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.
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.
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?
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.
[*] https://www.cnbc.com/2025/12/13/youtubes-ceo-is-latest-tech-...
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.
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.
The end goal is to dismantle public education and route public money to religious and private schools.
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?
This ruins “search and topic and write about it”
If there are no guidance teachers and schools can do what they want and some teachers would probably go to far to early
This is happening at schools nationwide. It is unstoppable at this point. It's a bizarre charade.
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.
I’m open to the idea! Show me the evidence. Then we can roll it out to our kids.
Seems like there's no benefit even if it's used "correctly"?
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.)
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.
Do we really need to force technology into everything or are we just used to doing it so see it as necessary?
AI is indeed dangerous. It gives super abilities when in the right hands. Some people don't like it as it creates competition for their mere existence. They start gaslighting campaigns - "AI is bad, dangerous, does not work, consumes too much energy etc". This is luddism of our century, but also a form of psychopathy. When everybody is being gaslit, some of the very same players who spreads false narratives use AI to their own benefit.
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.
Clearly not, given that you seem to believe this despite it being incorrect. Every single bit of evidence gathered so far indicates LLMs are worse teachers than humans or every self-directed learning.
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.
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?"
I... are you an LLM? The distance doesn't matter - you probably don't want to walk to the car wash.
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.
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 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.
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.
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.
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.
And also you may be above average there.
I have two kids and can confidently say eight year olds generally have good language skills, are capable of expressing themselves just fine, and have good comprehension of the parts of the world that they've been exposed to.
No one knows how to use either.
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!
What, in your professional opinion as an educator, should schools do about AI in schools?
AI could help with that.
FYI: all Norwegians learn at least one foreign language in school. It is mandatory. That means you have to, in case that’s a big word for you.
If you so prefer you will have the option to learn one or two more in middle school and one or two more in high school.
By the time I was 15 I spoke three languages. Everyone at that age would know at least two. Some would know four.
Is it so much to ask that you at least consult the AI tools you speak so fondly of before babbling about something you so obviously are deeply ignorant of? If nothing else then at least in lieu of growing a brain?
this tech is unsustainable by design
So, it's no surprise they're going to opt out of a system that's investing trillions to make education useless.
Even if the people building this world are wrong -- not all students are equipped to call some of the wealthiest people in the world complete bullshitters. Not all adults are ready to call them out as bullshitters, for that matter.
Every special event flyer I get from my kids' school now seems to be AI generated. I'd be surprised if quizzes and worksheets don't head the same way.
1. Wisdom through dogma. 2. Wisdom through reasoning. 3. Wisdom through experience.
AI is just stage 1. Instant and easily digestible. Traditional learning forces you to go to Stage 2, because you are often given too much information and you need to compress it to memorise it. And the best way to compress it, is by finding some kind of logical structure in the information.
I learned it as: 1 Knowing what it is. 2 Knowing how it works. 3 Knowing what it can become.
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.
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.
I think that is what is at risk.
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.
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.
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".
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.
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.
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...
Can you point to it?
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
This can be reduced to "Do students have access to a phone". With a 10 year old cell phone camera you can finish homework in seconds. Good luck after 12 with that.
Nobody seems to care enough to do anything about 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
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.
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.
I think the evidence so far is all students lose learning and cognition, but the brighter students lose less.
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 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.
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.
Would hate to dissect this just off a paragraph.
It’s tiresome.
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 successful outcomes in a largely random system and extrapolating from there because it makes them feel in control.
It is, quite literally, superstition.
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 misunderstanding.
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".
private school money with homeschool paperwork and an app doing the teaching.
https://www.wired.com/story/alpha-schools-new-york-city-camp...
It’s a fully centrally controlled technology that reduces your ability and makes you dependent on it to perform all daily and business functions with a huge environmental and economic impact. The economic impact is both the risks imposed by it failing and the risks imposed by it being successful.
It’s not Ludditism, it’s a good attitude to risk.
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.
These are elementary school kids...if they start using AI in 6th grade, they have 6 years to learn AI before graduating high school.
Essentially the entire value proposition for AI, particularly as it advances, is that you don't need to learn how to do things anymore.
Mine couldn’t until they were much older. And I have more so perhaps that’s more statistically valid?
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.
Literacy is a worldwide problem.
Yes and AI isn't to blame for that as adults predate AI. 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.
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.
the report also visualizes not only inter country but also intra country outcomes correlating socio economic influences (age, parents, family migration history, ...) and level of education (school, high school, college and higher) with test outcome (literacy, numerics problem solving)
it also has 10y ago/now comparison.
a trove for the Q "how are we doing, capability wise?"
thanks for pointing to the study!!
[piaacs report] https://www.oecd.org/content/dam/oecd/en/publications/report...
Correlation, causation, and all that
Plus "Generative AI" isn't one single thing. Using it to write your essay is cognitive offloading but using it as a Socratic tutor that gives immediate feedback and adapts to the student is closer to the thing education research says works.
There's an equity angle as well. A school ban doesn't ban AI at home. It bans the equalizing version. Kids in educated, rich households will get AI exposure from parents. Kids without that won't get it anywhere, because the one place where the field is leveled has opted out. If AI fluency becomes a differentiator in the labor market infrastructure which is very likely a 7 year exposure gap sorted by household class is the opposite of what public education is supposed to be for.
(edit: By AI fluency I mean basically knowing how to drive the tools, an intuition for what the tools can and can't do, when to use AI vs doing it yourself, plus detecting when output is wrong, knowing what to verify, etc.)
The problem is, a lot of the parents have bought into the digital parenting age too. They were told ipads etc were part of getting the best education for their kid. Now they're fighting hard on rolling it back (not least because they can't comprehend that it's a problem, that their child can't focus 5 minutes without a device)
[1] https://www.npr.org/sections/health-shots/2024/05/11/1250529... (Article is fine, but more importantly has multiple study links)
Sounds like following the evidence.
And having no TV and no smartphone at home and at school is likely the best way to acquire it.
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?)
Writing developed thousands of years BCE. So, considering we as a species have been successfully teaching our offspring how to read for hundreds of generations, I'd say we're probably pretty decent at it.
this happens constantly, every day. a current implementation of a technology isnt optimal so the entire class of anything related to that technology is treated as equally flawed.
the solution here is better tools, not preventing better tools from being created.
What kids need to learn to read is an adult to engage with them, listen to how they read and engage them on the contents of the book.
For anyone who still thinks kids should use AI, another argument to make is we are still figuring out AI (hence the constant debate on it, hype, uncertainty, boundaries of its capabilities etc etc). I don't think anyone with right mind can disagree with that. Keeping that mind, wouldn't it make sense to at-the-very-least tread with caution when it comes to kids.
A big hooray for that."
I don't suppose they are allowed to use physical violence again, still I would like to know what exactly you are cheering here for?
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.
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.
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.
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.
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.
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.
jenthoven•7h ago
embedding-shape•2h ago
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.
wwweston•1h ago
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.