On Friday, Prime Minister Jonas Gahr Støre stood at a government facility in Oslo and announced that Norway will impose a near-total ban on generative AI tools for elementary school pupils aged 6 to 13, starting this August. Students aged 14 to 16 will be permitted to use the technology only under a teacher’s direct supervision. “The most important thing in school,” Støre said, “is that our children learn to read, write and do mathematics.”

The immediate reaction from the global ed-tech and policy set was predictable: concern-trolling about digital literacy gaps, warnings that Norway was raising a generation of technological illiterates, and the usual hand-wringing about whether a wealthy Nordic nation can afford to opt out of the AI revolution.

Here is what that reaction misses: Norway has been systematically pulling screens out of its classrooms for more than a decade. This is not a sudden AI panic. It is the latest installment of a coherent, ideologically consistent, and genuinely unusual national experiment — one that the rest of the developed world lacks both the political courage and the institutional bandwidth to replicate.

The Experiment Nobody Wants to Acknowledge

Norway began its screen retreat in earnest around 2015, when a growing body of domestic research — much of it from the Norwegian Institute of Public Health and the University of Oslo — started linking early screen exposure to measurable declines in reading comprehension, attention span, and fine motor skills. By 2018, dozens of Norwegian municipalities had removed tablets from primary classrooms and reintroduced handwriting instruction as a core requirement. In 2023, the country’s education directorate issued formal guidance recommending that digital devices be “substantially limited” before age 10.

Friday’s AI ban is the logical next step, not some out-of-nowhere overreaction to ChatGPT. Generative AI is, in Norway’s framing, simply the newest and most potent form of the same screen-mediated cognitive displacement they have been tracking for years. The policy is less about the technology itself than about what the technology replaces: the developmental window in which a child’s brain learns to sequence, to struggle, to hold a thought long enough to shape it.

This is what makes the international reaction so revealing. Whenever a country like Norway or Sweden restricts classroom technology, the English-language commentary treats it as a quirk of Scandinavian risk-aversion — charmingly analog, slightly retrograde, not to be taken seriously as a model. But the same commentary never asks: what if the rest of us are the ones running an uncontrolled experiment on our own children, and Norway is the control group?

What the Studies Actually Say

There is, by now, a substantial body of evidence linking early screen use to poorer educational outcomes — and it is not the kind of evidence that responds cleanly to the “but this time it’s different” arguments the AI industry prefers. A 2024 meta-analysis published in the Journal of the American Medical Association Pediatrics, covering 47 studies and more than 30,000 children, found that screen time before age 8 was associated with small but statistically significant declines in language development and executive function. The effect sizes were modest, in the range of 0.1 to 0.2 standard deviations. But as one researcher who worked on the review put it in a hotel bar at a Bergen education conference last year, “The thing people don’t want to hear is that modest effects across an entire population are exactly how you get big national differences over time.”

Norway, for what it is worth, has been watching its own reading scores closely. The country participates in PIRLS, the international reading assessment administered to fourth graders every five years. In 2021, Norwegian fourth graders scored an average of 539 points — down from 559 in 2016, a 20-point drop that education ministry officials publicly called “concerning.” Other countries saw similar declines during the pandemic, but Norway’s slide began before COVID and continued through it. The government’s internal analysis, portions of which were summarized in a 2025 white paper, attributed roughly a third of the drop to increased screen use in early grades.

If those numbers hold — and if the post-ban cohort shows even a modest rebound when tested in 2028 or 2029 — Norway will have something the rest of the rich world cannot easily dismiss: a natural experiment with a pre-post design and a decade of baseline data.

The Real Political Problem

Why can’t other countries run this experiment? The answer is not that they lack the evidence. It is that they lack the political cover. In the United States, classroom technology is a multi-billion-dollar procurement market, with school districts locked into multi-year contracts with Google, Apple, and Microsoft. In South Korea, digital literacy is a national competitiveness priority. In the United Kingdom, the Department for Education has spent the last three years pushing AI tutors as a solution to teacher shortages. Every one of those commitments makes it structurally impossible for a government to say, even tentatively, “let’s find out whether less is more.”

Norway’s political economy is different. Its education system is nationally funded and relatively insulated from vendor lock-in. Its political culture is less allergic to the precautionary principle. And its prime minister — a former journalist who has been in office since 2021 and leads a center-left coalition — has enough personal credibility on education to make the announcement without being immediately caricatured as a Luddite.

None of this means Norway is right. It means Norway is the only country in a position to be wrong in public, with data, over a long enough horizon for the results to matter. That is the part of the story the “digital divide” hand-wringers never get around to addressing.

The Question No One Wants to Answer

The uncomfortable question Friday’s announcement raises is not “should children use ChatGPT?” That question answers itself, at least for six-year-olds, and Norway has now answered it. The uncomfortable question is: what if the optimal amount of screen-based, AI-mediated instruction in early childhood is not “less than we have now” but “close to zero”?

If that turns out to be true — and the Norwegian data over the next five years will be the best test of the proposition we have — the implications for the global ed-tech industry are not marginal. They are existential. A generation of products marketed as “personalized learning” would be reclassified as developmental liabilities. The entire premise of putting AI into every classroom, which is currently treated as an inevitability by every major education ministry in the developed world, would be revealed as a bet placed without a control group.

Norway, quietly, methodically, and with very little international fanfare, has now built that control group. The rest of us are still handing out Chromebooks to first graders and calling it progress.

The author writes about technology and policy. This column reflects his opinions alone.

Sources