> Pupils from first through seventh grade, aged 6 to 13, should as a general rule not be using AI, while those in lower secondary school, aged 14 to 16, can cautiously adopt tools under teachers' supervision, the government said.
Sounds right to me. Kids under 13 need to learn to read, write and comprehend text. Generative AI is not going to help them with those skills.
They can play with AI at home, and after 13 they can learn how to use AI productively and, ideally, in a way that enhances rather than detracts from their education.
Also from the story:
> Facing a broad decline in education test scores, the government in 2024 banned smartphones from schools and has given teachers back more powers to enforce discipline in the classroom.
A big hooray for that. Will be interesting to see what impact that has on Norway education - a quick search just now didn't turn up any detailed studies, presumably those will show up eventually.
You mean chatgpt style AI won't help them with those skills?
If a human parent or teacher can help with skills like reading, an AI system can too, once it's trained and designed to do so. (How good are humans at teaching reading anyway?)
Norway spent two decades digitizing classrooms and is now unwinding it. Seems a bit shortsighted and reactionary although I think they are trying to do the right thing.
Kids can read, write, and comprehend text at 8. I don’t even like LLMs and I’m against this mess. Imagine having regulations rolled out when we were 8 saying “you can’t use the internet!” And I was running my own websites by 10 years old.
Let’s stop pretending this tech is as interesting as we wish it was. If we want to ban models in school, ban laptops/chromebooks with internet. I don’t see the difference at this point.
> Between 2017 and 2023, there were increases in the percentages of adults performing at the lowest proficiency level (Level 1 or below) in both literacy and numeracy: in literacy this percentage increased from 19 to 28 percent and in numeracy from 29 to 34 percent.
> Adults at level 1 are able to locate information on a text page, find a relevant link from a website, and identify relevant text among multiple options when the relevant information is explicitly cued. They can understand the meaning of short texts, as well as the organization of lists or multiple sections within a single page.
>A sizable portion of the US adult population effectively can't read, write and comprehend text.
Yes and AI isn't to blame for that. It's the governments, schools, teachers, parents, teacher's unios, who taught them(or more accurately didn't teach them) and graduated them out of school anyway regardless just so they don't look bad in statistics. Sorry but if you graduate people out of high school who can't read you should be trialed for fraud. Simple as.
People blaming AI for adults unable to read puts us back to the 90s when Doom was to blame for school shootings or back to 60s when rock music was to blame for juvenile delinquency, all of them being wrong, and they're wrong here too. People always want to blame a third party external scapegoat that isn't' the parents and isn't the government, for the problems of their kids.
AI hasn't had a chance to demonstrate if it helps or hurts education yet.
That's the big problem with education in general. If you introduce a new factor to children's education you can't realistically measure the effect it has had for about five years, because you need to wait for a cohort of kids to go through that system and then see how they did.
This means that if you introduce something with clear negative effects it will be five years before you spot them!
That's pretty catastrophic given that ChatGPT only emerged in late 2022 and only got good around early 2024.
Nobody is blaming AI. The point is we don’t have the luxury of throwing nonsense at our kids when they’re illiterate. Particularly not nonsense where all the evidence shows it harms on average more than it helps.
In this case it's the US that's shitting on the US. These numbers don't compare the US with other countries, they compare the US in 2023 with the US in 2017. And the numbers are from the US government National Center for Education Statistics.
From what I can tell the average school at best can aspire to teach kids how to work and how to socialize. That's it. I'd personally be very happy with computers mostly going away from school too. Most actual learning and exploring will hopefully happen at home.
This is really not true. Kids do actually learn a lot in school - includong weak students. And you actually see huge difference between places with and without schools.
If you think an 8 year old can comprehend text at the same level of a 13 year old (or an 18 year old for that matter), I don't know what to tell you. Reading comprehension doesn't peak at 8.
Yes, there's also a push to ban all computers in classrooms because data is showing that it's of no benefit and if anything is a negative effect on education.
I think this is basically right. You don’t hand out calculators before kids understand arithmetic. LLM version is sneakier because skipping the work still produces something that looks finished.
Not quite. The LLM gives you a statistically probable sounding token stream. The calculator gives you a qualified answer within documented and deterministic limits of the device.
Teacher uses AI, creates lesson plan -> AI creates assignments -> Teacher gives assignments to students -> Students use AI to do assignment -> Teacher grades with AI -> Principal uses AI to monitor teachers progress.
This is happening at schools nationwide. It is unstoppable at this point. It's a bizarre charade.
You forgot the parasite companies that sit between the teachers/student/principal which pretend they detect the big bad AI assisted/generated work to punish individuals using AI (with a majority of false positives and always late by 1 generation of models). The charade wouldn't be complete without rent seeking intermediaries.
This isn't about teachers using Ai to make their life easier, otherwise it wouldn't only apply to elementary school, and from what I read it applies to students.
Students likely aren't allowed to use AI anyway for assignments. Or are they? That's the question, what is actually being banned if anything
AI in 1:1 tutor mode with proper hardware (live scanning pen and paper), harness and guardrails should be wildly successful (in terms of education outcomes) especially in elementary school.
Disagree. AI has no business being used in 1:1 tutor mode before the hallucination and sycophancy issues are completely resolved. As is, I can easily see it being a hindrance to building actual understanding.
Just one example - it's very common to see ChatGPT and the like respond with "you're absolutely correct! Great insight" to something that is a complete misunderstand.
Teachers hallucinate too. I’ve had creationists and communists and tin-foil-hat (chem trails, 5g, etc) teachers. Surely you can imagine an AI tutor that is higher than zero ROI.
> I’ve had creationists and communists and tin-foil-hat (chem trails, 5g, etc) teachers.
I certainly have, too, but there is still a difference between a person who has a factually incorrect but consistent worldview and an LLM which simply reflects the worldview of the user or even changes between queries.
I don't think creationists have any business being in schools either, for what it's worth, but I think it's easier for a teenager to sort out "Mr. Smith has no clue what he's talking about" vs "I have no clue what's true because the LLM everyone expects me to learn from just confirms everything I ask regardless of what I'm asking".
I think AI should be used in higher level schools but with the added requirement that the output will be held to a much higher standard and that it's fact checked. Teach the students to use AI to reach a higher level while mitigating the inherent issues like hallucination and sycophancy.
“AI adoption raises homework scores by 18% and reduces completion time by 30%, but lowers monthly exam scores by 20% within six months. High-stakes entrance-exam scores fall by 18 and 24%, with the full penalty emerging only after about two years.”
Yup. Short-term metrics juice. Actual comprehension and cognition falls. This seems to be the case across the board, including with adults.
I’m genuinely optimistic that there is a way to make AI helpful in education. I just don’t think we’ve found it yet. (We certainly haven’t demonstrated it.)
This is probably the big problem, or at least one of them.. If you use less time on learning, it will probably be harder to remember what you learned also. We need to spend some time to make it stick
The behavioral issue I see is that LLM users tend to immediately reach for an LLM and do their thinking in concert with it.
This tempts users to approach problems by first feeding them into the LLM and then simply following the route the LLM lays out, which does improve task completion time for tasks that the LLM can simply regurgitate, but it stops the user from developing the actual critical thinking skills that school is supposed to teach.
AI users who maintain similar homework completion time as non-AI users experience small learning losses.
This was a surprise too me. I would have thought otherwise.
Would love to see some evidence about if more or less people fall behind and have worse results. In my head the AI should be able to get the weakest students a bit highere.
I think AI could (and by some students probably already is) be used to help a student better understand the material, and faster than you could before. I still recall some parts of Physics taking a while to click, and often having to reread different sections of a textbook to try and understand the what and why behind something.
The biggest issue is a child has to want to do that, since they also could just ask the AI for the answer and then go back to playing video games. End of the day past age 13 or so I just don't see legislation making any difference, they'll find a way past any law blocking them from using AI. Like a lot of education it'll probably come down to parenting.
> I think AI could (and by some students probably already is) be used to help a student better understand the material, and faster than you could before
I think so too. But we haven’t demonstrated we’ve found how, in kids or in adults.
> biggest issue is
We genuinely don’t know what the biggest issue is. We just know it doesn’t work. There is zero quality evidence for AI helping with learning or cognition in kids or adults. (Happy to be proven wrong. This is a fast-moving and big field.)
> they'll find a way past any law blocking them from using AI. Like a lot of education it'll probably come down to parenting
And community. Rich towns restrict devices in school, monitor use at home and thus will have less of an issue with AI exposure.
>I think so too. But we haven’t demonstrated we’ve found how, in kids or in adults.
Ask chatgpt or claude, on their highest model (probably unnecessary but I'm sensing a vibe) to explain a simple linear algebra problem, and if you don't understand it, ask about what part you don't understand.
And if you truly believe it made something up, prove it.
This is seriously the easiest thing to prove out there, you can see for yourself in the next 5 minutes.
> And if you truly believe it made something up, prove it.
You seem to be assuming that the issue is around factual correctness, and that may be the case but the evidence we have so far doesn't support jumping to such a narrow cause.
Is the poor performance because the LLMs are frequently wrong? Unknown.
Is it because the LLMs are sycophantic? Unknown.
Is it because the chat interface is a poor one for learning? Unknown.
What we do know is that students who rely on LLMs learn less and perform worse in the long term. And that alone is enough evidence to support a ban. If better tools come along in the future and are shown to aid learning, then the ban can be re-evaluated.
Sure. I already know linear algebra. If it’s a new branch of mathematics, this is a terrible way to learn.
Again, the research points almost exclusively in one direction when it comes to learning and cognition around AI. You’ll solve more problems more quickly but wind up learning and thinking less.
Honestly, what do you think a teacher does that an llm can't? If you want to learn you absolutely can ask an llm how to _solve_ x and explain the steps and why.
My leaving out the word solve seems to have led some of you astray, I apologize.
Again the problem is you have the option to solve your problem and move on without understanding it. That does not mean you can not use the tool to understand the problem and how to _solve_ it.
I live in fear that instead of learning how to use the tool, some might just vote to ban the tool.
> what do you think a teacher does that an llm can't?
I don’t know! It’s an interesting question. All we know is it does.
> That does not mean you can not use the tool to understand the problem and how to _solve_ it
It doesn’t. But we have no evidence it can.
We have lots of evidence of people thinking they’ve learned something, taking a benchmarking test, and being found wholly deficient compared to folks who worked through a textbook, went to a class or even solved problems off YouTube videos or instructional websites.
> to explain a simple linear algebra problem, and if you don't understand it, ask about what part you don't understand.
The goal is not to understand a linear algebra problem. The goal is to learn how to solve it using lessons and techniques taught beforehand. Aka not to get a fish, but learning how to fish.
Considering that the paper concludes that even students who take the long approach and use LLMs in the most appropriate way for learning still retain less over the long term than students who simply don't use LLMs, I think it's likely they didn't read the paper in the first place.
We thought the same of electronic devices in general and digital learning content specifically. In actual practice both result in lowered test scores and declining critical thinking skills.
"You just have to repeat the prompt 3 times and then spin around counter-clockwise twice! That always works for me. You obviously just don't know how to prompt the model correctly."
Every time I see LLM enjoyers yapping on like this, it just reminds me of people trying to read tea leaves. There's all these goofy little rules about how to structure the prompt and how mean or nice to be to get it to work optimally, but I think it's obvious that most of these users are just seeing incidental more successful outcomes in a largely random system and extrapolating from there because it makes them feel in control.
I'm glad this is the case. It is the correct outcome.
I've heard students actually discussing that they will just use an LLM to shortcut work. I even have friends in their 50s who can barely think for themselves now without having to refer to "AI". And at least two of them are teachers.
Leading on from that, the staff are the most dangerous. My daughter has had generated exercises provided to her from multiple teachers, which are quite frankly entirely wrong. This was hilariously pointed out after I called a meeting with her mathematics teacher over it. They questioned my knowledge on the matter with the insane assumption that "AI is foolproof". I had to hit them with a clue stick then.
No one taught anything of value. No one learned anything of value. I am very worried we'll see a lost generation at some point rippling through the ages.
US legislators have almost no say in how schools are run. The DoE is a husk, and states call the shots. In my state of Oregon, you will have wildly different curricula and standards depending on your district, because there's almost no state oversight either.
They made a choice - US legislators effectively gave up their control of how schools are run when the conservative coalition allowed the trump administration to dismantle the DoE without congressional approval. Congress itself could legally retain control of education, but if congress refuses to assert that power then it is meaningless.
The end goal is to dismantle public education and route public money to religious and private schools.
I think it’s quite tricky. On one side, writing is a form of thinking and cognitive training.
Just NOT doing that work by having AI simulate it is not good for anyone’s cognitive development.
At the same time, anyone growing up today will be using LLMs for massive parts of the jobs they grow up to do. So they should learn about it.
I really feel for teachers/educators right now. It must be hard to remain demanding and insist on educating kids well while also preparing them for the future they’ll actually live in.
> anyone growing up today will be using LLMs for massive parts of the jobs they grow up to do. So they should learn about it
Whatever AI looks like in 20 years is going to be so different from what it is today as to make distracting from basic skill-building an almost-certainly net negative educational effort.
I agree to some degree. But by that logic, should kids in the 2000s not have learned about the internet because the internet fundamentally changed between then and now?
I think that if anything, it’s really good I learned how to operate a computer and the Internet BEFORE what the Internet became.
I pity the generation who don’t understand a computer’s folder structure because they grew up with smartphones and TikTok.
> At the same time, anyone growing up today will be using LLMs for massive parts of the jobs they grow up to do. So they should learn about it.
There's not so much to learn they can't put it into a high school course. Adults currently in the workforce haven't been using AI since they were in elementary school, and they're adjusting fine.
I would disagree that they’re adjusting fine. So much of the stuff we see now is full AI slop clearly created with the first output of ChatGPT. It’s like saying we don’t need to teach kids about the internet because older generations grew up without it and they’re scrolling TikTok.
a) what’s the actual percentage of professionals who actively use AI? It’s much smaller than we think in tech.
b) what percentage of those people understand the very basics of how LLMs work (e.g. token prediction, context windows, etc)
c) what percentage of those people understand AI Agents (or any of their ingredients (APIs, credentials, etc.)
You quickly arrive at a tiny fraction that has a real clue about what they’re doing.
Surprised to read this as Norway also has Sikt AI for schools, where teachers can monitor how AI systems are used. Seems like it has both embraced AI and banned it.
> Seems like it has both embraced AI and banned it.
Indeed, seemingly they done so by age/educational progression:
> Pupils from first through seventh grade, aged 6 to 13, should as a general rule not be using AI, while those in lower secondary school, aged 14 to 16, can cautiously adopt tools under teachers' supervision, the government said. In upper secondary education, from ages 17 to 19, students should learn to use AI appropriately so that they are prepared for further education and work, it added.
I wonder if the struggle is really comprehending thoughtful selective adoption.
Schools are the place where the product is a more fully developed person. There's no LLM shortcut for generating that. There are many ways you could use LLMs that would discourage it. There may be some that can encourage it.
Personally, I can see aggressively keeping kids away from LLMs until they've learned effort, living in tension/frustration, the pleasure of breaking through to discovery, trust evaluation, hypothesis/test cycles, and good socratic dialogue from the learner's side.
It may be possible at intermediate phases to prime some models to help with this process.
Social media execs are already known for keeping their children off their platforms* and even phones so my question is: Do the leading ML/AI people let their children interact with LLMs yes or no?
Broadly speaking, the wealthier the parents the more restrictive device is in general in the household.
This used to be a tech/non-tech line. It shifted to class sometime over the last ten years. The iPad kids are probably getting served slop. The AI-employed parents don’t have to directly police AI exposure because their kids’ device use is already controlled; at school, at extra-curriculars and at home.
The question is do the allow them to use their laptops to access llms. Like I assume before ai they where not allowing the kids to fry their brains doom scrolling but would allow programming. They weren't blocking all tech as they aren't raising subsistence farmers.
The difference is between asking to explain x^y and reading and understanding the why and maybe go through a test, vs asking to do my homework for me and then go back to video games quickly. If the whole idea of using LLMs is to reduce effort on the part of the pupil, you are probably holding it wrong.
"..ban on smart phones.." "..ban on gen AI.." "..ban on social media.."
yes
YES
YASSSSSSSSSSSS
Ban all the things for kids. I don't want to be interviewing people in 10 years and decline every candidate because they can't correctly answer the question "You are 50ft away from the car wash. Do you walk or drive?"
The question is slightly vague (since I could be going there not to wash my car), but I'm pretty sure in the intended interpretation, the actual distance is irrelevant. :)
The traditional schooling system can't stand that they are being outcompeted by AI and are trying to use the government to maintain their monopoly and keep those tax dollars flowing.
> The traditional schooling system can't stand that they are being outcompeted by AI and are trying to use the government to maintain their monopoly and keep those tax dollars flowing.
Learning is a conspiracy by Big Knowing, it's all a myth. Let's just ask an LLM to all our thinking, no need to be a functional human.
Yep… I mean scientific studies seem to suggest that you are completely wrong. But big science is all a conspiracy against those AI companies that work so hard for the betterment of mankind after all.
I already saw it in my life: a ban on calculators, a ban on computers. But after a short period of rejection, everybody starts to embrace the new tech. Instead of bans we were getting computer classes in schools.
> after a short period of rejection, everybody starts to embrace the new tech
We’re banning cell phones in school after seeing the evidence, albeit along a class gradient. We’ll probably see something similar with AI. Poor kids get AI in school (and unmonitored at home). Rich kids do not.
I did mathematics, and a fair bit of computer science on the side. This was nearly all on paper, without computers. And I'd go a week or two without a calculator sometimes, mostly employing it when I had a bad hangover.
Do we really need to force technology into everything or are we just used to doing it so see it as necessary?
Great, yet another "no child and no teacher left behind!"
I was gifted kid, bored to death at basic school. I was reading books under the table, and was lucky to have tolerant teacher. Total ban would just push me to misbehaving and disrupting the class.
AI is amazing tool for learning, if Norway can not harness it, there is something deeply wrong with the educational system. Perhaps teachers unions?
Norway has a big problem with young immigrant kids at school not speaking Norwegian. Right now other kids are expected to teach them basic language, holding back their own development (like learning reading and writing)! Again, AI could provide amazing help here!
I would allow schooling in other languages in primary schools in Norway. Basic education is a human right, and it is not right to force kids out of their mother tongue. Insisting on some obscure language is just misplaced nationalism!
I wonder if at some point we will look back on stuff like this as back in the 1990s schools banning internet research and search engines. Obviously that seems ridiculous now, but the internet was big and scary back then too.
Those bans were implemented without evidence. We have evidence AI exposure reduces learning and cognition. There are probably situations where it enhances it. But we haven’t delineated those yet, and so shouldn’t be rolling out a half-baked system more likely to hurt than to help.
Were schools banning search engines? I remember teachers recommending Dogpile (because it would combine search engines), and we did some computer lab stuff in the 00s, but there was no one saying that we should ignore search or the internet altogether.
In middle school I remember being assigned a book report that would include the author's biography. I'd just finished a book (The Gammage Cup) and of course my local library did not have any information about the author. So in that situation it was assumed that you would learn traditional research methods, but also that you would just pick a classic book where the information was readily available.
Few schools allow unrestricted internet access these days.
The general question here is risks vs rewards and with any new technology both are unknown making caution perfectly reasonable be that internet searches or anything else.
So sure in 30 years the policy will look different, but that doesn’t mean they are making the wrong decision.
I would assume if children are allowed to use AI without rails as a shortcut it will undermine their learning, and it's used for feedback and as a patient tutor it would accelerate their learning?
It seems like the problem is that they don't have the science and tooling to use it constructively at scale, so the desperate solution is to ban it outright until a scalable constructive approach is understood?
The article doesn't explain any of this directly...
It's frustrating to me when bold statements are directed at "AI" holistically and vaguely, completely ignoring any nuance.
There is a massive gap between letting elementary students free reign use chatGPT 3.5 (hallucinations and all) to do whatever, vs using a very guard-railed pedagogically optimized app powered by a SOTA model to support students in a specific way that accelerates good outcomes.
Most respectful interpretation is that the leaders know this and have a plan to figure it out, but for some reason it's not making it's way into this article. Is the absence representative of the truth of the situation, or some editors choice to pile on to a holistic anti-ai narrative?
> it's used for feedback and as a patient tutor it would accelerate their learning?
We have mounting evidence AI hurts learning and cognition in many circumstances. I have not yet seen similar-quality evidence for it helping.
Given that balance, restricting AI in education in the general population (while studying how to best deploy it) seems prudent. Especially given the Norwegian approach, which gradually introduces AI as kids get older.
I'm actually not really criticizing the decision so much the article and communications around it. If student learning outcomes are crashing and they desperately want to turn them around I understand why they would take dramatic action.
Giving students uncontrolled access to generic LLMs probably would hurt outcomes. Research process is slow (IRB and all that) so they are dealing with data from years ago (models that confident hallucinated a lot more than current SOTA) so if thats what they are basing it on its reasonable.
My frustration isn't with the decision (hey all teachers - no more chatGPT in the classroom). My frustration is with the reporting / nuance of "until we can research this better and figure out how to harness AI to improve outcomes and not undermine them".
> the reporting / nuance of "until we can research this better and figure out how to harness AI to improve outcomes and not undermine them"
It’s balancing the irrationally exuberant narrative of the tech bros and AI pushers. You have to stop the bleeding before you can dress the wound to promote healing.
I don't think one balances a lack of nuance with more lack of nuance.
One avoids nuance for clicks or to propagate a narrative, sew division, distract, etc.
Again. As I said in both my comments, I'm not criticizing the ban, I'm criticizing the absence of any communication regarding a plan for researching potentially constructive uses. As a reader, I can't tell if the Norwegian leaders have no plan, or if they didn't communicate a plan, or if they did and Reuters chose not to include it in the article.
Not everything has to be a culture war. When we are talking about our children's future it would be cool to do so pragmatically.
Eh, if the politician thinks this is the clarity of language necessary to send the message, I think that’s fine. Studying this and that can come later. It’s not like anyone is banning that research.
And I guess that's a big part of my frustration. I don't know what the politician actually said. I don't see any link to the/an official statement in the article.
I'm just an old man shaking fist at clouds.
I'm sitting on a mountain of evidence (n=44,000) that used in a very specific way and context AI accelerates and improves lasting learning outcomes. Th3 data is new, but the science that explains it actually goes back decades, predating AI - it's based on pedagogy from texts such as How People Learn (NRC).
My data also shows that students using AI the wrong way perform way worse - the performance gap is widening between students who want to learn and struggle (and use AI to optimize struggle) and students who want instant gratification and use AI for shortcuts.
So I know that if they truly slammed the door on this potential then they threw the baby out with the bath water.
But I don't know the truth because Reuters doesn't report the truth, and that's what tips me from concerned to frustrated. But I guess by complaining about modern journalism standards in a thread about banning AI I'm breaking HN guidelines. Time for me to log off...
> but helps me learn things much more quickly than reading textbooks
Have you tested this against an external metric of competence? The research seems to show that AI is great at making you feel you know something. But I think the studies looking at language learning found those using AI extensively tested below peers using traditional methods.
I don't care what the studies say. It's an incredibly good tutor.
AI helps me fill gaps in my knowledge quickly rather than hunting around for hours for exactly the right chapter which kind-of-but-not-quite explains the concept I am hunting for.
It feels like a good tutor. If you aren’t externally benchmarking your comprehension, you really can’t say.
> helps me fill gaps in my knowledge quickly rather than hunting around for hours for exactly the right chapter
Have you considered that learning to phrase your questions is part of indexing and thus learning a subject?
I’m not saying AI can’t help with that search process. But we have no evidence it helps and lots of evidence it hurts, and everyone with anecdotes to the contrary seems to be going off vibes around how much they learned without any external reference.
Be careful. Speed of learning isn't necessarily the goal. Durability is another metric. I learn more quickly with LLMs, too, but it's certainly questionable if that learning is as durable or deep as learning through struggling with a book.
The students of lowly-rated profs had better 10-year outcomes than those with highly-rated profs according to a study that I think came out of the Naval Institute a decade or two ago. "No shine without friction."
We need more data. Certainly turning students loose with AI stunts them. There's probably some happy medium. But where kids need the most practice with fundamentals when they're young, a blanket ban for now seems sensible. And it also seems like a good plan to introduce it when they get older. I suspect we'll learn a lot from this Norwegian experiment.
Yes, but (I assume) that you are old enough to have already gone through elementary school and perhaps further and learned and internalized a model for learning and retaining information.
Oddly enough, I never was able to learn things from textbooks, except in the context of a traditional classroom lecture course. I've also met maybe one or two people in my life who were able to learn the subjects of my college majors -- math and physics -- at any level from a textbook alone.
> Do you delegate all opinions to only those which have studies?
For policy decisions around something like education standards, something for which we have an established status quo (which works in Norway)? Yes. I don’t think waiting for evidence to act is imprudent in that situation.
> Pupils from first through seventh grade, aged 6 to 13, should as a general rule not be using AI, while those in lower secondary school, aged 14 to 16, can cautiously adopt tools under teachers' supervision, the government said.
Sounds right to me. Kids under 13 need to learn to read, write and comprehend text. Generative AI is not going to help them with those skills.
They can play with AI at home, and after 13 they can learn how to use AI productively and, ideally, in a way that enhances rather than detracts from their education.
Also from the story:
> Facing a broad decline in education test scores, the government in 2024 banned smartphones from schools and has given teachers back more powers to enforce discipline in the classroom.
A big hooray for that. Will be interesting to see what impact that has on Norway education - a quick search just now didn't turn up any detailed studies, presumably those will show up eventually.
You mean chatgpt style AI won't help them with those skills?
If a human parent or teacher can help with skills like reading, an AI system can too, once it's trained and designed to do so. (How good are humans at teaching reading anyway?)
Norway spent two decades digitizing classrooms and is now unwinding it. Seems a bit shortsighted and reactionary although I think they are trying to do the right thing.
I feel like two decades of data that it doesn’t work seems the opposite of shortsighted and reactionary.
Kids can read, write, and comprehend text at 8. I don’t even like LLMs and I’m against this mess. Imagine having regulations rolled out when we were 8 saying “you can’t use the internet!” And I was running my own websites by 10 years old.
Let’s stop pretending this tech is as interesting as we wish it was. If we want to ban models in school, ban laptops/chromebooks with internet. I don’t see the difference at this point.
> Kids can read, write, and comprehend text at 8
A sizable portion of the US adult population effectively can't read, write and comprehend text.
https://nces.ed.gov/surveys/piaac/2023/national_results.asp for 2023:
> Between 2017 and 2023, there were increases in the percentages of adults performing at the lowest proficiency level (Level 1 or below) in both literacy and numeracy: in literacy this percentage increased from 19 to 28 percent and in numeracy from 29 to 34 percent.
The literacy proficiency levels section on https://nces.ed.gov/surveys/piaac/measure.asp describes what Level 1 means:
> Adults at level 1 are able to locate information on a text page, find a relevant link from a website, and identify relevant text among multiple options when the relevant information is explicitly cued. They can understand the meaning of short texts, as well as the organization of lists or multiple sections within a single page.
28% of US adults are just at or below that level.
>A sizable portion of the US adult population effectively can't read, write and comprehend text.
Yes and AI isn't to blame for that. It's the governments, schools, teachers, parents, teacher's unios, who taught them(or more accurately didn't teach them) and graduated them out of school anyway regardless just so they don't look bad in statistics. Sorry but if you graduate people out of high school who can't read you should be trialed for fraud. Simple as.
People blaming AI for adults unable to read puts us back to the 90s when Doom was to blame for school shootings or back to 60s when rock music was to blame for juvenile delinquency, all of them being wrong, and they're wrong here too. People always want to blame a third party external scapegoat that isn't' the parents and isn't the government, for the problems of their kids.
AI hasn't had a chance to demonstrate if it helps or hurts education yet.
That's the big problem with education in general. If you introduce a new factor to children's education you can't realistically measure the effect it has had for about five years, because you need to wait for a cohort of kids to go through that system and then see how they did.
This means that if you introduce something with clear negative effects it will be five years before you spot them!
That's pretty catastrophic given that ChatGPT only emerged in late 2022 and only got good around early 2024.
Nobody is blaming AI. The point is we don’t have the luxury of throwing nonsense at our kids when they’re illiterate. Particularly not nonsense where all the evidence shows it harms on average more than it helps.
Right, AI isn't to blame for that, but cell phones might be? The bad number increased from 19 to 28 percent between 2017 and 2023.
Someone always finds a way to shit on the US. Every single time.
When quoting a factual statistic is "shitting on the US", you're losing the ability to address issues.
The US is a context that is generally relevant to HN, and for which we have lots of data.
Literacy is a worldwide problem.
In this case it's the US that's shitting on the US. These numbers don't compare the US with other countries, they compare the US in 2023 with the US in 2017. And the numbers are from the US government National Center for Education Statistics.
Price of ruling the world I guess
From what I can tell the average school at best can aspire to teach kids how to work and how to socialize. That's it. I'd personally be very happy with computers mostly going away from school too. Most actual learning and exploring will hopefully happen at home.
Many kids in Norwegian schools do not speak Norwegian or English. Kids need "computers" just to translate what other kid is saying.
This is really not true. Kids do actually learn a lot in school - includong weak students. And you actually see huge difference between places with and without schools.
At eight they have limited comprehension of the world around them and limited language skills. They need a lot longer to develop those in tandem.
And also you may be above average there.
When I was 8, I could use the Internet at school but every website was whitelisted.
If you think an 8 year old can comprehend text at the same level of a 13 year old (or an 18 year old for that matter), I don't know what to tell you. Reading comprehension doesn't peak at 8.
Yes, there's also a push to ban all computers in classrooms because data is showing that it's of no benefit and if anything is a negative effect on education.
I think this is basically right. You don’t hand out calculators before kids understand arithmetic. LLM version is sneakier because skipping the work still produces something that looks finished.
A calculator gives you an answer. An LLM gives you an answer that sounds like it already checked itself.
Not quite. The LLM gives you a statistically probable sounding token stream. The calculator gives you a qualified answer within documented and deterministic limits of the device.
No one knows how to use either.
Im confused, are there tasks given to 6 to 13 year old to use AI?
In the classroom, are they just throwing gpt in front of them? Is that the modern equivalent of watching a vhs?
Or do they have homework to vibe code something or given some prompts to ask at home and save somewhere?
Serious question, what does this mean?
There where no clear rules on the matter. Now the PM has given some guidance to schools so they know when they should use it and not.
If there are no guidance teachers and schools can do what they want and some teachers would probably go to far to early
Well one can no longer search for information in the big search engines without it just giving you the answer.
This ruins “search and topic and write about it”
I guess Norwegian schools will have to use smaller / alternative search engines now?
Teacher uses AI, creates lesson plan -> AI creates assignments -> Teacher gives assignments to students -> Students use AI to do assignment -> Teacher grades with AI -> Principal uses AI to monitor teachers progress.
This is happening at schools nationwide. It is unstoppable at this point. It's a bizarre charade.
You forgot the parasite companies that sit between the teachers/student/principal which pretend they detect the big bad AI assisted/generated work to punish individuals using AI (with a majority of false positives and always late by 1 generation of models). The charade wouldn't be complete without rent seeking intermediaries.
This isn't about teachers using Ai to make their life easier, otherwise it wouldn't only apply to elementary school, and from what I read it applies to students.
Students likely aren't allowed to use AI anyway for assignments. Or are they? That's the question, what is actually being banned if anything
It speaks to the nonsensical structure of the education system, that it functions just fine having optimized out the education.
Nobody seems to care enough to do anything about it.
AI used to write homework should be banned.
AI in 1:1 tutor mode with proper hardware (live scanning pen and paper), harness and guardrails should be wildly successful (in terms of education outcomes) especially in elementary school.
Disagree. AI has no business being used in 1:1 tutor mode before the hallucination and sycophancy issues are completely resolved. As is, I can easily see it being a hindrance to building actual understanding.
Just one example - it's very common to see ChatGPT and the like respond with "you're absolutely correct! Great insight" to something that is a complete misunderstand.
Just realized 1:1 AI is 90s self-esteem medals-for-everyone parenting on steroids.
Teachers hallucinate too. I’ve had creationists and communists and tin-foil-hat (chem trails, 5g, etc) teachers. Surely you can imagine an AI tutor that is higher than zero ROI.
> I’ve had creationists and communists and tin-foil-hat (chem trails, 5g, etc) teachers.
I certainly have, too, but there is still a difference between a person who has a factually incorrect but consistent worldview and an LLM which simply reflects the worldview of the user or even changes between queries.
I don't think creationists have any business being in schools either, for what it's worth, but I think it's easier for a teenager to sort out "Mr. Smith has no clue what he's talking about" vs "I have no clue what's true because the LLM everyone expects me to learn from just confirms everything I ask regardless of what I'm asking".
A bit part of education is (should be) independent learning with textbooks and reading. You don't need to be "tutored".
That’s rather disingenuous. But it seems nowadays that words have lost meaning… so, I don’t blame you. I blame the LLMs for this deterioration.
fwiw, Alpha School is the supervised version. the New York campus is $65k/yr and not legally a school.
private school money with homeschool paperwork and an app doing the teaching.
https://www.wired.com/story/alpha-schools-new-york-city-camp...
I think AI should be used in higher level schools but with the added requirement that the output will be held to a much higher standard and that it's fact checked. Teach the students to use AI to reach a higher level while mitigating the inherent issues like hallucination and sycophancy.
> AI in 1:1 tutor mode with proper hardware, harness and guardrails should be wildly successful
I’m open to the idea! Show me the evidence. Then we can roll it out to our kids.
Glad you’re asking. https://cepr.org/publications/dp21577
“AI adoption raises homework scores by 18% and reduces completion time by 30%, but lowers monthly exam scores by 20% within six months. High-stakes entrance-exam scores fall by 18 and 24%, with the full penalty emerging only after about two years.”
Yup. Short-term metrics juice. Actual comprehension and cognition falls. This seems to be the case across the board, including with adults.
I’m genuinely optimistic that there is a way to make AI helpful in education. I just don’t think we’ve found it yet. (We certainly haven’t demonstrated it.)
> reduces completion time by 30%
This is probably the big problem, or at least one of them.. If you use less time on learning, it will probably be harder to remember what you learned also. We need to spend some time to make it stick
The behavioral issue I see is that LLM users tend to immediately reach for an LLM and do their thinking in concert with it.
This tempts users to approach problems by first feeding them into the LLM and then simply following the route the LLM lays out, which does improve task completion time for tasks that the LLM can simply regurgitate, but it stops the user from developing the actual critical thinking skills that school is supposed to teach.
Can agree on that.
The description of the paper also said:
AI users who maintain similar homework completion time as non-AI users experience small learning losses.
This was a surprise too me. I would have thought otherwise.
Would love to see some evidence about if more or less people fall behind and have worse results. In my head the AI should be able to get the weakest students a bit highere.
> In my head the AI should be able to get the weakest students a bit higher
I think the evidence so far is all students lose learning and cognition, but the brighter students lose less.
I think AI could (and by some students probably already is) be used to help a student better understand the material, and faster than you could before. I still recall some parts of Physics taking a while to click, and often having to reread different sections of a textbook to try and understand the what and why behind something.
The biggest issue is a child has to want to do that, since they also could just ask the AI for the answer and then go back to playing video games. End of the day past age 13 or so I just don't see legislation making any difference, they'll find a way past any law blocking them from using AI. Like a lot of education it'll probably come down to parenting.
> I think AI could (and by some students probably already is) be used to help a student better understand the material, and faster than you could before
I think so too. But we haven’t demonstrated we’ve found how, in kids or in adults.
> biggest issue is
We genuinely don’t know what the biggest issue is. We just know it doesn’t work. There is zero quality evidence for AI helping with learning or cognition in kids or adults. (Happy to be proven wrong. This is a fast-moving and big field.)
> they'll find a way past any law blocking them from using AI. Like a lot of education it'll probably come down to parenting
And community. Rich towns restrict devices in school, monitor use at home and thus will have less of an issue with AI exposure.
>I think so too. But we haven’t demonstrated we’ve found how, in kids or in adults.
Ask chatgpt or claude, on their highest model (probably unnecessary but I'm sensing a vibe) to explain a simple linear algebra problem, and if you don't understand it, ask about what part you don't understand.
And if you truly believe it made something up, prove it.
This is seriously the easiest thing to prove out there, you can see for yourself in the next 5 minutes.
> And if you truly believe it made something up, prove it.
You seem to be assuming that the issue is around factual correctness, and that may be the case but the evidence we have so far doesn't support jumping to such a narrow cause.
Is the poor performance because the LLMs are frequently wrong? Unknown.
Is it because the LLMs are sycophantic? Unknown.
Is it because the chat interface is a poor one for learning? Unknown.
What we do know is that students who rely on LLMs learn less and perform worse in the long term. And that alone is enough evidence to support a ban. If better tools come along in the future and are shown to aid learning, then the ban can be re-evaluated.
Sure. I already know linear algebra. If it’s a new branch of mathematics, this is a terrible way to learn.
Again, the research points almost exclusively in one direction when it comes to learning and cognition around AI. You’ll solve more problems more quickly but wind up learning and thinking less.
Honestly, what do you think a teacher does that an llm can't? If you want to learn you absolutely can ask an llm how to _solve_ x and explain the steps and why.
My leaving out the word solve seems to have led some of you astray, I apologize.
Again the problem is you have the option to solve your problem and move on without understanding it. That does not mean you can not use the tool to understand the problem and how to _solve_ it.
I live in fear that instead of learning how to use the tool, some might just vote to ban the tool.
> what do you think a teacher does that an llm can't?
I don’t know! It’s an interesting question. All we know is it does.
> That does not mean you can not use the tool to understand the problem and how to _solve_ it
It doesn’t. But we have no evidence it can.
We have lots of evidence of people thinking they’ve learned something, taking a benchmarking test, and being found wholly deficient compared to folks who worked through a textbook, went to a class or even solved problems off YouTube videos or instructional websites.
> to explain a simple linear algebra problem, and if you don't understand it, ask about what part you don't understand.
The goal is not to understand a linear algebra problem. The goal is to learn how to solve it using lessons and techniques taught beforehand. Aka not to get a fish, but learning how to fish.
I'm sorry the wording of my post didn't match what you wanted.
Type in "Explain how to solve a simple linear algebra problem" into the AI of your choice instead.
> Type in "Explain how to solve a simple linear algebra problem" into the AI of your choice instead
I’m more interested in seeing how someone who teaches themselves with this approach scores on a standardized exam of linear-algebra competence.
A crucial part of learning is struggling with understanding and overcoming problems by yourself. AI removes that part.
>AI users who maintain similar homework completion time as non-AI users experience small learning losses.
Seems like there's no benefit even if it's used "correctly"?
Care to give us the bits you found interesting in the paper to spare me plonking down £6?
Would hate to dissect this just off a paragraph.
Considering that the paper concludes that even students who take the long approach and use LLMs in the most appropriate way for learning still retain less over the long term than students who simply don't use LLMs, I think it's likely they didn't read the paper in the first place.
We thought the same of electronic devices in general and digital learning content specifically. In actual practice both result in lowered test scores and declining critical thinking skills.
Idk why you screeching AI touts are so confident about its ‘wild’ success in all areas given absolutely zero evidence to that effect.
It’s tiresome.
It's inevitably your fault for prompting incorrectly or using the wrong model.
"You just have to repeat the prompt 3 times and then spin around counter-clockwise twice! That always works for me. You obviously just don't know how to prompt the model correctly."
Every time I see LLM enjoyers yapping on like this, it just reminds me of people trying to read tea leaves. There's all these goofy little rules about how to structure the prompt and how mean or nice to be to get it to work optimally, but I think it's obvious that most of these users are just seeing incidental more successful outcomes in a largely random system and extrapolating from there because it makes them feel in control.
It is, quite literally, superstition.
Instead of prompts, let’s call them incantations.
just ban everyone from using ai completely lets go back to how it used to be
I'm glad this is the case. It is the correct outcome.
I've heard students actually discussing that they will just use an LLM to shortcut work. I even have friends in their 50s who can barely think for themselves now without having to refer to "AI". And at least two of them are teachers.
Leading on from that, the staff are the most dangerous. My daughter has had generated exercises provided to her from multiple teachers, which are quite frankly entirely wrong. This was hilariously pointed out after I called a meeting with her mathematics teacher over it. They questioned my knowledge on the matter with the insane assumption that "AI is foolproof". I had to hit them with a clue stick then.
No one taught anything of value. No one learned anything of value. I am very worried we'll see a lost generation at some point rippling through the ages.
Good, wholeheartedly agree with this. Too bad US legislators are impotent.
US legislators have almost no say in how schools are run. The DoE is a husk, and states call the shots. In my state of Oregon, you will have wildly different curricula and standards depending on your district, because there's almost no state oversight either.
They made a choice - US legislators effectively gave up their control of how schools are run when the conservative coalition allowed the trump administration to dismantle the DoE without congressional approval. Congress itself could legally retain control of education, but if congress refuses to assert that power then it is meaningless.
The end goal is to dismantle public education and route public money to religious and private schools.
Yes, that is what impotent means.
I think it’s quite tricky. On one side, writing is a form of thinking and cognitive training.
Just NOT doing that work by having AI simulate it is not good for anyone’s cognitive development.
At the same time, anyone growing up today will be using LLMs for massive parts of the jobs they grow up to do. So they should learn about it.
I really feel for teachers/educators right now. It must be hard to remain demanding and insist on educating kids well while also preparing them for the future they’ll actually live in.
> anyone growing up today will be using LLMs for massive parts of the jobs they grow up to do. So they should learn about it
Whatever AI looks like in 20 years is going to be so different from what it is today as to make distracting from basic skill-building an almost-certainly net negative educational effort.
I agree to some degree. But by that logic, should kids in the 2000s not have learned about the internet because the internet fundamentally changed between then and now?
I think that if anything, it’s really good I learned how to operate a computer and the Internet BEFORE what the Internet became.
I pity the generation who don’t understand a computer’s folder structure because they grew up with smartphones and TikTok.
> At the same time, anyone growing up today will be using LLMs for massive parts of the jobs they grow up to do. So they should learn about it.
There's not so much to learn they can't put it into a high school course. Adults currently in the workforce haven't been using AI since they were in elementary school, and they're adjusting fine.
I would disagree that they’re adjusting fine. So much of the stuff we see now is full AI slop clearly created with the first output of ChatGPT. It’s like saying we don’t need to teach kids about the internet because older generations grew up without it and they’re scrolling TikTok.
a) what’s the actual percentage of professionals who actively use AI? It’s much smaller than we think in tech.
b) what percentage of those people understand the very basics of how LLMs work (e.g. token prediction, context windows, etc)
c) what percentage of those people understand AI Agents (or any of their ingredients (APIs, credentials, etc.)
You quickly arrive at a tiny fraction that has a real clue about what they’re doing.
> anyone growing up today will be using LLMs for massive parts of the jobs they grow up to do. So they should learn about it
Essentially the entire value proposition for AI, particularly as it advances, is that you don't need to learn how to do things anymore.
>using LLMs for massive parts of the jobs they grow up to do
These are elementary school kids...if they start using AI in 6th grade, they have 6 years to learn AI before graduating high school.
Fair point!
Surprised to read this as Norway also has Sikt AI for schools, where teachers can monitor how AI systems are used. Seems like it has both embraced AI and banned it.
> Seems like it has both embraced AI and banned it.
Indeed, seemingly they done so by age/educational progression:
> Pupils from first through seventh grade, aged 6 to 13, should as a general rule not be using AI, while those in lower secondary school, aged 14 to 16, can cautiously adopt tools under teachers' supervision, the government said. In upper secondary education, from ages 17 to 19, students should learn to use AI appropriately so that they are prepared for further education and work, it added.
I wonder if the struggle is really comprehending thoughtful selective adoption.
Schools are the place where the product is a more fully developed person. There's no LLM shortcut for generating that. There are many ways you could use LLMs that would discourage it. There may be some that can encourage it.
Personally, I can see aggressively keeping kids away from LLMs until they've learned effort, living in tension/frustration, the pleasure of breaking through to discovery, trust evaluation, hypothesis/test cycles, and good socratic dialogue from the learner's side.
It may be possible at intermediate phases to prime some models to help with this process.
Social media execs are already known for keeping their children off their platforms* and even phones so my question is: Do the leading ML/AI people let their children interact with LLMs yes or no?
[*] https://www.cnbc.com/2025/12/13/youtubes-ceo-is-latest-tech-...
Broadly speaking, the wealthier the parents the more restrictive device is in general in the household.
This used to be a tech/non-tech line. It shifted to class sometime over the last ten years. The iPad kids are probably getting served slop. The AI-employed parents don’t have to directly police AI exposure because their kids’ device use is already controlled; at school, at extra-curriculars and at home.
The question is do the allow them to use their laptops to access llms. Like I assume before ai they where not allowing the kids to fry their brains doom scrolling but would allow programming. They weren't blocking all tech as they aren't raising subsistence farmers.
> I assume before ai they where not allowing the kids to fry their brains doom scrolling but would allow programming
Young kids don’t need to learn programming. They need to learn math, reading, spatial reasoning and social skills. (Among other things.)
The kids who were being taught Java in elementary school ten years ago aren’t particularly better off for it.
I figure they let them use Wikipedia, google etc. before though so I am curious where the line exists now.
there is a difference in a kid asking a LLM to explain 0! or 0^0 and to ask it to be its friend.
The difference is between asking to explain x^y and reading and understanding the why and maybe go through a test, vs asking to do my homework for me and then go back to video games quickly. If the whole idea of using LLMs is to reduce effort on the part of the pupil, you are probably holding it wrong.
"..ban on smart phones.." "..ban on gen AI.." "..ban on social media.."
yes YES YASSSSSSSSSSSS
Ban all the things for kids. I don't want to be interviewing people in 10 years and decline every candidate because they can't correctly answer the question "You are 50ft away from the car wash. Do you walk or drive?"
To be fair, norwegian kids would probably fail because they don't have a clue how much 50 feet are.
> To be fair, norwegian kids would probably fail because they don't have a clue how much 50 feet are.
I... are you an LLM? The distance doesn't matter - you probably don't want to walk to the car wash.
The question is slightly vague (since I could be going there not to wash my car), but I'm pretty sure in the intended interpretation, the actual distance is irrelevant. :)
Got'em! Yes, this one right here officer; that's the clanker!
The traditional schooling system can't stand that they are being outcompeted by AI and are trying to use the government to maintain their monopoly and keep those tax dollars flowing.
> The traditional schooling system can't stand that they are being outcompeted by AI and are trying to use the government to maintain their monopoly and keep those tax dollars flowing.
Learning is a conspiracy by Big Knowing, it's all a myth. Let's just ask an LLM to all our thinking, no need to be a functional human.
LLMs can teach better than the human teacher could. Big Knowing wants to be the one to teach, but that is not the only option.
Yep… I mean scientific studies seem to suggest that you are completely wrong. But big science is all a conspiracy against those AI companies that work so hard for the betterment of mankind after all.
I believe my eyes more than studies. I have learned so much, so fast through the process of talking with LLMs.
I already saw it in my life: a ban on calculators, a ban on computers. But after a short period of rejection, everybody starts to embrace the new tech. Instead of bans we were getting computer classes in schools.
> after a short period of rejection, everybody starts to embrace the new tech
We’re banning cell phones in school after seeing the evidence, albeit along a class gradient. We’ll probably see something similar with AI. Poor kids get AI in school (and unmonitored at home). Rich kids do not.
I did mathematics, and a fair bit of computer science on the side. This was nearly all on paper, without computers. And I'd go a week or two without a calculator sometimes, mostly employing it when I had a bad hangover.
Do we really need to force technology into everything or are we just used to doing it so see it as necessary?
Great, yet another "no child and no teacher left behind!"
I was gifted kid, bored to death at basic school. I was reading books under the table, and was lucky to have tolerant teacher. Total ban would just push me to misbehaving and disrupting the class.
AI is amazing tool for learning, if Norway can not harness it, there is something deeply wrong with the educational system. Perhaps teachers unions?
Norway has a big problem with young immigrant kids at school not speaking Norwegian. Right now other kids are expected to teach them basic language, holding back their own development (like learning reading and writing)! Again, AI could provide amazing help here!
So I take it you have some expertise in teaching? At what level? K-12? Middle school?
What, in your professional opinion as an educator, should schools do about AI in schools?
I would allow schooling in other languages in primary schools in Norway. Basic education is a human right, and it is not right to force kids out of their mother tongue. Insisting on some obscure language is just misplaced nationalism!
AI could help with that.
I wonder if at some point we will look back on stuff like this as back in the 1990s schools banning internet research and search engines. Obviously that seems ridiculous now, but the internet was big and scary back then too.
Those bans were implemented without evidence. We have evidence AI exposure reduces learning and cognition. There are probably situations where it enhances it. But we haven’t delineated those yet, and so shouldn’t be rolling out a half-baked system more likely to hurt than to help.
what is the evidence
Were schools banning search engines? I remember teachers recommending Dogpile (because it would combine search engines), and we did some computer lab stuff in the 00s, but there was no one saying that we should ignore search or the internet altogether.
In middle school I remember being assigned a book report that would include the author's biography. I'd just finished a book (The Gammage Cup) and of course my local library did not have any information about the author. So in that situation it was assumed that you would learn traditional research methods, but also that you would just pick a classic book where the information was readily available.
Few schools allow unrestricted internet access these days.
The general question here is risks vs rewards and with any new technology both are unknown making caution perfectly reasonable be that internet searches or anything else.
So sure in 30 years the policy will look different, but that doesn’t mean they are making the wrong decision.
Wouldn't it matter a lot how AI is being used?
I would assume if children are allowed to use AI without rails as a shortcut it will undermine their learning, and it's used for feedback and as a patient tutor it would accelerate their learning?
It seems like the problem is that they don't have the science and tooling to use it constructively at scale, so the desperate solution is to ban it outright until a scalable constructive approach is understood?
The article doesn't explain any of this directly...
It's frustrating to me when bold statements are directed at "AI" holistically and vaguely, completely ignoring any nuance.
There is a massive gap between letting elementary students free reign use chatGPT 3.5 (hallucinations and all) to do whatever, vs using a very guard-railed pedagogically optimized app powered by a SOTA model to support students in a specific way that accelerates good outcomes.
Most respectful interpretation is that the leaders know this and have a plan to figure it out, but for some reason it's not making it's way into this article. Is the absence representative of the truth of the situation, or some editors choice to pile on to a holistic anti-ai narrative?
> it's used for feedback and as a patient tutor it would accelerate their learning?
We have mounting evidence AI hurts learning and cognition in many circumstances. I have not yet seen similar-quality evidence for it helping.
Given that balance, restricting AI in education in the general population (while studying how to best deploy it) seems prudent. Especially given the Norwegian approach, which gradually introduces AI as kids get older.
I'm actually not really criticizing the decision so much the article and communications around it. If student learning outcomes are crashing and they desperately want to turn them around I understand why they would take dramatic action.
Giving students uncontrolled access to generic LLMs probably would hurt outcomes. Research process is slow (IRB and all that) so they are dealing with data from years ago (models that confident hallucinated a lot more than current SOTA) so if thats what they are basing it on its reasonable.
My frustration isn't with the decision (hey all teachers - no more chatGPT in the classroom). My frustration is with the reporting / nuance of "until we can research this better and figure out how to harness AI to improve outcomes and not undermine them".
> the reporting / nuance of "until we can research this better and figure out how to harness AI to improve outcomes and not undermine them"
It’s balancing the irrationally exuberant narrative of the tech bros and AI pushers. You have to stop the bleeding before you can dress the wound to promote healing.
I don't think one balances a lack of nuance with more lack of nuance.
One avoids nuance for clicks or to propagate a narrative, sew division, distract, etc.
Again. As I said in both my comments, I'm not criticizing the ban, I'm criticizing the absence of any communication regarding a plan for researching potentially constructive uses. As a reader, I can't tell if the Norwegian leaders have no plan, or if they didn't communicate a plan, or if they did and Reuters chose not to include it in the article.
Not everything has to be a culture war. When we are talking about our children's future it would be cool to do so pragmatically.
Eh, if the politician thinks this is the clarity of language necessary to send the message, I think that’s fine. Studying this and that can come later. It’s not like anyone is banning that research.
And I guess that's a big part of my frustration. I don't know what the politician actually said. I don't see any link to the/an official statement in the article.
I'm just an old man shaking fist at clouds.
I'm sitting on a mountain of evidence (n=44,000) that used in a very specific way and context AI accelerates and improves lasting learning outcomes. Th3 data is new, but the science that explains it actually goes back decades, predating AI - it's based on pedagogy from texts such as How People Learn (NRC).
My data also shows that students using AI the wrong way perform way worse - the performance gap is widening between students who want to learn and struggle (and use AI to optimize struggle) and students who want instant gratification and use AI for shortcuts.
So I know that if they truly slammed the door on this potential then they threw the baby out with the bath water.
But I don't know the truth because Reuters doesn't report the truth, and that's what tips me from concerned to frustrated. But I guess by complaining about modern journalism standards in a thread about banning AI I'm breaking HN guidelines. Time for me to log off...
> sitting on a mountain of evidence (n=44,000) that used in a very specific way and context AI accelerates and improves lasting learning outcomes
Can you point to it?
AI makes me worse at programming but helps me learn things much more quickly than reading textbooks. Both can be true.
> but helps me learn things much more quickly than reading textbooks
Have you tested this against an external metric of competence? The research seems to show that AI is great at making you feel you know something. But I think the studies looking at language learning found those using AI extensively tested below peers using traditional methods.
I don't care what the studies say. It's an incredibly good tutor.
AI helps me fill gaps in my knowledge quickly rather than hunting around for hours for exactly the right chapter which kind-of-but-not-quite explains the concept I am hunting for.
> It's an incredibly good tutor
It feels like a good tutor. If you aren’t externally benchmarking your comprehension, you really can’t say.
> helps me fill gaps in my knowledge quickly rather than hunting around for hours for exactly the right chapter
Have you considered that learning to phrase your questions is part of indexing and thus learning a subject?
I’m not saying AI can’t help with that search process. But we have no evidence it helps and lots of evidence it hurts, and everyone with anecdotes to the contrary seems to be going off vibes around how much they learned without any external reference.
Be careful. Speed of learning isn't necessarily the goal. Durability is another metric. I learn more quickly with LLMs, too, but it's certainly questionable if that learning is as durable or deep as learning through struggling with a book.
The students of lowly-rated profs had better 10-year outcomes than those with highly-rated profs according to a study that I think came out of the Naval Institute a decade or two ago. "No shine without friction."
We need more data. Certainly turning students loose with AI stunts them. There's probably some happy medium. But where kids need the most practice with fundamentals when they're young, a blanket ban for now seems sensible. And it also seems like a good plan to introduce it when they get older. I suspect we'll learn a lot from this Norwegian experiment.
Yes, but (I assume) that you are old enough to have already gone through elementary school and perhaps further and learned and internalized a model for learning and retaining information.
I think that is what is at risk.
Oddly enough, I never was able to learn things from textbooks, except in the context of a traditional classroom lecture course. I've also met maybe one or two people in my life who were able to learn the subjects of my college majors -- math and physics -- at any level from a textbook alone.
Do you delegate all opinions to only those which have studies?
> Do you delegate all opinions to only those which have studies?
For policy decisions around something like education standards, something for which we have an established status quo (which works in Norway)? Yes. I don’t think waiting for evidence to act is imprudent in that situation.