AI boosted homework scores, then exam scores dropped: study
economist.com
economist.com
I believe AI is basically an amplifier of bad and good. I’m cynical about the world and assume it will be used more for bad than good, but I don’t doubt some of the best people in every field will be using AI to amplify their work in good ways.
Edit: Why don't they have lower-emission power sources?
It's good to think about second order impacts but this is not useful.
Just one 20 GW AI data center which is being built with gas powered turbine generator power plant, will be the largest fossil fuel power plant in the world... How exactly is that the same as one coal mine
Raising the noise floor like this only makes it that much harder to find "Smart" people, which we were already doing terrible at.
I use Claude every single day, but this is such a bad tradeoff. Maybe it will help me standup a quick fix when that is needed. Maybe it can help me dig through documentation to find relevant bits and figure out the unstated assumptions underlying it. Maybe it helps me generate test cases.
Meanwhile, my day to day life is now noise. All social media is noise. All content is noise. Slop pours onto me from all directions. Writing more test cases isn't helping me.
Am I smart? Am I dumb? I don't care, right now I'm deafened
I'm using Claude at work myself and am impressed with the product, but notice that this is the only reason I need to use it at all. Our product pages were shit to begin with, now they're AI-generated and somehow even worse. Our procedures are incomprehensible spaghetti with enough arbitrary context switching to give a sadistic Soviet municipal administrator an erection at the thought of watching anyone try to actually follow them.
Use AI to create inefficiencies, then use AI to bypass them. Those who can't do the latter will struggle to survive.
Clarification: to value “smart” people, which we were already doing terrible at.
It does give us a new heuristic, though: people who are willing to completely cut generative AI out of their lives (cold-turkey, if you ever started using it) are a much smaller group of, predominantly thoughtful, people. You do have to give up Claude to be part of this group, but from what you say, that's no great loss, and no longer being deafened is worth it.
This has considerable advantages over conventional elitism, because the barrier-to-entry is negative in almost all cases.
The one exception I've found is assistive tech, where the state-of-the-art is so poor that vibecoded slop is genuinely an improvement over the state-of-the-art, and in many cases the tooling simply isn't available to make your own assistive tech (unless you want to bootstrap an entire networked computing environment, which isn't very helpful when you want to do your online banking and do not, in fact, work at your bank).
But there are not many principled exceptions where you could seriously argue that the trade-off is worth it. Take mathematics, for example, which we often see touted as a "good use-case" of generative AI. The primary advantage of generative AI in mathematics is being able to search though a vast corpus of ivory towers and inconsistent terminology (without proper attribution) to locate and connect ideas that can help solve problems. The deficiency this is addressing is elitism, inadequate communication, and inadequate indexing within academic mathematics. This problem is entirely created by the academic mathematicians, and has been known for nearly a century (per https://en.wikipedia.org/w/index.php?title=Nicolas_Bourbaki&...):
> Bourbaki was founded in response to the effects of the First World War which caused the death of a generation of French mathematicians; as a result, young university instructors were forced to use dated texts. While teaching at the University of Strasbourg, Henri Cartan complained to his colleague André Weil of the inadequacy of available course material, which prompted Weil to propose a meeting with others in Paris to collectively write a modern analysis textbook.
To my knowledge, this is the only organised project to clean up and improve mathematical communication. Everything else (Metamath, Mizar, AFP, Lean) is yet another ivory tower. The Wikipedia article on this topic (https://en.wikipedia.org/wiki/Mathematical_knowledge_managem...) risks deletion as non-notable, that's how little anyone's actually trying. They made their own bed, and generative AI will only provide a brief respite from having to lie in it. (I was surprised how many other "compelling" use-cases evaporated when I applied this razor to them: the sibling comment https://news.ycombinator.com/item?id=49392265 points out one such.)
Vibe-coding assistive tech which doesn't yet exist, as a temporary scaffold to improve the quality-of-life of yourself and others in a social world dominated by non-essential access barriers is, to my knowledge, the only exception to this principle that can be justified. If you treat people who make other excuses, or who don't even bother with excuses, as not worth listening to, you lose little – and doubly-so, if you make your stance clear, so that others know the "cost" of gaining your attention.
... including ultimately stuff needed to make the smart people.
LLMs make significant mistakes frequently and smart people have no way of judging those mistakes outside their domain expertise. They are also sycophantic and great at being an echo chamber which makes people feel smart even if they are not.
So I think the burden of proof is on you to prove that they somehow amplify intelligence, it seems highly unlikely.
Smart people know LLMs confabulate and tell them they’re Absolutely Right! Smart people don’t want to be embarrassed by trusting the hallucination machine and revealing their gullibility to others.
All of those sound like flaws and defects of dumb people?
Great summary of the flaws with LLM "research"/"reasoning". It's always trying to con you, and I question the literacy and intelligence of the people who can't see this.
In software particularly, what AI does do is give me back my time from the drudge work that I don't care about. Keeping my build system configured and my tests up to date and my documentation synchronized is a good use of AI because I mostly don't care about how crummy the result is as long as it "works".
This is "AI"s forte. Producing more bad work at lower cost.
Suppose it's good to learn how elastic the brain is, in both directions, at a young age where it doesn't matter.
Doesn't it matter the most at a young age?
The "slightly higher" performance is based on statistically insignificant samples (between 4 and 20 students, depending on the context, out of the total population of 26,000): https://bsky.app/profile/benjaminjriley.bsky.social/post/3mt...
Citation needed? I have no clue where you got this from. I hadn't even heard of it as a conjecture, let alone as something anyone accepted, let alone as generally accepted...
1. They don't do any homework.
2. All the in-class time is split between the teacher babysitting and playing social worker to problem students, and lecturing, with little to no opportunity to actually practice what they've learned?
I understand that some students don't have home environments that are conductive to doing homework well. I understand that some students are enrolled in five hours a day of extracurricular university-application-padding activities. I understand that some students have incredibly poor screen discipline and impulse control.
But I don't understand that anyone has magically figured out how to teach complicated things to students, and have it stick without them spending a lot of time practicing what they are learning.
As anyone who has tried to do something hard knows, the first step to being good at something is to spend a lot of time being pretty shit at it.
A student who has written and received feedback on 500,000 written words is going to be way better at writing than that same student who wrote 50,000, just like someone who has put 5,000 hours of focused practice into playing the piano is going to be better than my dumb ass, who has only put 100 hours in.
(If you found the solution to get good at stuff without practicing it, I'd love to get good at piano without putting any homework in on it.)
I'm pretty sure there is a way to fine-tune this [0] to auto-complete away any of your hesitations or mistakes :)
Art is similar. You can look up skills as necessary to try to make the piece you want. Adult art learners often just start with the type of art they want to make.
In both cases, the finished piece might not be the quality you want, but in neither case are you necessarily doing repetitive stuff to learn (playing scales over and over or sketching the same bits over and over). You can retry a piece or move on - the skills will still carry over.
None of these reflect homework in school. Now, I know I graduated school decades ago, but math homework was repetitive with seemingly no real-world application and absolutely no help if I needed it. I took a math course online some years later and it was much better: Instant feedback if I got the problem wrong and instant help to walk through the problem if I needed it - then a different but similar problem was given for the homework. This was actual practice in ways traditional homework wasn't. Homework in the traditional sense isn't practice - they are all miniature tests that affect your grade.
I didn't have to practice writing - I just did the papers assigned and rushed through them. You can grade papers for other subjects on prose and grammar instead of just doing it for language courses. You don't have to read classics to read better if you just read a variety of things you are interested in. You'll read plenty of boring things for other subjects and get that sort of practice.
With most of this stuff, having actual homework isn't necessary as long as students are given time for practice. Practice is what makes you good at something, not homework, and practice can take many forms.
You can absolutely learn to play some piano by just noodling. Your learning will be much slower, and you'll miss important fundamentals that are present in any piano method.
That's perfectly fine for learning things that you do for enjoyment, or things that don't really matter. It's a lot less fine for learning things that... do.
What I remember being questioned is does it make sense to do those exercises as homework or would they be better in school.
Or on the flip side should school get out early like 11 or noon, like I think the german gymnasium does and have all the exercises as homework.
The US system where children get out of school at 15:30 and still have a bunch of homework seems a little lopsided someplace.
The 3 month break in-between school years is definitely questionable.
I think the answer is a complex one because it intersects with personality and neurodivergence.
Depending on who you are and your family situation any form of homework can be a real challenge. Not because of what you are studying but because of how difficult it is to sit down and do anything you aren't passionate about. Certainly anyone with an executive function disability will have that challenge.
Industrialized mass education has always suffered from a unit economics problem: the labor required to assign individually-tailored problem sets and manually grade them in the volume needed for most students to actually learn the material is prohibitively expensive.
"At the other end of the distribution, AI students who spend more than 65 minutes on their homework receive homework and exam scores similar to those of non-AI students, suggesting that these students do not use generative AI for homework assignments. However, this group consists entirely of students who adopted generative AI no more than Öve months. Six months after adoption, no AI student spends more than 65 minutes completing their homework (see Figure A5). This is consistent with the gradual process of learning how to use AI tools. It also suggests that AI crowds out the highest level of e§ort."
"Interestingly, in the range of 50-65 minutes, the median and the interquartile range of exam scores of AI and non-AI students are similar. This implies that, in the range where AI students and non-AI students have overlapping homework times, students who spend the same amount of time completing homework on average receive similar exam scores."
"This pattern shows that students who spend the same amount of time on homework learn similarly, with or without generative AI. In other words, generative AI reduces time spent learning for the majority of AI students but not learning efficiency for those who spend the same time studying as the non-AI students."
they're not designed to measure general aptitude, or function as admissions criteria, or screen for job applications, or any other numerous things they are used for.
there can be many questions of pedagogy. one of them is, what do our exams measure and how do we use them? professors who say, "My exam is designed to measure who studies, not be used for all these other purposes that they are actually used for" - I don't buy it. It's the same as late night comedians saying they are not responsible for solutions, even when spending 90% of their air time making political jokes.
THIS is the pedagogical issue, that pedagogy has NEVER caught up with the scope of responsibilities. This is acute in STEM - I mean, the humanities departments are generally pretty well run, all things considered, in this regard. Generative AI is accelerating that pre-existing crisis.
Huh? They're designed to measure how much you know. They can't see how much you study, nor would they have reason to be interested.
At the end of the day though what matters is what you know. Furthermore, if it's a serious subject, it shouldn't matter whether you learned it from this teacher or from another school and teacher, as long as your knowledge is correct. Knowing the idiosyncracies of this particular teacher should not factor into the grade. A serious subject can be learned on one continent and examined on another. Bullshit courses are all about learning pet peeves and hobby horses of a particular teacher.
let's imagine a different study. we instead compare AI-users and non-users on a Wechsler (IQ-adjacent) test.
overall, it would be surprising if AI usage impacted your Wechsler scores. someone has done this study and the impact is quite quite small. BUT. do we care? We don't use Wechsler scores for admissions, we don't use them for jobs, we don't use them for... are you getting it now? A Wechsler family test is measuring something real, just like a university exam measures something. But what do we USE them for? Wechsler and a typical university exam are, in some senses, EQUALLY vague in terms of their fitness for purpose for answering a question like, "should we hire this guy?"
Like there is an association between IQ and earnings but it is actually surprisingly small! There is an association with math education and earnings and it is also surprisingly small. And consider how many people get by just fine without using a single piece of math education once they have finished school - like what if maximizing your earnings isn't all that it is about? Are you getting it now?
The issue isn't the AI usage. I can find tests that are immune to AI usage. The issue is using tests for things that they are not designed for. We pick and choose, for some subtle but nonetheless pervasive cultural reasons, which tests we use for which purpose, and very frequently, not because they are calibrated for the chosen purpose. This is coming from someone who scores very well on all these tests, and have kids, so I have a very strong incentive to buy into the status quo, and I'm telling you: academic testing has been fucked up for a long, long time.
Two weeks of ADHD-fueled research later, I concluded that academia is actively resistant to implementing assessment reform because it would expose the utter pointlessness of most of what happens in university classrooms.
The reality is that we have no idea what most university exams measure because they are ad hoc, written by amateurs (yes, most professors are untrained in pedagogical methods) with zero psychometric validity analysis.
This is a reasonable opinion after two semesters.
IMO, with more experience you should have modified your stance.
If all classes were pointless, students would leave at the same point they entered. Since we observe that is not the case, something is happening in class. Do more of that.
The explicitly stated goal of most courses is mastery of a well-defined set of domain-specific knowledge and skills. The only way to objectively determine if the course achieved that goal for a given student is comprehensive assessment. Bad assessment design, much like bad experiment design in science, is worse than useless—it actively misleads and confuses.
A well-designed, repeatable, reliable assessment is a prerequisite before one can even consider the effects of particular pedagogical methods inside or outside of the classroom itself.
"The negative learning effects are larger for students with higher initial achievement. The differences in the estimated full (6-10 month average) effects are substantial, with a 50% gap between the most negative effect (-24 percent) for the highest tercile and the least negative (-16 percent) for the lowest tercile. "
Not top 10% as you asked, but the closest to what you asked. My working hypothesis is that top performance is highly correlated with willingness to work hard, and AI decreases the motivation to work hard.
I think if you take a physics class where the student is intelligent and intrinsically motivated through their own interest (I admit this is rare) then AI probably helps.
That last part is key. Intrinsic motivation doesn't mean you pursue it outside normal bounds. My kid loves soccer, its her second favorite thing in the world, she has an absurdly high tolerance for physical discomfort while playing, but she doesn't play it at home. There's other things she's rather do, such as play with her toys.
When you move the bar to something even less interesting to most kids like science, you're going to have a pretty huge falloff. You're basically selecting for kids who choose to do it in their spare time. I know a lot of smart kids (I run a boyscout troop, my wife a girlscout troop, both with lots of high achievers), and none of them do this.
I'm confident it's an amplifier for people who know how learning works and already do a lot of it, successfully. However the level of "learning fluency" I'm talking about isn't reached for many until late college or grad school, and sometimes not at all. So I'm not surprised by the quoted results for 12-18 year olds.
The discussion is about "AI", so common sense is out the window. These people's professional reputations depend on addict-level "AI" usage remaining socially acceptable.
How would you prove/disprove this assumption without falling into a True Scotsman fallacy?
It would not prove that "the results depend on the operator". To prove this (and figure out the traits that make a proficient operator) you would either need a massive dataset to which you'd apply some machine learning to figure out the correlations, or at least you'd need a hypothesis on the traits you want to test.
Do left-handed people perform better with LLMs? Analytical thinkers? Dyslexic people?
It's not enough to just claim "some individuals will perform better with these tools" without any notion of who does, and whether it's possible to become one of these people, and what is the expected gain in this group.
Otherwise, sure, we could sell unsecured chainsaws as a tool for making ice sculptures and observe that "some people" are indeed more proficient at not getting their face cut off ; but if the tool is being marketed to schools, such a vague statement is not helpful in making a case for it.
My only point was I think it's possible to do. I personally think it's pretty clear. And there's nothing special about AI. In rural places, you could replace it with "access to books" or "access to a good tutor" and so on.
I also think you'd see the same effect with "motivation" (if you could test it), with performance on standardized tests, and with grades. AI will boost the better (by these metrics) students more.
It seems that the students who had the most skill and motivation from the start, have the most to lose.
You might say "it's good to learn research skills" and that's true to an extent, but tutors have always made people better students. And AI is a tutor you can message at any time, day or night, for free.
The good thing about a tutor is that he or she isn't always available and knows they won't be in the future, so they instill good habits and independence in you.
Hence the whole "AI is basically an amplifier of bad and good" argument made earlier. The ones who do this because they want to understand, don't need to cultivate this at all, it naturally happens with chatbots. They don't just ask for the answer to a question, but then dig into why it's like that and what not. But the ones that don't care, now have to do even less to get the fast answer without understanding.
I wasted several days trying to have AI teach me containers. I would have been much better off just reading the docs.
I wouldn’t be surprised if you did and still had the same issue, but if you relied on the training data recall alone, you definitely shortchanged yourself.
I’ve had decent results with tasking a model to run pre-research and summarization as a checklist, then have a second instance of the model(s) review and then develop my learning plan learn, versus the times I just asked Claude or ChatGPT to explain something to me “from memory.”
which makes it not a tutor. if it is true that tutors have always made people better students, then those tutors are definitely imposing some limits on how many answers they give you and requiring you to do some thinking. I think you could have premised your same argument by, "copying a smart kid's answers has always made people better students...."
First, there's a certain amount of baseline knowledge we'd like students to possess. Without a certain prerequisite amount of underlying information committed to memory, it gets far more difficult to achieve fluency in a topic.
But more than that, there's other skills we're trying to build: frustration tolerance, processing contradictory information, disciplined problem solving. You only really develop these skills through productive struggle. If you find a way to shortcut the productive struggle, students truggle.
> but tutors have always made people better students.
Sure. Bloom showed us that students taught with a combination of tutorial and mastery methods, one-on-one, outperform students in a normal classroom by roughly 2 sigma.
The paradox has always been-- why hasn't technology unlocked these gains for students in normal classrooms? If we could boost everyone's performance by this amount, it would be huge for society-- but society can't afford to teach everyone with tutorial methods.
Since the 1970s, we've invested in edtech towards trying to make this happen, but most of it has actually had net-negative effects as best as we can measure. AI, so far, looks to be much worse.
I think part of the answer is that a big part of what makes a conventional classroom work are social pressures. So far, it looks like AI (and edtech in general) does more to dismantle conventional pedagogy and to break down the social fabric of the classroom, than it has improved differentiation or unlocked this tutorial effect more broadly.
I was assuming this was a typo and was thinking about making a joke about it, but it does appear to be slang that fits the context:
https://www.urbandictionary.com/define.php?term=Truggle
> 1. the standard of perpetual intellectual failure made by an individual.
> Truggle; the standard defenintion of a person who is a failureat everything.
Was it a typo or did you actually mean this?
The answer is obvious: you're doing homework type activity (learning is not a goal), not exam type activity (learning is the goal).
Social stigma against just juices people to stick to socially desirable answers; no doubt a whole bunch who self reported as non-AI users actually used AI
"e§ort" == effort
"Öve" == not sure. Maybe "over five"?
We need to shift the incentives by adding ruinous penalties for things that are currently quite commonplace if they are done by large players. Some dude training his own AI on his own computer can scrape and train. The fine for OpenAI or Meta using a single copyrighted book without permission should be in the tens or hundreds of millions.
What we're seeing currently in our society is a "loophole inversion" where the rules have an effect mainly via their loopholes. The most profitable activity is to find loopholes and exploit them as frenetically as possible to gain as much advantage as you can before the loophole is closed, or get people hooked on the loophole so it's retroactively legalized. Entities that are big enough to do this are big enough because they have lots of money behind them. Entities doing the same kinds of things without lots of money are not really doing much harm. So the best approach is to adopt a "sliding scale" in which even tiny violations by wealthy actors result in penalties enormously greater than fairly large violations by small players.
I think we don't yet know the long term view - what is harmful or damaging in the long run. Too early to tell and pass judgements. Every new technology from writing to internet had its detractors and they all pointed out negative externalities, as they seemed to appear at the moment.
Same can be said of technology in general tbh.
We could probably cook up thousands of examples of the same problem (YouTube DIY tutorials, GPS navigation, etc.)
It’s alarming how near universally AI has been a cataclysm in class rooms.
On homework the short answer is is that we give kids home work exercises.
AI helps learners reduce the effort expended to exercise and get results. This is making homework moot.
The data point around 80 minutes seems like noise to me. Looks like there isn't enough data/students who spend that much time and also used AI.
It would be nice if AI was a force for good as well as bad, but the data here doesn't support it
A huge proportion of people who use AI use it to outsource their thinking.
It can be used for good for this purpose. You can use it to find primary source information. You can use it to generate challenges for you. You can use it to critique your work when there’s nobody else available. Etc etc.
Evidently the AI users think hard things are beneath them. They want to win without doing any work. Sympathy, I have not.
If it is an amplifier of "good" vs "bad", I don't know, and I do not know what that means exactly. I do not think some students are just "bad" or "lazy" by character and that explains all. I think that their environment has a big role in shaping their actions. I don't think that AI cannot be used beneficially (and the articles talks about some attempts that maybe worked well), but it requires careful considerations, guidance and intentionality from educational institutions themselves. The same goes with all kinds of technology, and AI is nothing new in that sense, but often new technologies are introduced in education without careful considerations or evidence based policies. Which is why imo articles like this are very important.
We need to restructure the system to treat failure as a signal instead of a disaster. Grades should come from hard randomized exams with unlimited retakes so one bad day won't hurt you. Homework should be optional material for self study, evaluated by teachers if you choose to do it but never forced.
Hold back students for individual classes instead of a whole grade so failing one can't ruin your social life and teachers are more willing to do it. F students will realize they have to study, start actually learning and then pass on the second time. No big deal. It happened to my friends in college, no reason they can't do it in high schools.
Discipline is a skill and it's one you have to get from experience. If you try and force kids to study when they don't want to "for their own good" you're not actually helping them. Everyone needs to find their own path. Let people fail.
this rhetoric is pretending to be an alternative to coercion. IMO the ideas you are talking about are well trodden and are still coercion nonetheless.
Do you have evidence for this beyond your friends (who were accepted into college)?
At Caltech, homework was assigned but had no bearing on your grade. The grades were based on the midterm and final exams.
But not mastering the homework usually resulted in flunking the exams. There were "retch" sessions after each homework assignment that was staffed by a grad student, and the purpose was to help the students understand the homework problems. I knew only one person (Hal Finney) who was so smart he didn't need to do the homework.
I learned the hard way that the path to success was:
1. never miss a lecture, no matter what
2. take notes by hand during lecture
3. do the homework on time, and make sure you understand every problem. Take advantage of the retch sessions.
And that worked for me.
The general pattern at Caltech was 2 hours of study for every hour of lecture. Which was quite a shock to me.
I found that (1) I didn't need to take anywhere near as many notes during lecture and (2) I could ask way, way, way more relevant questions.
This also helped tremendously when it came to studying for the actuarial exams.
As a student of a top-tier French Master's degree, I consulted with a teacher to deal with exhaustion and to ask for a class rescheduling for my case. Explaining my situation, the teacher looks at me and interjected:
—Waitwaitwait. You... you went to all lectures!?
And he was right. Rather than following the curriculum, I should have developed my taste for various engineering topics and only used the classes as entertainment.
When I was there, a long time ago, exams were timed and were usually open book open note. Blue books were filled in. You were trusted to adhere by those rules, and most students did their exams in their dorm rooms.
The evidence that the students honored the rules was some exams resulted in a 50% failure rate.
As for me, I went there because I wanted to learn the material. I did not care about getting a diploma. (Mine is in the basement somewhere.) I did not take any "easy A" classes, because I wanted a return on my time and tuition investment. (Though, easy A classes were hard to find at Caltech.) I wasn't even going to attend graduation, but my parents showed up and I attended to please them.
The classes, year by year, were dependent on mastering the previous year's classes. So if you cheat with AI, you're digging yourself into a bigger and bigger hole. Caltech rewires your brain. If you don't learn the stuff, you're going to be one of those EEs who carries around a card with V=A*R, V/A=R, V/R=A printed on it.
So the endgame was figuring out what the tests in previous years looked like (cause it was likely gonna be a copy paste affair), do a targeted study run for those exercises and 9/10 you would pass.
We didn't call them retch sessions, and they were taught by the instructors though. We also were encouraged to peer tutor and since we were all restricted to one building the homework was always group work allowed.
Also had badges to track time spent in the building for required study hours, though some people gave up and just slept at their desks when they started sliding down the grade scale and the hours racked up.
Generally I think I did 30 hours of studying/homework (went up and down depending on what was being taught, but was around that) a week (for 12-15 hours of actual lecturing), with some of my friends putting in 50% more. Generally the only day we weren't there was Saturdays. Most of the day Sunday was usually spent in class preparing for the next week.
But you had tutorials each week, and if your tutors thought you weren't doing enough work they could set you exams mid-course called 'penal collections' and if you failed them you could be thrown out. They were rare but definitely not unknown.
Education system, contrary to popular belief/name, isn't tailored to educate but to select winners and losers which then will be picked on the job market.
That's why it seems absurd when you think about it as an institution that aims to educate. That's because that isn't the real purpose of it. The purpose is to stratify and classify early.
Schools do not operate in a vacuum; they serve as credentialing gatekeepers for a hyper-competitive capitalist job market. If everyone could easily retake exams until they got an A, grades would lose their primary utility for employers and universities: differentiation. Society relies on schools to provide a neat hierarchy of candidates that for one reason or another thrived in difficult environment of adolescent schooling.
The system often prioritizes compliance, endurance of boredom, and social maneuvering over actual critical thinking precisely because those traits align with corporate hierarchies.
Compliance and endurance of boredom I can somewhat agree with (thought not to a great extent), but social manoeuvring? As long as a student completes their classwork and homework, and does well in their exams (none of which mandate any social interaction apart from team projects), they easily pass through the school system. In fact, I used to hear about how the school system disproportionately benefited studious, shut-in students over street-wise, socially-savvy ones.
Socially savvy students on the other hand will get plenty of help from teachers and other students even if they are not so good themselves.
This is why the so called “intelligence” where I will define it as skill of acquiring skills right now, is not so important for success. You can be very good at learning but if your social aptitude and charisma sucks then you will just end up as depressed, miserable and poor and probably alone too.
If you have a kid It’s honestly very important and way more important than homeworks or exams, to care about kid’s social development and how you can help so your kid gains social confidence and never loses it at crucial moments
Optional homework is often a disaster. At best, students would do it right before an exam and the goal of education is not to just pass exams. They’d probably still get a lower score than if they did the homework when they were supposed to.
What I think is better is to have a due date, but just make the maximum 10% each day it is late. So after 2 days, the highest score you could receive would be 80%.
I liked that system because it gave some flexibility with deadlines while still encouraging you to turn things in on time.
My view is this: a student can turn in an incomplete assignment. Work they completed get evaluated and graded. Anything not completed gets zero credit. Homework should be an investment in effort over time anyway. Unfortunately, procrastination is a deeply rooted problem that usually kills this.
> Homework should be optional material for self study, evaluated by teachers if you choose to do it but never forced.
My wife is a teacher and her school does this. The result is that nobody does homework. The kids who need the extra practice don't get it and they fall into a spiral of failure and apathy. This is especially bad for subjects like math that build on themselves.
IMO the better solution is to require and grade shorter homework and provide students with optional, supplementary assignments that can be used to make up for missed credit on homework for the same material. That way students who grasp the material quickly can demonstrate and move on while others who need practice are naturally encouraged to get it without being permanently punished for struggling initially.
I much prefer the first.
School shouldn't be impossible, but you shouldn't have a charade of schooling.
Obviously, extra practice is a good idea.
Assignments students do in class isn't "homework". It's just classwork. All my math classes had both.
Perhaps you mean study halls? Those were uncommon until highschool where I grew up. And most of the time people spent there was not studying.
Exams will have to be a lot longer if you allow unlimited retakes. Generally exams work on a sample principle, but this breaks with retakes.
This simply does not work for most, as pupils are too immature. You put sunscreen on children, keep them away from the street and brush their teeth. Having everyone re-learn the same stupid lesson is wasteful, it's better to institutionalize this knowledge in the form of mandatory participation.
Study design: "David Stromberg of Stockholm University and Victor Lei and Wu Yanhui of the University of Hong Kong set out to fill the gap. They tracked 27,000 pupils aged 12-18 in China, where ai adoption has been fast. Around 80% reported using models such as Doubao and DeepSeek; the other 20% formed the control group."
As with most training the journey is the point, not the destination.
That said, I think smart use of AI could help. It could explain concepts in a way that might help you understand better, it could probe your knowledge in a more dynamic way by tailoring questions, and so on. This requires the AI be restrained by some harness, not free to write down the answers for you.
How would this magic harness look like and why would anyone use it?
Put that in your agents.md.
I meant: what's going to stop the average lazy student from ignoring whatever harness the university recommends and instead use an unrestricted LLM, thus learning nothing?
Nothing until they start failing exams and maybe seeing the error of their ways. My suggestions wasn't for you, it was for the smart student who wants to use AI to enhance their learning but not have it do all the work.
But cheating isn't new. People have been cheating in school for centuries. It's easier now, but the consequences were always the same. At some point the chickens come home to roost and you pay the piper.
My concern is that with LLMs and above average bullshitting abilities, the chickens might never come home to roost.
People say that all the time, but does anyone really think a lack of good explanations for things is a limiting factor in 2026? Or even 2010?
Kids won't be using LLM tooling as a personal tutor following some sort of Socratic method, they'll ask it to solve their home/coursework for them and blindly copy/paste the answer. Hell, they'll just manually copy down what's on their screen if copy/pasting isn't possible for whatever reason.
Obviously exceptions exist, but I'd wager from being an ex-kid myself the type of kid who would genuinely use these tools for actual proper self-tutoring would be an extreme rarity.
AI gives you a way to get the same help without asking another human. Say of that what you will, but not everyone was comfortable asking other humans for help even back then.
The act of building the mental muscle is what creates education.
Learning the material (using the text book and videos) was what they had to do at night and class time was spent working through problems or applying the material in some way.
Your answer - middle school - I find it extremely young for this, which is extra interesting.
That year I sent an email to her teacher to let them know that I'm the one responsible for her not doing all the work and if that's a problem, we need to talk. It wasn't a problem.
The one class that was flipped was a relief because she could breeze through the lesson much faster than would normally be spent on it in class and the amount of time on exercises was limited to the class time slot.
I do like it though, and I like that college generally leaned that way more (and was better balanced, as it had substantially less time in classrooms, so you could study during the day).
IMHO there's a lot to recommend this style. A lot of students learn better when they can wrestle with the material at their own pace, on their own time, in their own setting. It quells a lot of anxiety about "I'm not following what the teacher is saying, am I stuped, will I look bad in front of all my peers?" And then class time can be spent identifying holes in your knowledge and getting instant feedback from an expert, which is where they are most useful.
If you skip this, well your brain won't develop as much at period of life it's able to do so.
AI is a big change; pedagogy is going to change too.
I mean, does it? The way I see it, the best LLMs have to offer is infinite patience (until they inevitably and unpredictably start to confabulate, which should be a gigantic red flag, anyhow), I don't see how LLMs can be used to explore teaching paradigms that haven't been explored before. We know pretty well from centuries of empirical experimentation how children's brains develop under different stimuli. It's not exactly something the software industry needed to "hack".
It is really a combination: you need to memorize a large corpus of information to operationalize knowledge — understanding a bunch of theorems in mathematics does not help much (even if you are able to prove them when you see them) if you can't remember the boundary conditions they hold under.
Or having good understanding of foreign language grammar won't help you if you do not memorize words that you need to express your thoughts.
Good literature will show you some of life's challenges and potentially let you think through them in a non-stressful situation (other than "I've got to finish the last 200 pages by Monday" ;)).
At the outskirts of one's knowledge this is inevitably the case. My impression it is still useful to know that a certain implication is possible - and one can look up the exact conditions.
It's just a low pass filter for the job market. Students aren't actually interested in learning anything, they have been condition for a long time to jump through the hoops.
It would be much better if we actually returned to exam and performance based evaluation of students, and stopped giving everyone As just for trying hard. But parents look at college fees as buying their kid a job opportunity, and so grade inflation has basically ruined the SNR of academic performance for all but the lowest performers.
Using the forklift as a spotter and to assist in loading weights increased gains. Then again, a human can do all those things, and provide real human connection.
This is a pedagogical problem that AI merely exposed. Educators need to figure out How to make students choose the scenic route instead of having them optimize for the most efficient completion of a task.
All of this was at university of which 3 of them were at the same university.
Especially the artsy game design program definitely did not feel like it was preparing me to be a cog in some giant corporate wheel.
Career exams tend to be a mix but a lot less ideology.
Your career is more akin to the totality of school than it is to any specific facet of it imo.
The social aspects are more important than the exam sitting most of the time.
> The social aspects are more important than the exam sitting most of the time.
To spell it out more clearly: every single aspect of your career is an exam. The interview is an exam, quite literally. The day-to-day responsibilities (meetings, planning, problem-solving, collaborating) are parts of the exam. If you're failing at those or automating them away, then what is your role? The assignments (solve X bug or add Y feature) are parts of the exam too. Failing to do these things will mean that, yes, you are failing the exam.
And I could also call most of those things assignments instead of exams and nobody would bat an eye.
Your analogy was both thoughtless and stands up to no scrutiny
You don't need a final answer but you can't just answer "I don't know" otherwise you risk being slimed (shout out to anyone who still remembers "You Can't Do That on Television")
Is an acceptable answer.
Asking people to be conclusive on the spot is a recipe for acting on bad data. There are too many biases related to stature, they interfere with accuracy.
The answer is "accepted", in the sense that it won't get you fired. But the meeting plows on, and decisions are made in absence of the answer to the question! — and never revisited once the answer is known.
And like 90% of the time it comes up, it's because the data contradicts the decision.
If cohorts are graduating into the workforce with decreasing stand-alone skills, that implies the economic value-add of labour is shifting to AI, which implies broadly falling standards of living, falling political power, and an ongoing transfer of power and wealth to capital.
If students are taking shortcuts with AI when their brains have the highest capacity for learning they may be permanently weakening their prospects. It doesn't seem likely that whatever "AI management skills" they incidentally learn will make up for, over their lifetime, the economic value loss they suffer due to reduced cognitive skills.
On an individual basis, in the next few years, this might not matter. Across decades, economies and polities, it will.
Students have been cheating and taking shortcuts on their work long before the current generation of AI tools existed (and probably for as long as graded exams have existed).
Somehow, civilization has managed to survive.
Just to add to this, if AI is as epochal a change as its boosters say (and I am open to that argument), then surely it must have the potential for awful consequences, in proportion to its potential for good.
unless you're just blindly passing it on for someone else to deal with
Try it for yourself by making a prompt like this (adapt as needed):
Pretend I am an undergraduate student of Computer Science. I am learning about early microprocessors from the 1970s. I want you to ask me an examination question as if you were doing a viva voce exam with me, to test my understanding of concepts. I want you to receive my answer and then based on what I said I want you to ask me a more specific question to probe my understanding. Repeat this interaction up to 5 times. Then grade my understanding so far, by giving me a pass, merit, credit, or distinction. Can you explain how you arrive at the grade based on my answers and your expectation of undergraduate knowledge of microprocessor theory?
A name people use for this phenomenon now with AI is cognitive offloading.
I coincidentally took advantage of the cheap DeepSeek Flash pricing the past couple of weeks to compile & synthesize a knowledge base/wiki of hundreds of educational AI research studies like this one here: https://edtechdev.github.io/aied/ Here's the summary of this article: https://edtechdev.github.io/aied/articles/stromberg-generati...
In the case of some management, is causing them to unlearn, forgetting about proper review and maintenance practices.
Open book exams existed for a long time and even take home assignments where the net was usable to complete the tasks. Difference was one actually had to do the reseach and learn something along the way regarding research skills or the topic itself.
LLms just make the automation and laziness with a lower barrier to polished output.
Pretty sure the way social media is designed contributes more to short attention spans and atrophy of the cerebral cortex.
https://www.bbc.co.uk/future/article/20240517-the-human-brai...
Same for social media. The negative effects have been apparent for years.
I don't think we can go back to a world without books, without social media and without AI.
That is catastrophic because I heard several CEOs that said that we cannot do anything about it and that the technology is here to stay. So, I suppose that we will just need to adapt to a world with declining mental cognition for millions of children and a future with diminishing production capacity because ... checks notes ... this quarterly profits.
When I was in school there were multiple instances where I would be stuck on a problem for nearly an hr but I always learnt something from it. They key I think is to accurately identify when to and when not to use AI
When the signal is removed, this is what you get.
> AI users who maintain similar homework completion time as non-AI users experience small learning losses.
If the pure completion time of homework goes drops significantly with an LLM tool, I suspect these students are spending the extra time turning the material over in different ways to internalize, which is interesting.
What is the point of using of AI if that is not going to speed up learning.
It tries to link them together with no obvious cause and effect.
The score could have dropped due to worse environment at schools or kids who were stuck at home studying online and now in physical locations could not adapt fast enough (meaning, possibly just temporary drops). We do not yet have enough metrics gathered yet.
This type of article bashing AI for the root of any problems, I find it appalling without any evidence.
It's nothing but a conjecture.
AI shouldn't be "helping" with homework. AI should be the homework.
Start with any topic or problem, students should be encouraged interact with AI, and learn stuff using Socratic method.
Teachers should rate the chat session instead.
And iterate the process to high profeciency.
Then I stopped caring about tests too.
I did the same thing through college. The only reason they passed me and I got a degree is that I built the school’s website and I built personal ecommerce sites for the head of art and his wife to sell their paintings.
I haven’t done shit since like 7th grade.
Worked at Facebook, Apple, Microsoft, same behavior there - did basically nothing for them while making thousands off my games on App Store.
Fuck authority
True or not, it supports a narrative of "kids these days...", of a morally bankrupt educational system, of indulgent adults spoiling children, of just desserts to the wicked and lazy.
Again, I'm not saying the article is right or wrong. I'm saying I'm skeptical because it so tidily aligns with the Economist's editorial world view.
As with any skill, practice is necessary to achieve competence. If homework is no longer a viable way to force students to practice, you'll need to find some replacement, you can't just abolish it and expect there to be no negative consequences.
By what means? Have you got a magic wand that will turn 3rd graders into mature functional adults who will study on their own without homework?
> The study followed 27,000 pupils aged 12 to 18
(Third grade is 9-10 years old I believe.)
What are you basing that claim on?
And American grad school does have homework. I’m not sure why you think it does not.
This sounds really strange, never heard of.
BTW / Offtopic: I have read some weeks ago, today if you are hosting a party for your kids birthday, parents are preparing "give away bags" for handing over when leaving the party
Really well thought out goodie bags are a treat, they follow the theme of the party and they have memorable (even if low priced) gifts.
Most goodie bags are filled with disposable toys that break after a few minutes of play.
Please also abolish PowerPoint and force oral exams while you are at it.
Teachers are also luddites as a social class. Math teachers cried about calculators yet everyone of them who refused to adapt did disservice to their students (especially anyone doing statistics). My state's education system is utter trash despite being well off economically because of strong teachers unions keeping them from undergoing even a tiny bit of scrutiny.
They do not deserve the social credit/grace that they get. They are ultimately gatekeepers, whose extremely biased decisions in grading/treatment of their flocks decide which kid grows up on the streets and which kid grows up to eat lobster thermador every day.
Teachers also should not be disciplinarians. Children with discipline problems bad enough to disrupt a whole class should be thrown out of the class room as basically the only thing a teacher does for "discipline". The "right to education for the shit kids means we can't do that" framing is such trash for those who want an education and will inevitably be disrupted by a small minority of terrible students.
Also it's interesting that one of the only other things Max Stirner wrote about besides philosophy was education and his myriad problems with it (given his history as a school teacher).
https://theanarchistlibrary.org/library/max-stirner-the-fals...
I see our current approach like a driving school training its students at fast running.
Yes, using a calculator leads to worse outcomes in your learning. But also, yes, everyone always carries a calculator on their person 24/7 these days and you'll never be as fast and as accurate as a calculator.
I think it would have been kinda cool if you could have unlocked the privilege of using a calculator in school through good grades in math class. Prove you can do what the tool can, then you get to use it. Would have also provided at least one incentive for students to get good grades in at least one class.
Does it? https://www.jstor.org/stable/749255
If you don't have to solve a problem, your brain does not have that work to do.
Your brain working less means it has less opportunities to train.
The consequence is worse cognitive abilities which is tied to worse learning outcomes.
-
From your link:
> Sustained calculator use in Grade 4 appears to hinder the development of basic skills in average students.
It noted the following positive outcome:
> apparently improves the average student's basic skills with paper and pencil, both in working exercises and in problem solving
But this only measured the outcome for solving math. Yes, offloading problems will make it easier to solve problems, at the cost of not training your brain, which leads to worse learning outcomes across all subjects.
My own experience is that college is fun. You can spend the time wisely, but school stuff, is not what makes you money.
Make student loans dischargeable in bankruptcy and make colleges under write them. The problem will get solved quickly.
AI just gave the students a way to dodge the slog.
But university was never intended to teach the bleeding edge. That would really be impossible in practice. Pre-phd, it is supposed to teach ways to efficiently attack a problem. you can take a bunch of "play courses", and succeeding at any of those requires pretty much only that one skill.
Nowadays, the challenge they face, is to keep teaching "problem attack methods" in a way that can't be trivialized by AI.
Though fundamentally, if you go to university, and evade learning the one thing you can learn there - thats your loss.
GP said exactly the same thing.
I agree with GP’s broader point, which is that you cannot have understanding without a specific set of baseline facts. The student studying only those facts and never achieving the understanding they are meant to support is a tragedy of our education system.
There are people who need to understand microbiology at a specific level and they cannot derive it all on the fly. They can achieve better memorization through understanding but I personally have found rote memorization can be an essential first step in some disciplines.
"Wax on. Wax off" -- memorization and repetition are the basis for understanding.
How useful it is to remember useless information that one day turns out to not be so useless after all.
These people have fallen into OpenAI and Anthropic's trap and are illiterate without the assistance of coin-operated slop generators.
Thank god for china!
If our society is organized on just delegating all authority to "the best answer" we can drop all jobs - and only the surgeon has some authority over the machine. When we have a surgery robot, then they can go to. And at some point we'll be recommending and forcing surgery for athletes foot because some kind of best practice or viral post going out of control and a bunch of openclaw bots randomly decide to make tons of recommendations online for amputation for athletes foot and GPT 8.6 agrees this is the way to go.
OR we can somehow decide there is a boundary, there is some room for human memory and decisions.
All podiatrists are not the same. All physicians in their specialties are often specializing in their own niche and are up-to-date on that niche.
I've had lisfranc ligament injury. My podiatrist wanted to perform a surgery quickly, which in 90% of cases is advised. I decided to seek a 2nd opinion because I realized she did the X-rays wrong, as she did them non-weight bearing, which is useless if there are no fractures.
The specialist was excited to see me, and up to date on all the lisfranc research. I've had a really bad sprain and he said conservative treatment should be sufficient. It took months but I healed up.
I've had quite a few injuries from playing sports and often found doctors to be "wrong" simply because their knowledge was out of date.
We should establish a single practice of care, MedicineGPT. It may be a rough start but over time the gene pool will adapt and the human race will live harmoniously with the machines (sorry, with the GPUs). Plus our insurance premiums should finally go down.
However, if you went to an AI agent which has access to medical research, you would get the most up to date research.
Why would I seek a 2nd opinion if the initial doctor knew what to do?
Ironically I had a friend who went to med school and in his own words roughly 90% or more of the students there are heavily relying on GPT. They even had a few instances where people were using it to cheat during exams. And supposedly during the clinical years of uni if they don't know something they literally just go to a separate room and start asking GPT for answers. Make of that what you will.
Thinking is infinitely more complex than "critical thinking" vs. rote functions. You want students to get to critical, eventually, but the lower orders on bloom's taxonomy serve important roles, too.
Course design built around multiple choice questions, essays, and a lot of the historical assessments that we're used to are largely not great for learning in the first place—–they're used because they can be auto-graded or score assigned quickly. AI, conversely, is kind of revealing the limitations of these outmoded learning design practices.
Think of the meme going around right now from Milennials to Gen Z: "I wrote a 5 page essay on a book I never read!" <- They're saying this to AI generated essays as if there's any difference in the critical thinking that comes out of an exercise like that (read: absolutely none).
AI is a new tech—so of course people are still figuring out the limits of it. In 10 years a lot of the paradigm will be set. Until then, students are unfortunately kind of in the thick of it.
As somebody that read every mandatory reading book, I wish I had had the critical thinking skills to not read them. In retrospect, they were a waste of time and made me absolutely hate reading.
This is something that AI cannot replace. Viewing things as memorizing is the wrong take.
Open book tests and allowed cheat sheets already solve the memorization-is-not-required gap. It seems like allowing use of AI pushes that closer to thinking-is-not-required. And if no thinking is required, then how can you prove personal mastery?
Absolute waste of time getting babies to walk now that we have cars and electric scooters!
Yes, you do need to memorize and know things about the subject in order to be a master of it. What is the definition of "mastery" that you are using? What are the Things being memorized?
Yes you do???
You can't play an instrument without muscle memory.
Why are commercial airline pilots required to spend hundreds of hours in simulators? So that when things go wrong they don't have time to think, they just react. And their reaction will be the correct course of action because that's what they have drilled on.
How many times have you gone "Oh, this new thing I have just encountered reminds me of this other thing?". How can you make such leaps without knowing that other thing?
How do you show mastery of a topic where you neither understand fundamentals nor have skills to do a task without help?
Mastery is about demonstrating skills. Skills take practice, hands down.
AI can help us build better tools, provide tutoring, answer broad ranging questions, help relate concepts, and do reasoning for us. All of those things can be good in that they can help us acquire skills faster, e.g. with better practices and exercises. But it cannot replace actual time spent doing a thing.
Wanna learn knitting? You're going to have to knit some stuff.