HN Reader

Unfortunately, this kind of story will continue to be a popular one in newspapers and magazines, garnering lots of clicks. It feeds into the "everything is different now" sort of desperate helplessness people seem primed to adopt with respect to AI sometimes.

Obviously the answer to testing and grading is to do it in the classroom. If a computer is required, it can't connect to the internet.

Caught with a cellphone, you fail the test. Caught twice you fail the class.

The non-story beatings will continue until morale and common sense improve.

Someone I know (not SWIM, it actually wasn't me) did a little bit of work for a high-school-level tutoring company. The company got all their leads from TikTok. They did numbers on TikTok.

The company was tutoring English Literature as one of its subjects.

They were generating English Literature exam problems - for their users - using the ChatGPT web UI.

They would upload the marking spec, and say: "Give me an excerpt from something that might be on this syllabus, and an appropriate question about it".

Naturally, their users - the high school students - were getting, often, hallucinated excerpts from hallucinated works by existing authors.

I think the kids will be fine - it'll be their world, at some point, and that world will look a lot different to now. Maybe that's too optimistic!

I would hope, in that world, LLM literacy amongst adults has increased.

Because I feel really, really bad for all the kids who are beating themselves up about getting badly marked by ChatGPT (I assume) on an imaginary excerpt of an imaginary Wordsworth poem by their functionally imaginary tutor.

It makes me laugh, and reminds me of one of my favourite jokes, about the inflatable boy who - being of a rebellious nature - takes a safety pin to the inflatable school. Chaos ensues. Afterwards, the inflatable boy's inflatable teacher says:

"You've let me down; you've let the school down, but worst of all, you've let yourself down."

I guess I'm suspicious of the linked article. Call me full of hot air, but is it actually a safety pin? Or is it just designed to look really good on an application for an inflatable college?

An older analog is calculators. My college intro to stats course didn't allow them. We did simple arithmetic by hand and looked up things like roots and logs in tables. I still have my copy of this: https://www.amazon.com/dp/0849306922

I just tutored my nephew through his college intro to stats course. Not only are calculators allowed, but they had a course web app so that all they did was select a dataset, select columns from those datasets, and enter some parameters. They were expected to be able to pick the right technique in the app, select the right things, and interpret the results. Because of the time savings, they covered far more techniques than we did in my day because they weren't spending so much time doing arithmetic.

Despite lots of cries about "who will know how to make calculators?", this transition to calculators (and computers) being allowed was unavoidable because that's how these subjects would be applied later on in students' careers. The same is true of AI, so students need to learn to use it effectively (e.g., not blindly accepting AI answers as truth). It will be difficult for the teachers to make their lesson plans deeper, but I think that's where we're headed.

Another lesson we can draw from the adoption of calculators is that not all kids could afford calculators, so schools sometimes needed to provide them. Schools might need to provide access to AI as well. Maybe you are required to use the school's version and it logs every student's usage as the modern version of "show your work"? And it could intentionally spit out bad answers occasionally to test that students are examining the output. There's a lot to figure out, but we can find inspiration in past transitions.

I can definitely vouch for this based on stories from my wife stories (teacher at a private school), her friends (fellow teachers), and my experience working at coworking spaces and coffee shops. [^1]

LLMs can be amazing [^0] as an assistive technology, but using them as a "do it for me" button is just way too easy, so that's how they are de facto used.

I believe it will take about 5-10 years for us to fully comprehend how damaging unplanned remote classrooms and unchecked LLM use in the classroom was. Like heroin, it will be extremely hard to undo our dependence on them by that point. I'm pretty scared for how our students will fare on the global scale in the coming years.

[^0] I strongly believe that 60% of the value of LLMs can be realized by learning how to use a search engine properly. Probably more. Nonetheless, I've fully embraced my accidentally-acquired curmudgeon identity and know that I'm in the minority about this.

[^1] You won't believe how many people leave their laptops unlocked and their screen's contents visible for everyone to see. Committing identity theft has to be easier than ever these days. This basic infosec principle seems to be something we've lost since the great WFH migration.

I'm a sysadmin for a public school district and the admins are working on rolling out Gemini for students/staff. I've shared all the studies I can find about cognitive decline associated with LLM use, but it seems like it's falling on deaf ears.

Makes you realize how fast this has happened. This is a "get off my lawn" article by a high school senior.

It's part of the job of education to instill some common culture. (Which common culture varies, but not all that much outside political topics.) For students, questions about that culture are new issues. LLMs have digested a huge amount of existing material on it. LLMs are thus really good at things students are graded upon.

This gives students the impression that LLMs are very smart. Which probably says more about educational practice than LLMs. The big problem is not cheating. It's that the areas schools cover are ones where LLMs are really good.

There's no easy fix for this.

"The technology is producing a generation of eternal novices, unable to think or perform for themselves."

A quite possible future: you're surrounded by dead-eyed humans with AI implants who mindlessly repeat whatever the chatbot tells them.

A high schooler writes about how absurd is a school system based on a fixed one-size-fits-all education and how AI is making this so obvious.

> We used to share memes about pounding away at the keyboard at 11:57, anxiously rushing to complete our work on time. These moments were not fun, exactly, but they did draw students together in a shared academic experience.

This reminds me of type 1 vs type 2 fun. Type 1 fun is fun in the moment; drinks with friends. Type 2 isn’t fun in the moment but is fun in retrospect. Generally people choose type 1 if given a choice but type 2 I find is the most rewarding. It’s what you’ll talk about with your friends at the bar. I know it’s very much old man, well I guess this high schooler is too, yelling at clouds but I do worry what the elimination of challenge does to our ability to learn and form relationships. I’d expect there to be a sweet spot. Obviously too much challenge and people shut down.

> The technology has also led students to focus on external results at the expense of internal growth.

But this has already been the case. We have all been running behind numbers for so long. Nobody gives a damn about actually learning.

I started learning after I got my first job. Started focusimg on literature, arts and languages a lot more after I started working. AI only amplifies this to the next level.

There are certain aspects like disciplinary and on time scenarios which I can agree with. But the education system has not been about education since for a long time. Sure, premium institutions had something going on. But maybe that is what will be takenover by AI as well?

You can avoid this problem easily not letting students using phones or computers in the classroom and making doing more tasks at the classroom than at home.

If you're going to say "but in a working environment you use a computer", then teach them how to use text processing and spreadsheets int the computer room, a thing that didn't happen today in most schools btw.

The examples used in this student's classes demonstrate how inept the teacher is at designing a classroom promoted for learning rather than performance. If I were a teacher and I send around sheets of paper of exercises, my first directive is "Close all laptops" (or never have them there in the first place) followed by "If I see any use of technology to cheat, you have failed the exam." This is EXACTLY what is was like when I was in school only a few years ago.

Similarly for the debate club - why are teams allowed to have any technology in the hall in the first place?

Education is supposed to be difficult - that's how we learn!! Teachers seem to pander more and more to students who complain that "This is too difficult". As if easy learning was ever a thing!

They seem surprised about this, but the cheating kids were always there. They just use different (and arguably easier) tools now. Blocking the AI won’t stop them from cheating, it’ll just make them revert to the old way of doing it.

Maybe they can allow AI for writing but raise the bar on quality so the blind copy-paste submissions still fail. I've still never read a good AI-generated doc at work, it's always verbose and aimless. At this point I close the doc if I catch a whiff. Unlike the AI code which is fine.

It's probably either that or ban it and do everything in-person, which might have to be the stopgap solution.

New York originally banned ChatGPT

https://www.nbcnews.com/tech/tech-news/new-york-city-public-...

Then reverseed the ban

https://www.nbcnews.com/tech/chatgpt-ban-dropped-new-york-ci...

An ex-gf of mine spent four years going through university to become an Occupational Therapist. She's severely dyslexic, so the university provided her with all sorts of assistance to get through her degree, from scribes in the exams to extra time for tests. She passed, and became a qualified Occupational Therapist. She landed a job in a local hospital. And on day one, was handed a huge pile of paperwork to complete. No scribes, no assistance, just "this is the job, get on with it". She failed the job, left after 3 months, spent a couple of years rethinking her entire life, and switched to a completely different career with less paperwork.

My point is that education has to be aligned with the actual world outside.

Everyone uses AI now, for all sorts of tasks. And if they don't now, they will in the next few years. Trying to exclude AI from education is not only pointless, it's doing the kids a disservice: AI is going to be a large part of their future, so it needs to be a large part of their education.

If we follow the implied course of TFA we'd reduce AI use in schools and go back to old-skool teaching methods. Then that cohort of kids would get their first job and on day one they'd be handed an AI and told "this is the job, get on with it". Like with my ex-gf, everything they were taught would be useless because the basic foundation is different.

I know education is not entirely vocational, but if it moves too far from the world of work that everyone actually spends most of their time in, then it gets too theoretical and academic. AI is part of it, education needs to change.

So I am going to be critical of how this piece is being presented. I'd like to start that by saying it is well written by an obviously talented person who will do well in life. I'm prefacing my comments with this because I really don't want my comments to be considered disrespectful or antagonistic. I just don't think this can be considered a representative view from 'A High Schooler'

I don't think this is necessarily wrong, but over the years I have seen many high achieving senior students writing about or being interviewed about topics where they are less representing the community they are a member of, but the opinions that supports those who give them praise, support, and opportunities.

I don't think it should reflect poorly on a student that does that, but I also don't think you can draw significant conclusions from their stated opinions. Most people like this have not yet found their own voice, what you hear is often the voice that they think they are supposed to have. For many, tertiary education is as much about finding that voice as it is studying specific fields

Measuring what is best for students is an incredibly complex task, not least because 'best' can mean different things to different people, and often the wellbeing of the student is not considered high enough. There is science here, but given the importance of the field, way less than there should be. Changing education for the better is extremely difficult when the science conflicts with public opinion. There are forces at play that know that their only path to success is through swaying public opinion because the science is against them. The science of education can be laborious, slow, and full of difficult to express nuance. It is also the only sure process by which we can find out what actually works.

So by all means follow the argument that it makes, but don't mistake the source as being representative. The author expresses their love for debating and development. I imagine that they would respect the sentiment that the work should stand on what was said, rather than who said it.

[as a final thought]

It would actually be an interesting research project to find articles like this written on contentious issues over the years and locating the writers to get their opinions on them with the benefit of hindsight.

I've been working hard at this problem over at https://kurnell.ai. Our thesis is that going forward it's unrealistic to ban or detect-and-reprimand AI, therefore we need to meet students where they're at and democratize access to the best AI for all.

We have great traction with universities in USA and Australia. The flywheel that we've constructed means that students are being prepared for industry + research in a Post-AI world, and professors can see exactly how students are using AI tooling. Our findings are that knowledge of how students are using AI goes a long way to helping institutions adapt.

Keen to chat and share our findings - reach out at hamish(at)kurnell.ai !

Idk this doesn’t really click for me. That first cheating example was entirely possible before GenAI (through sparknotes). And yet, we learned. Learning has always been a choice

In the novel Tom Brown's School Days [0], published in mid 19-th century and depicting a public school in the first quarter of that century, there is a scene describing how students those days used "vulgus-books": collections of previous years' students' homeworks in Latin that new students copied from for their assignments in Latin composition. There was a boy named Arthur in that story who refused to copy from other students' essays, and worked on the compositions himself. He also tried to convince his friends to abandon that practice of copying, and to write their assignments on their own.

This is what this article reminded me of. The student writes how her classmates use help from AI as if she cannot decide for herself to do the work on her own if she cares about learning. She writes as if she is devoid of agency.

The Atlantic published a post on reddit about this article, titled "I’m a High Schooler. AI Is Demolishing My Education." [1] And yet, it is the other students that the author primarily focuses on. Why does other students' cheating demolish _her_ education?

[0] - https://gutenberg.org/cache/epub/1480/pg1480.txt

[1] - https://www.reddit.com/r/ArtificialInteligence/comments/1n7o...

Multiple uses of em-dash. Must be written by AI.

It seems we should go back to low tech (books, handwritten assignments, paper exams...) for all courses except the technology ones (which are still a must, but need to be improved).

Of course, they could still AI to help them with homework but people were already copying the homework from their mates. But if they just copy and don't learn, that would be surfaced during the exams.

Instead of not letting the students to use AI, how about we adjust the assessment method, and learning objective, to let the students do something that AI along cannot do, so that they can do it w/wo AI but still build up their own skill.

In the workplace, we're using AI anyway.

I'm not sure if this direction is suitable for kids, like we still learn to do calculation even when we have calculator (which is needed for some cases, but for complex math, we opt for tools)

to me this is an indication that the priorities of these formative years should change, and it doesn't look like what "school" looks like

the montessori and sudbury school model always seemed closest to what was necessary, although now I wonder if even those are cracking at the seams with outsourced thinking

regardless, I think a re-evaluation of the point is absolutely necessary.

self-motivated children are rare and require a specific environment and support system to thrive in, but will always be there to escape the more obvious return to serfs working on fiefs, unless born into capital themselves

Two things can be true: keep phones/AI out during class, and admit they exist at home. A workable pattern: handwritten in-class checks; take-home projects with an artifact trail (sources, drafts, prompts); a 5-minute oral defense. AI becomes something to critique, not a crutch.

It's weird. All of our attributes which we hold and value, and develop via a mix of genes and training - intelligence, but also strength, stamina, reflexes - we acquired, if you strip it all off to basics, to feed and to procreate. That's all evolution cares about.

Now, we are social animals, and we grew to value these thing for their own right. Societies valued strength and bravery, as virtues, but I guess ultimately because having brave strong soldiers made for more food and babies.

So over time, we tamed beasts and built tools, and most of these virtues kind of faded away. In our world of prosperity and machine power on tap, strength and bravery are not really extolled so much anymore. We work out because it makes us healthy and attractive, not because our societies demand this. We're happy to replace the hard work with a prosthetic.

Intelligence all these millenia was the outlier. The thing separating us from the animals. It was so inconceivable that it might be replaced that it is very deeply ingrained in us.

But if suddenly we don't need it? Or at least 95% of the population doesn't? Is it "ok" to lose it, like engineers of today don't rely on strength like blacksmiths used to? Maybe. Maybe it's ok that in 100 years we will all let our brains rot, occasionally doing a crossword as a work out. It feels sad, but maybe only in the way decline of swordsmanship felt to a Napoleonic veteran. The world moved on and we don't care anymore.

We lost so many skills that were once so key: the average person can't farm, can't forage, can't start a fire or ride a horse. And maybe it's ok. Or, who knows, maybe not.

I’m an AI engineer but I think schools need a nuclear option

Banish tech in schools (including cell phones) (except during comp classes) but allow it at home

Ie in high school only allow paper and pencil/pen

Go back to written exams (handwriting based)

Be lenient on spelling and grammer

Allow homework, digital tutoring AI assistants and AI only when it not primary- ie for homework not in class work

Bring back oral exams (in a limited way)

Encourage study groups in school but don’t allow digital tech in those groups in class or libraries only outside of campus or in computer labs

Give up iPads and Chromebooks and Pearson etc

> or that they remove a sense of urgency from academics

That was one of my frustrations with "prep" school: An artificial sense of urgency that does not, in any way, reflect how one leads a happy, healthy, and successful life; nor does one need a sense of urgency in academics to grow into an adult who makes a positive contribution to society.

> Some students may use these tools to develop their understanding or explore topics more deeply, ... can also be used as a study aid

I think the same can be said about internet searches. Altavista came around when I was in high school; and I lost all motivation to memorize arcane facts. The same can also be said about books and libraries.

Instead, it's important to realize that a lot of topics taught in schools have to do with someone's agenda and opinion about what's important to know, and even political agendas; and then accept that many lessons from school are forgotten.

> Student assessments should be focused on tasks that are not easily delegated to technology: oral exams ... or personalized writing assignments ... Portfolio-based or presentational grading

Those are all time consuming; but they miss a bigger point: What's the real point of grades anyway?

Perhaps its time to focus on quality instead of quantity in education?

The real irony is that this article was clearly, edited and formatted by an LLM.

Well, we can hope this is only a transitional issue. Here is why.

AI will, like previous technologies, enable some of us to become more productive. In fact, it raises the bar on productivity, since an experienced programmer can now create much more code. (An inexperienced one can create much more mess, so you might not see it in aggregate statistics).

When it comes to the classroom, we should do the same. We raise the bar so that in fact, you cannot do anything without using AI. Much as you would run out of time if you didn't have a spreadsheet in a stats course 20 years ago, or pandas 10 years ago. The new tech enables more work to get done in the form of learning more high level things, while relegating lower level things to just building blocks that can be understood in the same way we understand reference texts, ie "I've seen the principal once, and I can find it again if I get to that level of abstraction".

Teaching needs to change. Perhaps the thing to do is have an Oxford tutorial rather than traditional class. For those who didn't attend, a tutorial is basically two students and a professor in a room, talking. You can't hide. You can prepare however you like, and you should spend quite a lot of hours if you're sparring with a politics or math professor. But once you're in the room, it becomes painfully obvious if you are unprepared. This is a way to get accountability.

At the moment, we have this high school system testing that is a factory. Every test is done as a thing that is easily marked. Multiple choice, or short answer, or short essay. It encourages superficial learning when you know you can dance around the important topics and just pick up the easy points, as well as simply avoiding silly errors. You can also win by simply learning the likely questions, and aping the answers.

Have a weekly small-group session with an expert, and they can find your limits. Yes, it will cost money.

There's going to be so much more tests in classrooms.

The only thing I'll say that's good is it might lead to less homework, which I always thought was poorly designed and mostly busywork.

For reference, a similar HN story from about a month ago [1] on teacher usage of AI tools in the classroom.

It's pretty fucking dire. I think we're failing an entire generation of kids and the ramifications of this is going to be real bad in 5-10 years. I've heard similar stories from friends of mine whom are teachers.

[1] https://news.ycombinator.com/item?id=44808122

A prescient cartoon from 15 years ago: https://archive.rall.com/main.php?g2_itemId=14184&g2_imageVi...

I honestly think school systems will use AI as a way to pass "difficult" students. It has been a real problem dealing with students that are failing, and school systems have just basically looked the other way. Now they can look the other way as far as AI and most kids will just magically pass everything and move along their merry way.

It's awful, but I think we'll see it happen, sadly.

I recently found out that my nephew's school had no take-home homework before high school, instead having kids complete assignments during class time. At first, I was flabbergasted that they would deny kids the discipline building of managing unstructured time without direct supervision. Homework- at home- seemed like such a fundamental part of the schooling experience.

Now, I'm thinking that was pretty much they only way they could think of to ensure kids were doing things themselves.

I know it was a rough transition for my nephew, though, and I don't know that I would have handled it very well either. I'm not sure what would be a better option, though, given how much of a disservice such easy access to a mental crutch is.

The lesson here is adapt or die. The things they thought were important or difficult or impressive are no longer any of those things. Regurgitating information on a test or generating prose from the notes you took in class are tasks which are easy to stereotype, and now readily automated.

Rather than framing this as destroying education, it should be interpreted as proof that these tasks were always shallow. AI is still much worse than humans at important things, why not focus on those things instead?

The school systems are clearly not keeping up. Any kid who isn't doing project oriented creative work, aided by an LLM as needed, is not preparing for the the world they will likely inherit.

This take from a Hermione-type High School senior shed next to zero new light on the subject. Yes, we know AI is redefining school and jobs and daily life. The perspective of an obnoxious A+ type student isn't helping, especially because you kind of can read between the lines that she isn't friends with these kids using AI, which would give her a deeper perspective of why and how they are using AI.

Is this what The Atlantic has come down to, publishing a complain-y piece by the class president?

EDIT: For anyone struggling with my criticism of the article, I very much agree that there is a problem of AI in education. Her suggestion which is "maybe more oral exams and less essays?" I'm sure has never been considered by teachers around the world rolls eyes.

As for how to tackle this, I think the only solution is accept the fact that AI is going nowhere and integrate it into the class. Show kids in the class how to use AI properly, compare what different AI models say, and compare what they say to what scholars and authors have written, to what kids in the past have written in their essays.

You don't have to fight AI to instill critical thinking in kids. You can embrace it to teach them its limitations.

Maybe AGI is impossible not for any technical reasons, but because AI is a self-limiting technology? As AI improves, it destroys the educational system that creates the researchers that develop it.

Proves the simulation of AI never required the virtual plug in pod. It's simply pervasive in our reality as Baudrillard asserted (contrary to the Matrix dichotomy), and now requires its overthrow to obliterate automated seamless simulation from taking over. Say goodbye ML, we won't be missing you.

I struggled in high school. I went off my ADHD meds in favour of a social life, a choice I don’t regret. But it made classwork and homework absolute torture. The torture mostly sucked but I did help me grow. I’m convinced I would have cheated if I had access to ChatGPT, and that’s pretty upsetting to think about when thinking about today’s students.b

I hope that how we educate changes, forces by AI, improving in ways that would have helped people like me. I worry that might mean lessened access for all, if it requires the cost to go up.

Imagine back in the days when calculators were just invented. An 8 year old kid might have the similar complain: “my classmate finished a 4 digits number multiplication problem in 5 seconds which generally took 1mins.” People might say, in the long term, the kid who cheated would be less proficient in arithmetic, which turned out to be true. But when you think about it, it seems not the end of the world when most high schooler in US cannot do complicated arithmetic quickly and accurately without a calculator.

The future of high school tests is using an AI agent to research and build something amazing in a fixed time window, not knowing what it's going to be ahead of time. We just have to start thinking bigger with education, the trivial bullshit that we used to burden people with has been made obsolete. Education becomes a building workshop with a focus on helping students come to their own answers from first principles using modern tools.

i get it, but i also think about all the useful things i could have been doing (perhaps now assisted by AI) instead of pounding through Algebra homework and English essays all night...

The atlantic loves selling ai despair, alongside anti family rhetoric- I genuinely believe they have evil intentions and it is important to stay away from them.

It seems like the real problem here is the curriculum. The school should be removing students' ability to use AI on these assignments, and it really isn't that hard to do.

Phones shouldn't be in the classroom, and devices used in the classroom shouldn't have any access to AI.

Students shouldn't really have homework anyway so I think it's completely reasonable to just have kids doing work on pen and paper in the class for the most part.

What are we going to do to solve this?

I'd say oh-but-what-about-calculators-i-am-very-smart comments call for a drinking game, but no one should be doing this, as they'd die

So, its a "Kids these days", written by a kid.

I've seen the same commentary about:

Spellcheck

Typed material

Computer art programs

Calculators

One of the things that doesn't seem to be emphasised enough is that LLMs can be great as Tutors for students. E.g. They could really help students not understanding fully a section of Mathematics and really break down the logic.

Teachers can also use them to mark homework.

They are a boon as much as they are a bane.

Are high school teachers too lazy to do essays in class like they always were 20 years ago? Stop fucking using Google Drive or whatever. Start teaching these children like you used to do not so long ago. For FUCK'S sake.