We Paid to Make Our Children Less Capable? What EdTech Panic Reveals About Learning and AI
Sandra Buatti-Ramos defines the future of career strategy and career ecosystem design. As founder of Hyphen Innovation and Post-Traditional Careers, she architects research-driven, AI-powered frameworks that dismantle obsolete talent development models and unlock unstoppable potential for post-traditional professionals and organizations.
A powerful and increasingly fatalistic cultural narrative is taking shape around technology-enhanced education in the United States. Institutions spent billions modernizing their schools. Students grew more distracted, academic and cognitive performance slipped, and the very tools meant to improve education left a generation less cognitively capable.[1][2]
Fortune made the case bluntly in February 2026. Its headline announced that the United States had poured $30 billion into replacing textbooks with laptops and tablets, only to produce the first generation less cognitively capable than its parents.[1] The article leaned on neuroscientist Jared Cooney Horvath, whose written Senate testimony argued that Generation Z is the first cohort in modern history to score below its predecessors on standardized measures. Horvath cited falling results on the Programme for International Student Assessment (PISA) and attributed the pattern partly to distraction, task-switching, and the loss of productive difficulty.[1] By August, the argument had crossed to British readers, carried by The Guardian under a headline in Horvath’s own words: “Our children are less cognitively capable”.[2]
The cultural narrative’s appeal is no mystery. Spending on devices climbed. Distraction became hard to ignore. Scores dropped on international achievement tests such as PISA and on some intelligence tests. Research even partially supports the worry, showing that unstructured technology use in educational environments can disrupt learning.[3][4] Yet evidence that some technologies interfere with learning is far from proof that technology drove a generation-wide decline in cognitive capability. The narrative assumes the devices caused it. Whether the evidence can carry that assumption from plausible possibility to proof is another matter.
What the intelligence data demonstrate
The strongest support for the idea that younger generations are growing less cognitively capable comes from the reversal of the Flynn effect, the steady twentieth-century climb in intelligence-test scores from each generation to the next. Bratsberg and Rogeberg (2018) analyzed Norwegian conscription scores. They found a decline within families, between brothers, suggesting environmental causes rather than shifts in who makes up the population. Which environmental factor? The researchers could not say. Media exposure appeared on their list as only one candidate, next to education, nutrition, and health. The timeline poses a further problem for the narrative. The decline began with men born after 1975, who took their military entrance tests in the early 1990s.[5] By the time schools issued laptops and tablets to students, the trend was already well underway.
U.S. data muddy the picture further. When Dworak et al. (2023) analyzed the online test results of 394,378 U.S. adults between 2006 and 2018, they found no common trajectory across cognitive abilities. Matrix reasoning and letter-and-number series slipped; however, verbal reasoning held steady and three-dimensional spatial rotation rose in the researchers’ primary estimates. Cognitive abilities moving in different directions at once sit awkwardly with any theory of across-the-board decline in intelligence. The researchers, for their part, made no claim to know the cause. Working with self-selected volunteers, they noted they could not determine what drove the changes.[6] Read side by side, the Norwegian and American studies support a far more modest conclusion than recent headlines suggest. Some groups have lost ground on some measures, for environmental reasons no one has yet pinned down.
What PISA demonstrates
PISA has become one of the easiest pieces of evidence to point to in this debate. Between 2018 and 2022, average mathematics scores across OECD countries fell by nearly 15 points, the largest decline PISA had recorded.[7[ It is a troubling number; however, it does not explain the cause of the decline. The OECD (2023, p. 84) makes that limitation explicit, cautioning that “PISA results do not establish causality.” Looking further back makes a simple explanation even harder to sustain. Reading and science performance had been declining for years before the pandemic, leading the OECD to conclude that the pattern “must have deeper reasons”. There is also the question of how seriously students take the assessment itself. Seventy-one percent reported putting less effort into PISA than they would have put into a test that affected their grades.[7] So while the decline is real and worth taking seriously, the score data alone cannot tell us whether we are looking at changes in knowledge, motivation, educational conditions, or some combination of all three.
Findings on device use in educational contexts further complicate the cultural narrative. Students who reported using digital devices for learning for up to an hour a day scored 14 points higher in mathematics than students who reported no use, even after accounting for differences in socioeconomic background. Beyond that point, the relationship shifted, with heavier use associated with lower performance.[7] Because these data are correlational, they cannot tell us that a certain amount of technology improves learning or that too much causes scores to fall. They do suggest that simply counting hours may miss something important. As the OECD (2024) later emphasized, what students do with technology matters as much as how much time they spend using it.
When technology really does cost students
The worry has a real basis. At West Point, Carter et al. (2017) randomly assigned economics classrooms to permit or restrict laptops and tablets, and cadets allowed devices scored about 0.18 standard deviations lower on the final exam. In England, test scores rose after schools banned phones, most sharply among the lowest-achieving students.[3]
What each study revealed, though, was the cost of learner distraction. The West Point researchers noted that their results do not extend to classes in which devices are built into instruction, and Beland and Murphy (2016) warned against applying theirs to structured uses. Neither document discusses lasting damage to students’ minds.
The promising variable the narrative leaves out
Harvard provides a useful counterexample. In Kestin et al.’s (2025) physics study, students were randomly assigned to either an AI tutor built around structured problem solving or an in-person active-learning class. Students using the AI tutor performed better on the immediate assessment, with effects ranging from about 0.63 to 1.3 standard deviations. That finding differs sharply from what Carter et al. (2017) found at West Point. In that scenario, laptops and tablets were simply permitted during class, and student performance declined. At Harvard, the technology was deliberately woven into the learning process and engineered to keep students doing the thinking.[8] The AI tutor sequenced problems and embedded expert solutions. It also guided students through their reasoning rather than simply providing an answer, distinguishing it from general-purpose AI chatbots designed primarily to be helpful rather than to promote learning.[8] This distinction may seem small, but it changes the question we should be asking about educational technology. Rather than asking whether technology itself helps or hinders learning, the contrast between West Point and Harvard asks us to examine how the technology is designed, the role it plays in the learning process, and how much of the cognitive work it leaves for the student to do.[4][8]
What stands out about the Harvard tutor is how deliberately it was designed around the learner’s thinking. It presented problems in sequence, incorporated expert solutions, and guided students through the reasoning process rather than simply supplying answers. The researchers themselves drew an important distinction, noting that general-purpose chatbots “are generally designed to be helpful, not to promote learning”. However, an important limitation remains. Students were tested immediately after the lessons, so the study cannot tell us whether those gains held up over time.[8]
The risk worth acknowledging
Buried within the cultural narrative that classroom technology has made a generation less capable is a serious concern for today’s educators. Bjork and Bjork (2011) have shown that performance during learning can be “a highly unreliable index” of whether anything was learned, and that the labor of retrieving and generating answers is what makes knowledge stick.[9] A device or technology that recalls information and produces answers on students’ behalf may undermine the learning process. Early AI research raises this concern. Cross-sectional surveys and small laboratory studies, however, cannot show that AI has already diminished a generation.[10][11]
Better questions for a costly investment
Accountability starts with sharper questions. What does a given technology ask learners to do? Who does the thinking, the student or the machine? When is learning measured, given how little same-day performance reveals about what lasts?[9] Once the tool is gone, can learners do the work on their own?
What students are owed
The empirical evidence does not support the story of a generation whose minds technology has hollowed out. What it supports is narrower and more unsettled. Scores on some achievement and intelligence tests have fallen in some populations, for reasons still unexplained. In studies that test technology directly, the outcome depends on the learning architecture built around a tool and the cognitive work it still asks learners to do.
Purchasing technology was never the same as designing learning, and the two should not be conflated. Before adopting any classroom technology, the question worth asking is what it demands students think, practice, and decide, and what they will eventually be able to do without it.
Read more from Sandra Buatti-Ramos, ACRW, CLMC
Sandra Buatti-Ramos, ACRW, CLMC, Founder, Chief Learning Architect, & Lead Coach
Sandra Buatti-Ramos is a preeminent voice in career strategy and career education ecosystem design. As founder of Hyphen Innovation and Post-Traditional Careers, she develops research-driven, AI-powered frameworks that dismantle outdated talent development models and create scalable pathways to career mobility. A "Top Career Coach" known for her work with students, professionals, and forward-thinking organizations, she fuses dynamic coaching strategies with cutting-edge instructional design to accelerate workforce readiness transformation. Her portfolio spans award-winning career coaching initiatives, the creation of a workforce preparation Innovation Lab, and the launch of a first-of-its-kind AI-driven learning ecosystem.
References:
[1] Rogelberg, S. (2026, February 21). The U.S. spent $30 billion to ditch textbooks for laptops and tablets: The result is the first generation less cognitively capable than their parents. Fortune.
[2] Weale, S. (2026, August 16). ‘Our children are less cognitively capable,’ says neuroscientist who wants to cut screen use in schools. The Guardian.
[3] Beland, L.-P., & Murphy, R. (2016). Ill communication: Technology, distraction & student performance. Labour Economics, 41, 61-76.
[4] Carter, S. P., Greenberg, K., & Walker, M. S. (2017). The impact of computer usage on academic performance: Evidence from a randomized trial at the United States Military Academy. Economics of Education Review, 56, 118-132.
[5] Bratsberg, B., & Rogeberg, O. (2018). Flynn effect and its reversal are both environmentally caused. Proceedings of the National Academy of Sciences of the United States of America, 115(26), 6674-6678.
[6] Dworak, E. M., Revelle, W., & Condon, D. M. (2023). Looking for Flynn effects in a recent online U.S. adult sample: Examining shifts within the SAPA Project. Intelligence, 98, Article 101734.
[7] OECD. (2023). PISA 2022 results (Volume I): The state of learning and equity in education. OECD Publishing.
[8] Kestin, G., Miller, K., Klales, A., Milbourne, T., & Ponti, G. (2025). AI tutoring outperforms in-class active learning: An RCT introducing a novel research-based design in an authentic educational setting. Scientific Reports, 15, Article 17458.
[9] Bjork, E. L., & Bjork, R. A. (2011). Making things hard on yourself, but in a good way: Creating desirable difficulties to enhance learning. In Psychology and the real world (pp. 56-64).
[10] Gerlich, M. (2025). AI tools in society: Impacts on cognitive offloading and the future of critical thinking. Societies, 15, Article 6.
[11] Kosmyna, N., Hauptmann, E., Yuan, Y. T., Situ, J., Liao, X.-H., Beresnitzky, A. V., Braunstein, I., & Maes, P. (2025). Your brain on ChatGPT: Accumulation of cognitive debt when using an AI assistant for essay writing task [Preprint, under review].
[12] OECD. (2024). Students, digital devices and success (OECD Education Policy Perspectives No. 102). OECD Publishing.
