’re all correct to complete the lesson, or get high enough % correct to complete the lesson and move on. Whether you are actually comprehending it and integrating it as knowledge into an understanding of the particular language is a whole other question. And personally, I often feel like I’m understanding just enough to get by. I don’t feel like those apps tend to spend even close to long enough on any given concept and generally cram too much into one lesson. But also, the impersonal nature of it (not unlike the impersonal nature of large classes where the teacher doesn’t have much time for personal attention, but even worse than that, cause there’s zero personal attention) means there’s no way to “check in” and see how a person understands what they’re learning. And without that, you’re just sort of hoping that they’re getting something from it.
I remember some time back, someone I think in early childhood education talking about methods they used for understanding what the child was learning, so they could adjust if need be. I don’t remember the details now, but that kind of thing seems very important for education at all ages. And it’s something more organically present in tutoring from a human teacher, but doesn’t automatically come with the mass education setups.
In my experience this was more common 10+ years ago than it is today. Memorizing things temporarily can only get you so far. The problem is that grade inflation means you need to make an A or A+ or you just don’t get it at all. Getting a B means you are very mediocre and getting a C means you need to retake the course or explore a different field imo. If you are only really able to articulate that you understand 80% then you only know enough to be wrong in new creative ways.
I’m worried enough about good students falling into a state department trap, but these mediocre students will destroy us all one day.