The case for rethinking classroom tech

92% of Pre-K–8 school leaders strongly support restricting cell phones in school — yet most students still spend 1–4+ hours a day on school-issued devices, largely for testing.

Even leaders who want less screen time are boxed in by how schools test

A 2026 NAESP survey of principals found near-unanimous support for cell phone restrictions, with real reported benefits: more focus, better behavior, fewer disciplinary incidents. Yet 92% of these same schools still issue every student a device — mainly for testing and digital curriculum — and most students spend 1–2+ hours a day on them. The concern is widely shared: 85% of leaders, 81% of teachers, and 74% of parents surveyed all separately flagged it.

Parents support pausing AI in schools by 3 to 1 — and that holds across party lines, education levels, and grade.

This isn't a fringe position — it's a bipartisan majority

A 2026 Institute for Family Studies survey of 2,000 parents found 75% support pausing AI's introduction in schools, with only 25% opposed — strong across party lines and every grade level. Around the same time, New York City and Los Angeles, the country's two largest districts, both moved to pause or restrict classroom AI use, citing similar feedback.

"Stop turning schoolchildren into crash-test dummies."

The man who built the world's benchmark for school quality says we're moving too fast

Andreas Schleicher — the OECD's education director and founder of the PISA test — laid out the case in The Economist this September. American 15-year-olds' scores have fallen sharply since 2018 — 22% now miss the baseline in science alone, with wealthier students falling furthest. OECD-wide reading comprehension dropped by more than a year's worth of learning, driven partly by a shift to skimming and digital reading. Moderate device use for schoolwork tracks with better outcomes, but leisure use in class sharply lowers performance, and students who lean on chatbots score lower and report less motivation. His point isn't to reject AI — it's that we'd never let a new vaccine reach millions of children without rigorous trials first, yet we hand new technology to classrooms with remarkably little evidence of its effects.

77% of voters want government-set guardrails on classroom technology — only 23% support the current hands-off approach.

This isn't a fringe concern — it's one of the most consensus findings in American polling

A 2026 Century Foundation poll of 2,000+ registered voters found every technology concern tested drew at least 80% concern — including 85% worried about AI letting students skip learning the material, and 84% concerned about companies profiting from student data. Support for government rules was strong across party lines: 76% of Democrats, 74% of independents, 80% of Republicans. The concern tracks with real incidents: a 2024 breach at PowerSchool exposed 62 million students' records, and 96% of K-12 ed-tech apps share student data with third parties.

"At this point in its trajectory, the risks of utilizing generative AI in children's education overshadow its benefits."

A year-long global review of over 400 studies reaches a sobering conclusion

Brookings' Center for Universal Education spent a year on a "premortem" — studying generative AI's risks to kids before they play out at scale, rather than after. Drawing on over 500 stakeholders across 50 countries and 400+ studies, their conclusion: AI's risks right now can undermine kids' foundational development in ways that prevent its benefits from ever being realized. Their ask isn't a ban — it's real evidence, guardrails, and accountability, built with educators and families rather than imposed on them.

"Slowing down may ultimately be the fastest path to getting it right."

Going slow doesn't mean rejecting AI — it means using it where it's actually proven to help

Brookings education director Rebecca Winthrop argues districts pausing AI — including NYC and LA — are making the responsible choice, not an anti-technology one. Even Estonia's well-funded national AI program found real tradeoffs after a year: teachers saved time, but nearly three-quarters of them agreed students were turning in work they couldn't explain — outsourced thinking, not learning. Going slow isn't an all-or-nothing ban: AI can stay where it clearly helps — assistive tech, multilingual support, teacher tools — while pausing broad student use until there's a real plan. And 83% of parents believe kids need to think critically without leaning on AI; policy tends to fail when schools and families pull in opposite directions.

Much of today's classroom technology works against how kids actually learn.

The case for evaluating EdTech by learning outcomes, not novelty

Neuroscientist and educator Jared Cooney Horvath argues that much popular classroom technology — adopted with good intentions — actually undermines the deep, effortful thinking real learning requires, and lays out what works instead. A useful, non-partisan frame: judge classroom tools by what they do for learning, not by whether they're new.

"...too big and too integral ... to interrogate."

A Massachusetts district just found out what happens when a school hands its whole operation to Big Tech

Over Labor Day weekend, Springfield Public Schools — 23,000 students across roughly sixty schools in western Massachusetts — was hit by a cyberattack that shut classes for four days: staff lost access to transcripts and medical records, phone lines and security cameras went dark, and students and teachers were locked out of laptops and email. The district's own guidance told teachers to fall back on "books, notebooks, and paper." This happened to a nearby Massachusetts district, not somewhere far away. Reporting on Natasha Singer's book Coding Kids, The New Yorker traces the broader pattern: Google and Microsoft now dominate K-12 tech so completely that opting out is barely realistic, both have faced serious student-data allegations, and the "learn to code" pitch behind a decade of ed-tech spending has already been upended by the AI it helped fund. Even Bill Gates, who a year ago predicted AI tutors could replace most teachers, has since raised his own concerns about AI's effect on kids' critical thinking — even as his foundation commits a billion dollars to expand AI in education.

Source: Jessica Winter, "What We Lost When Big Tech Took Over the Classroom," The New Yorker, September 17, 2026. This piece is behind The New Yorker's subscriber paywall — cited here by title, author, and date rather than a link, since the archived copy we had access to won't stay live.

"We aren't sitting at the table. We're the meal."

The same playbook Big Pharma used on doctors, now aimed at teachers

In reporting behind Coding Kids, Natasha Singer describes a Microsoft-led "AI training" for 200 NYC teachers — product demo, Minecraft-branded video — that struck her as strikingly similar to pharma's marketing seminars for doctors. A Georgetown pharmacology professor summed it up: industry funds what helps industry. It's not just workshops — the AFT, the country's second-largest teachers' union, accepted $23 million from Microsoft, OpenAI, and Anthropic to train members on AI, a deal even its own president acknowledged created tension between student interests and companies "in it for profit." Even Microsoft's president admitted the public sector had failed to fill this role itself. The scale: Chromebooks are 59% of mobile devices shipped to U.S. schools, and Google has paid $170 million to settle complaints that YouTube illegally harvested children's data. Some districts are walking devices back — LA eliminated Chromebooks and iPads for pre-K through Grade 1 — and some educators are teaching "technology civics" instead: asking who built the technology, who profits, and how it affects them.