The PISA 2025 wake-up call: why the answer isn't less tech

The OECD just released its latest PISA results, and there's no way to sugarcoat them. Across 91 countries, 15-year-olds are reading, doing math, and thinking scientifically at the lowest levels this century. Reading scores across OECD countries fell 28 points between 2015 and 2025 — roughly equivalent to a year and a half of lost learning. Math dropped 22 points. One in five students now lacks the foundational skills they need for further education, a career, or just navigating daily life.

These are sobering numbers, and educators deserve to take them seriously. But the conversation that follows matters more than the headlines.

What the data actually says about technology

It'd be easy to look at these results and point the finger at screens. Students report spending an average of 3.4 hours per day on leisure screen time. The share of "hasty readers" — kids who rush through a passage and give a quick but wrong answer — nearly doubled between 2018 and 2025 (from 7% to 11%). The narrative writes itself: phones and apps are ruining our students.

But that's not what the PISA data says if you look closely.

Digital tools used for learning actually improve outcomes — up to a point. Students with limited to moderate technology use for learning outperformed both those with no access at all and those with heavy, unstructured use. The problem isn't technology in the classroom. The problem is technology in the classroom without a plan.

There's a world of difference between a student scrolling social media during a lesson and a student working through a structured, teacher-designed activity that uses AI to deepen their understanding of a text. PISA doesn't make the case against technology in education. It makes the case against undirected technology in education.

The AI finding that should change the conversation

Here's the data point that deserves the most attention: nearly half of students now use AI chatbots for learning. But students who use AI as a shortcut — drafting essays, summarizing readings, looking up answers instead of working through problems — scored about 20 points lower in science. That's roughly a full year of learning left on the table.

The flip side? Students who were taught how to critically evaluate AI-generated information performed better than their peers.

Same tool. Opposite outcomes. The difference was whether a teacher was guiding the experience.

This is the finding school leaders should be talking about in every staff meeting this fall. The question isn't whether to allow AI or ban it. It's whether students are using it in ways that deepen learning or skip it entirely.

Three things educators can do right now

Invest in sustained, challenging reading

The reading decline isn't just about comprehension scores — it's about stamina and synthesis. Students are losing the ability to sit with a complex text, work through it, and form their own interpretation. That skill doesn't rebuild itself through test prep. It comes back through regular practice with real, challenging material and a teacher who's there to push thinking when a student gets stuck.

Get specific about technology, not just restrictive

Nearly half of OECD schools have now banned phones on premises, and the data suggests that helps with distraction. But a phone ban by itself doesn't teach students how to use digital tools well. Schools need technology that's built for learning — with structure, teacher oversight, and guardrails — not just consumer apps repurposed for the classroom.

Treat AI literacy as core curriculum

The PISA data reveals an emerging "AI divide." Students from more advantaged backgrounds are more likely to receive instruction on evaluating AI-generated content. If schools don't deliberately teach every student how to think critically about what AI produces, this gap will grow. AI literacy belongs in every classroom, not just the ones with the most resources.

The real takeaway

There's a version of this conversation that ends in hand-wringing. But the PISA data points to something more hopeful than the headlines suggest: when teachers lead, students learn.

The countries and school systems that bucked the decline — the UK, Estonia, Singapore, among others — didn't do it by pulling the plug on technology. They did it by using it with clear intention: defined expectations, teacher-led activities, and instruction that builds the deep reading and reasoning skills the global results show are slipping.

Every school is now facing the same question: how do we put powerful tools in students' hands and make sure those tools serve learning instead of replacing it?

That's not a question technology answers on its own. It takes teachers who know their students, curriculum that demands real thinking, and a platform built to support both.

Flint is an AI learning platform designed for teachers, not around them. Every activity is teacher-designed, with built-in guardrails, real-time monitoring, and AI guided by educator expertise — so technology strengthens learning instead of shortcutting it.