The core problem the study addresses is what educational researchers have long called the difference between first-order and second-order change. First-order changes are visible and easy to count: a school buys tablets, installs interactive whiteboards, or secures a fast internet connection. Second-order changes are far harder to see because they concern beliefs, practices, and the deep structure of teaching and learning itself. When a school moves from digitizing its existing lessons to genuinely reorganizing learning around personalized, competence-based trajectories, it has undergone a second-order transformation. Traditional indicators, built for the infrastructure era, simply cannot register that shift, which is why a school can score highly on equipment measures while its classrooms remain pedagogically unchanged.

Uvarov’s framework, which he calls the school digital renewal process, describes this evolution as a progression of stages. In the earliest stages, digital technology is grafted onto traditional teaching: the same lectures, the same textbooks, and the same assessments, now delivered through screens. At intermediate stages, schools begin to adapt their content and routines, blending online and face-to-face work and experimenting with new formats. Only at the most advanced stages does the transformation become structural, with learning organized around individual competence development, flexible pacing, and the kind of personalized mastery approaches that researchers have pursued since Benjamin Bloom famously posed his two sigma problem in 1984, asking why classroom instruction could not be made as effective as one-to-one tutoring. Bloom’s revised taxonomy of educational objectives, later formalized by Anderson and Krathwohl, supplies the conceptual ladder the new study climbs: from remembering and understanding at the bottom to analyzing, evaluating, and creating at the top.

The methodological innovation at the heart of the paper is deceptively simple to state and radical in its implications. Instead of surveying teachers, interviewing principals, or auditing technology stocks, the approach gathers its evidence from publicly available digital resources that schools themselves publish: websites, course catalogs, shared repositories, project descriptions, and other artifacts of their everyday digital activity. Artificial intelligence tools, including large language models of the ChatGPT class, are then used to analyze this material at scale, classifying it and mapping it against the stages of the renewal framework. What a school says about itself in public, in other words, becomes a measurable trace of how far its transformation has actually progressed.

This is where the technical work of the study becomes concrete. Drawing on empirical data from international schools, the research demonstrates that the automated analysis of digital footprints can detect indicators of second-order change that conventional instruments miss. The publicly available resources of a school that has genuinely embraced personalized, competence-based learning look different from those of a school that has merely digitized its old practices: the tasks described require different levels of cognitive processing, the organization of learning is framed differently, and the language of the institution itself shifts. By anchoring the classification in Bloom’s revised taxonomy, the method gains a principled yardstick for judging whether the learning activities a school describes sit at the lower or upper rungs of the cognitive ladder, and whether the school’s stated practices match the deep pedagogical restructuring that the age of AI demands.

The significance of avoiding teacher surveys is easy to underestimate. Surveys are expensive, slow, and subject to well-documented biases: teachers may report what they believe is expected of them, may not recognize how far their practice departs from their self-image, or may simply be too burdened to answer carefully. Resistance to change, which researchers from Larry Cuban to Michael Fullan have documented across decades of reform efforts, further complicates self-reporting, because the very people whose transformation is being measured are asked to describe it themselves. By contrast, digital footprints are produced continuously, as a by-product of ordinary school activity, and can be re-analyzed whenever needed without imposing any additional workload on school staff. The approach is therefore not only cheaper but potentially more honest, capturing what schools actually do and display rather than what respondents recall or claim.

Scalability is the second major advantage. National governments and international organizations, from the OECD to UNESCO, have invested heavily in monitoring digital education, producing index reports and assessment toolkits that compare countries and systems. Yet these efforts overwhelmingly rely on aggregated self-reports and infrastructure statistics, which tell policymakers how many computers schools have but not whether learning has changed. An indicator system built on automated AI analysis of public digital resources could, in principle, track thousands of schools at a fraction of the current cost, producing comparable longitudinal data on the later and most important stages of digital renewal. For education systems struggling to justify enormous technology budgets, that kind of evidence would be transformative: it would show not whether schools have bought the tools, but whether the tools have changed the learning.

The timing of the study is not accidental. The arrival of generative AI has thrown the traditional school model into sharp relief. Students already use large language models for homework, research, and writing, as surveys such as the Digital Education Council’s global student survey have documented, and schools are scrambling to respond. Platforms like Summit Learning have demonstrated that personalized mastery ecosystems, organized around individual learning trajectories and competency-based progression, are technically feasible at scale, even as they have exposed the organizational and cultural difficulties of implementing them. Uvarov argues that in this environment the schools that matter to monitor are not those still acquiring equipment but those attempting the deepest form of renewal, and it is precisely those advanced stages for which existing indicators are least adequate.

The framework also reframes what educational scenarios mean for the AI era. If the goal of schooling shifts from transmitting a fixed canon of content to cultivating competencies that learners can apply and extend throughout their lives, then the indicators of success must shift as well. A school’s published digital presence, analyzed through the lens of cognitive complexity and learning organization, offers a window into that deeper reality. The study’s demonstration that this window can be opened automatically, using tools that are now widely available, suggests a future in which the monitoring of educational transformation becomes as continuous and data-rich as the monitoring of public health or the economy.

There are, of course, challenges the research does not claim to have solved. Digital footprints can be curated, and the gap between what a school displays and what happens in its classrooms will never be zero. The interpretation of public resources requires careful calibration to avoid penalizing schools with different levels of web presence or different cultural norms about self-presentation. Ethical questions about automated analysis of institutional data, even publicly available data, will need continued attention. But the study establishes something that has been missing from the field: proof of feasibility. It shows that the second-order changes, the pedagogical transformations that genuinely define digital renewal, can be detected without questionnaires, at scale, using the AI tools that themselves are driving the transformation. For a discipline that has spent forty years measuring the easy things because the hard things were unmeasurable, that is a genuine breakthrough, and one that policymakers, school leaders, and researchers monitoring the global wave of AI-era educational reform will be watching closely.

Subject of Research: Developing AI-based indicators to monitor the school digital renewal process and deep pedagogical transformation

Article Title: Developing Indicators for School Digital Renewal in the Age of AI

Article References: Developing Indicators for School Digital Renewal in the Age of AI. (n.d.). https://doi.org/10.1007/s44366-026-0080-4

Image Credits: AI Generated

DOI: 10.1007/s44366-026-0080-4

Keywords: school digital renewal, AI in education, digital transformation indicators, Bloom's revised taxonomy, personalized learning, ChatGPT, digital footprints, educational monitoring, second-order change, competence-based learning, Developing, Indicators