AI in schools: unmediated use is undermining student learning

Artificial intelligence has moved well past the classroom pilot stage. Across the globe, students are using AI tools daily for their schoolwork — and the data on what that actually looks like in practice is beginning to arrive. The picture is more complicated than either the optimists or the pessimists predicted, and it demands a serious, evidence-grounded response from educators, policymakers, and families alike.

01What the Data Actually Shows

A February 2026 survey by the Pew Research Center found that just over half of U.S. teenagers have used AI chatbots to help with schoolwork, with roughly one in ten reporting that AI handles most or all of their assignments. More striking is what students say about their peers: approximately 60% believe that students at their schools use AI to cheat "often." The concern is not hypothetical — it is already embedded in how young people understand their own academic environment. The risks extend well beyond academic dishonesty. A commentary by Morgan Polikoff, a professor at USC's Rossier School of Education, published by EdSource in April 2026, cites analyses showing that 20% of student interactions with AI in school settings involve cheating, bullying, or self-harm. That figure — one in five interactions — reframes the conversation. The question is no longer whether AI poses risks in educational settings, but how significant those risks are and who bears responsibility for managing them. The parent-child awareness gap compounds the problem. Polikoff's own research at USC found that only 7% of parents believed their teenagers used AI for schoolwork multiple times a week, while 27% of teens confirmed they did — and that figure is likely an undercount. Parents also overwhelmingly reported being unaware of their schools' AI policies. Without informed adults in the loop, the ethical scaffolding around student AI use collapses.

02Brazil: High Adoption, Uneven Guidance

Brazil presents a particularly instructive case. The 15th TIC Educação survey, cited in a Fundação Lemann analysis from April 2026, found that seven in ten high school students use AI to write essays, summarize texts, and complete assignments. Among students in the final years of elementary school, the figure is four in ten — a cohort that is developing study habits and cognitive routines while simultaneously relying on tools that can short-circuit those very processes. An Oxford University study referenced in the same analysis captures the dual nature of this dynamic: 90% of students reported developing skills with AI support, yet six in ten also admitted that the technology had a negative effect on their creative thinking, because it made problem-solving too easy. The same tool that scaffolds learning can also hollow it out, depending entirely on how it is used and whether an educator is present to shape that use. On the teacher side, Brazil is also moving quickly. Data from the OECD's 2024 Teaching and Learning International Survey (Talis), reported by Exame in April 2026, found that 56% of Brazilian teachers already use AI to improve their work — one of the higher rates among the 53 countries surveyed. Teachers are using AI for lesson planning, grading, and generating differentiated materials. The adoption is real, but the pedagogical frameworks to guide it are still catching up.

03The Ethical Imperative: Neither Ban nor Blank Check

The evidence points toward a path that is neither prohibition nor uncritical adoption. Both extremes fail students. Banning AI in schools ignores the reality that students will use it regardless, and risks deepening inequalities between those who learn to use it critically and those who do not. Unrestricted adoption, without pedagogical intention, risks producing a generation that can prompt a chatbot but cannot construct an argument. The more productive framing is one of intentional integration: designing assessments that value process over product, investing in teacher training that goes beyond tool tutorials to include ethical reasoning, and building transparent AI policies that families can actually understand and engage with. Learning has always been about the productive struggle — the effort of not knowing and working toward understanding. AI should support that struggle, not eliminate it. What should educators and policymakers watch in the months ahead? The development of assessment frameworks that are genuinely AI-resistant — not through surveillance, but through design — will be a key indicator of whether schools are responding to this moment with the seriousness it demands.