The central insight of the report is deceptively simple but profoundly important for anyone responsible for a child’s education. Engagement, as measured by time on screen, frequency of clicks, responsiveness to animations and the pursuit of in-app rewards, is a poor proxy for genuine learning. A student can be captivated by a colorful interface, motivated by points and badges, and visibly attentive to the task at hand, while acquiring little durable knowledge that transfers beyond the screen. What matters, the expert panel concludes, is whether students can apply new knowledge and skills in other contexts, away from the app or device where they first encountered them. That ability to transfer learning, long recognized in cognitive psychology as the gold standard of education, is the metric that should guide decisions about which technologies deserve a place in a child’s learning life.

The report was compiled by a multidisciplinary panel that included cognitive and educational psychologists, learning scientists and EdTech experts. Together they synthesized evidence spanning an unusually broad landscape of technologies, from educational apps and digital games to intelligent tutoring systems, generative AI tools and large digital learning platforms. Their conclusions are distilled into ten evidence-based recommendations aimed at families, educators, policymakers and developers, all designed to ensure that technology promotes meaningful learning rather than merely captivating attention. The recommendations call on schools, families and developers to prioritize tools that build durable knowledge and skills students can carry into other contexts, rather than evaluating apps on flashy features or superficial measures of interactivity.

Arthur C. Evans Jr., PhD, chief executive officer of the American Psychological Association, framed the report as a call for higher standards in a market that has grown faster than the science supporting it. Educational technology, he noted, has genuine potential to profoundly advance students’ learning and abilities, but parents, teachers and school districts need tools that are safe, supported by evidence and capable of genuinely helping students learn. The recommendations, grounded in psychological research, are intended to help determine whether EdTech products actually produce the learning outcomes they promise. This emphasis on evidence is a deliberate pushback against marketing claims that, in the panel’s assessment, routinely run ahead of what scientific studies can currently support.

One of the report’s most consequential sections concerns generative AI, the class of tools that can produce essays, answers and explanations on demand. The expert panel warns that these tools pose particular risks because they can improve students’ immediate performance without building the underlying knowledge or skills that performance is supposed to reflect. Nicole Barnes, PhD, APA’s executive lead psychologist for education, offered a vivid illustration of the problem. AI can help a student produce a better essay, she observed, but the improved final product does not necessarily mean the student has become a better writer. Just as educators need to separate engagement from learning, she argued, they will increasingly need to separate performance from learning. The distinction matters because performance, the quality of work a student produces at a given moment, has always been an imperfect indicator of learning, and generative AI widens that gap dramatically by allowing polished output to be produced without the cognitive effort that produces durable understanding.

Notably, the report does not call for banning AI tools from the lives of children and adolescents. Instead, it recommends teaching students to actively evaluate AI-generated responses. Students should learn to question the assumptions embedded in chatbots’ answers, identify gaps in their reasoning, compare responses against other sources, and consider how answers might change under different circumstances. In this vision, students should approach AI less as an answer machine and more as something to be continuously tested, challenged and learned from. This reframing is significant because it treats AI literacy not as a matter of technical skill but as an exercise in critical thinking, drawing on decades of research showing that active evaluation and comparison produce deeper learning than passive reception of information.

Another recurring theme in the recommendations is the importance of adult involvement. The panel urges that adults be made a significant part of EdTech implementation, whether that means teachers integrating app-based lessons into broader instruction, or parents sitting alongside younger children as they explore new tools. The reasoning reflects a well-established finding in learning science: children learn more deeply when a knowledgeable adult scaffolds their experience, asks questions, connects new material to prior knowledge and helps them reflect on what they have learned. Technology that replaces adult interaction, rather than augmenting it, tends to fall short of its promise. The panel also calls for investment in research that can keep pace with rapid technological change, an acknowledgment that the scientific evidence base moves far more slowly than the product cycle of the industry it is meant to evaluate.

Responsibility for closing that evidence gap, the report insists, lies primarily with the companies that develop EdTech products and the institutions that choose to adopt them. Developers should be prepared to demonstrate that their products improve learning, not merely that children enjoy using them. Schools and districts, for their part, should expect credible, independent evidence before investing in educational technology, rather than relying on vendor claims or anecdotal success stories. This placement of responsibility is a meaningful shift in framing. Rather than asking overwhelmed parents and teachers to become evaluators of learning science, the report locates the burden of proof with those who profit from and procure these products, a model more akin to how efficacy standards operate in medicine than to how software has traditionally been sold to schools.

For parents navigating a marketplace saturated with apps claiming to teach everything from phonics to calculus, the APA offers refreshingly simple advice: look and ask. The first step is to spend a minute simply watching what happens on the screen when a child uses an app. If much of the experience involves dressing a character, selecting accessories, collecting rewards or responding to flashing images, it is reasonable to question the app’s educational value. These design elements, borrowed directly from the attention economy of gaming and social media, are engineered to maximize engagement and retention rather than learning, and their prominence in an educational product is often a warning sign rather than a selling point. The second step is to ask a teacher, principal, early childhood leader or school board whether there is credible, independent evidence showing that the app actually leads to learning. That single question, posed consistently and persistently, could reshape purchasing decisions across an entire school district.

The timing of the report could hardly be more significant. Classrooms around the world have spent the past several years absorbing waves of new technology, first through pandemic-driven remote learning and now through the explosion of generative AI tools that students encounter both in school and at home. Billions of dollars flow annually into the EdTech sector, and schools face constant pressure to modernize, often with limited capacity to evaluate the scientific merit of the products they adopt. Against that backdrop, the APA’s message cuts through the noise with a principle that is easy to state and hard to ignore: what looks like learning on a screen is not the same as learning in a child’s mind. Durable knowledge, the kind that transfers to new problems, new contexts and new challenges, remains the true measure of educational success, whether the medium is a textbook, a teacher or the most sophisticated algorithm a company can build.

The report’s broader legacy may be in giving parents, educators and policymakers a shared vocabulary for asking harder questions about the technology entering children’s lives. By distinguishing engagement from learning, and performance from learning, the expert panel has articulated a framework that applies as much to a gamified math app as to an AI writing assistant. The challenge now, as the report makes clear, is ensuring that the products built for children are held to standards worthy of the promises made about them, and that the adults responsible for children’s education feel empowered to demand evidence before entrusting young minds to a screen.