Students sit in rows facing the front, while an AI system presents content from the teacher’s traditional position. Our illustration captures a tension in how we imagine education: the blue glow communicates modernity, yet the classroom retains its familiar design. Students receive content, the teacher controls its delivery, and the room directs attention toward a single source. The technology changes while the pedagogical architecture stays in place.
Yet report after report, including those from reputable institutions, gives us this beautiful, blue glow of bullshit. That repetition matters because trusted organizations help define what counts as educational progress. When their images place AI inside the familiar classroom, they carry assumptions about school time, assessment, and authority into the promise of change. The technology takes on an existing role while the design of education escapes scrutiny. Our illustration draws attention to that pattern; an image cannot establish how any one institution teaches.
Artificial intelligence exposes how much of education rests on fragile practices: assignments that confuse textual production with thinking, assessments that reward repetition before judgment, policies that outsource trust to administrative procedure, and claims of innovation whose effects on the lived experience of learning remain unclear.
Over the past few years, this tension has stopped being abstract. In staff rooms, leadership meetings, faculty committees, ministries, accreditation offices, and board discussions, AI has moved from technological curiosity to daily institutional problem. Teachers face student work whose authorship they struggle to assess. Institutions rewrite academic integrity policies under pressure. Educational systems are trying to respond to technologies that change faster than the institutional frameworks meant to govern them.
The rapid expansion of generative systems creates understandable pressure on educational organizations. School leaders, faculty, institutional managers, and policymakers must respond to immediate problems: academic integrity, curriculum redesign, data governance, and unequal access. Yet when the discussion remains limited to adoption, control, or instrumental training, artificial intelligence is reduced to a matter of implementation. That displacement impoverishes the debate, because choosing a tool commits institutions to decisions about how it reorganizes the conditions under which learning, teaching, assessment, and institutional governance take place.
In this context, international organizations, ministries of education, universities, schools, and technology companies have produced guidance documents, training programs, policy statements, and frameworks intended to bring order to the incorporation of artificial intelligence in education. Such work can be useful, but it does not by itself explain what is at stake. That work needs to connect the choice of tools with questions about knowledge, authority, assessment, and institutional governance.
The history of educational technology shows that this pattern did not begin with artificial intelligence (Tyack & Cuban, 1995). Textbooks, calculators, personal computers, the internet, and mobile phones have each been presented, at different moments, as threats or promises of transformation. In each case, urgency, resistance, regulation, and modernization followed.
Yet many schools have changed their devices while keeping their ways of organizing teaching and learning. In Tinkering toward Utopia, Tyack and Cuban (1995) examine how schools reshape reforms through their existing routines. Their account suggests that an institution can take up a new tool while preserving the practices that determine its role. This helps explain why a change in technology can leave the organization of learning intact.
This point is central to the present moment. Despite decades of technological innovation, many institutions continue to organize learning through rigid structures and compliance models. Even when platforms, interfaces, and policy vocabularies change, the basic ways schools manage time, define performance, and distribute authority can remain in place. Artificial intelligence makes this tension more visible because it alters the conditions of access, production, and circulation of knowledge, while many institutions attempt to respond from organizational models designed for another era.
The contemporary crisis in education involves technological and administrative deficits alongside a loss of institutional purpose, an erosion of trust, a reduction of teacher and learner agency, and a difficulty imagining different ways to organize learning. AI brings those tensions into view when institutions must decide how to respond.
The illusion of innovation
Many current AI initiatives present themselves as educational transformation, even though their primary uses continue to automate existing practices: generating exercises, correcting texts, producing summaries, monitoring academic misconduct, and accelerating administrative tasks (Xia et al., 2024; Kizilcec et al., 2024). In such cases, leaders define innovation through the incorporation of tools, leaving the harder institutional questions unresolved.
How does assessment change when generative systems can produce complex texts? How should faculty or teacher time be redistributed when administrative tasks can be automated? How is authority reorganized when knowledge circulates through distributed systems? These questions require a deeper institutional and political discussion than the production of usage guidelines.
Recent work on AI governance warns that the accelerated adoption of opaque systems can strengthen automation, surveillance, and nontransparent decision-making if effective public oversight is absent (Taeihagh, 2025; Williamson, Molnar, & Boninger, 2024). Institutions that use those systems can reinforce practices whose effects on trust and authority deserve scrutiny.
These questions reach beyond the classroom because the providers of AI systems can shape the choices available to teachers and students. Examining that influence requires us to connect the use of a tool with who controls it.
Questions for educational leaders
-
Which AI platforms does the institution use?
-
What student, faculty, or institutional data do those platforms collect?
-
Who can audit the systems being used?
-
Which educational decisions are shaped by external providers?
-
What professional capacities do leadership and teaching teams need to develop?
Platforms, power, and educational sovereignty
The second problem concerns the role of major technology platforms in reorganizing education. Much of the AI infrastructure entering schools, universities, training systems, and public agencies belongs to private companies that operate through closed development models, intensive data collection, and constant system updates. Educational institutions integrate tools whose internal mechanisms they lack the means to inspect, modify, or supervise. This situation exceeds software adoption because it affects the conditions under which knowledge is produced, circulated, and validated.
Platforms organize interactions, prioritize content, register behavior, and shape forms of intellectual work through the services they provide. In educational settings, those decisions carry pedagogical significance because they influence which questions are asked, which sources appear relevant, which forms of production become suspect, and which forms of participation become normalized. The institution may retain formal educational responsibility, but part of its operational authority shifts toward systems designed and governed by external actors.
This phenomenon reflects broader debates about the concentration of technological power and the extraction of data. Critical research on digital platforms has shown how everyday activities can become sources of data, algorithmic influence, and institutional dependency (Benjamin, 2019; Couldry & Mejías, 2019; van Dijck, Poell, & de Waal, 2018).
Education participates in this dynamic when ordinary academic activities become part of private systems that educational institutions do not control.
The thinking, exploration, and intellectual production of students and educators can become commercial resources for external platforms.
These uses of data link privacy to governance, because control over the system affects who can oversee its use. When private platforms control critical infrastructure for learning, institutions lose practical capacity to determine how their learning environments function. Updates follow corporate calendars, moderation criteria can change without local educational deliberation, and classification mechanisms remain opaque to students, educators, and institutional leaders.
Under those conditions, institutions end up managing consequences they cannot always explain or correct.
In Beyond the digital enclosure (Moravec, 2026a), I analyze this dynamic as a progressive loss of educational sovereignty: private platforms begin to define central aspects of institutional life that remained under local educational control. Contracts and licenses establish one basis for dependency, which grows through the incorporation of metrics, interfaces, and decision criteria designed outside educational communities.
This issue matters across regions, but its effects differ by context. In high-income countries, institutional dependency may emerge through procurement, platform lock-in, analytics dashboards, assessment systems, and outsourcing of educational judgment. In lower- and middle-income countries, the problem can intersect with infrastructure inequality, donor agendas, language marginalization, and the importation of models designed in other settings. In international education and development contexts, AI may deepen dependency when institutions adopt closed systems that organize knowledge, assessment, and interaction according to commercial or geopolitical priorities beyond local control.
The Global South faces particular vulnerabilities. Schools and universities may incorporate software produced in North America, Europe, or China while exposing student data, linguistic patterns, and traces of intellectual work to systems used to train external models. At the same time, institutions may import automated classification, assessment, and recommendation mechanisms built from cultural, linguistic, and economic assumptions that do not reflect local realities.
For this reason, the global debate should not remain limited to technical training or responsible use. It needs to address educational sovereignty, audit capacity, democratic control of digital infrastructure, language justice, public accountability, and the institutional right to define educational purposes.
The exhaustion of performative leadership
Many institutions produce strategies, manuals, policy statements, and events about artificial intelligence to demonstrate responsiveness to technological change. Such activity performs understandable institutional functions: it organizes internal conversations, communicates public concern, and signals that the organization is keeping up.
This activity can create a wide gap between discourse and practice when documents do not change the resources, responsibilities, or working conditions required to transform educational experience.
In many cases, these guidelines function as an institutional release valve. They allow the organization to show that it has done something, while much of the emotional, ethical, and operational work moves to individual teachers and faculty members who must decide without support how to assess, what to allow, what to prohibit, and how to interpret student use of generative tools.
Over recent decades, the education sector has produced an extensive institutional literature on technological innovation. Much of it shares a predictable vocabulary: teacher training, equity, ethical use, pilot programs, and personalized learning. These elements can have value, but they lose force when they are not connected to decisions about budget, professional time, assessment, infrastructure, data ownership, and participation of educators and students in technological governance.
Institutions create a gap between policy and practice when they treat producing a framework as evidence that they have changed the conditions of learning. When policies regulate individual behavior while leaving system governance unresolved, responsibility shifts toward educators and students while the technological systems that generate new risks remain opaque. As a result, people working and learning inside institutions absorb costs that policy documents need to account for: redesigning assessments, reviewing suspect work, adapting to changing platforms, and responding to contradictory expectations.
These costs connect policy with the conditions of teachers’ work. Institutions need to account for them when they assign responsibilities, set budgets, and decide who can question a system. The table below shows where those decisions can change the work behind an AI policy.
Table 1
Institutional Responses to AI: Compliance or Transformation
| Management Dimension | Performative Response (Compliance) | Institutional Transformation (Sovereignty) |
|---|---|---|
| Institutional Policy | Rapid publication of usage and sanction guidelines | Review of assessment formats and professional workload |
| Team Development | Isolated technical training on prompt use | Sustained development of institutional capacity and imagination |
| Institutional Climate | Deployment of detection and surveillance tools | Creation of environments grounded in trust, reflection, and judgment |
| Internal Processes | Administrative automation oriented toward control | Review of the educational purposes the institution is designed to serve |
Beyond compliance
These choices require institutions to develop the capacity to judge how new tools serve educational purposes under conditions of accelerated change. Many organizations respond to artificial intelligence through administrative categories designed to manage stability, control, and predictability.
Generative systems raise questions about assessment, authority, and learning that require educational judgment alongside technical rules. They require institutions to reconsider what assessment means, how educational authority is distributed, which human capacities gain value, and which forms of educational experience deserve protection.
The debate about artificial intelligence can become confined to uncritical adoption or defensive rejection. Both positions keep attention on the tool, which leaves the institutional conditions that give it meaning unexplored. Recent debates on AI literacy suggest a different path: institutions need to help people use generative systems and understand how those systems produce responses, which assumptions they carry, which risks they introduce, and how they affect the autonomy of educators and learners (Mills et al., 2024; Conrad & Kamperman, 2025).
Institutions can use that understanding to define the pedagogical, political, and organizational conditions for adopting AI. Those conditions determine whose authority the systems displace and which capacities they help develop.
Answering these questions requires institutions to revisit some of the assumptions that have organized much of contemporary schooling: transmission-centered teaching, standardized assessment, rigid administration of time, and separation between policy design, teaching, and learning. Many systems operate through models conceived to manage stability, scarcity of information, and predictable pathways. The expansion of algorithmic systems, automation, and distributed production of knowledge changes those conditions and makes many inherited institutional responses insufficient.
Assessment offers a clear example. When generative tools can produce coherent texts in seconds, mechanisms that treat individual reproduction of information as proof of learning become insufficient.
When access to knowledge no longer depends on memorization, the capacities education must value begin to change.
This shift requires institutions to reconsider:
-
the distribution of authority in learning environments;
-
the recognition of new forms of intellectual work;
-
the development of judgment in information-saturated contexts;
-
the place of creativity, collaboration, and interpretation in educational experience; and
-
the protection of autonomous thought in automated systems.
From this perspective, educational leaders need to connect technological choices with governance, institutional purpose, and shared work on alternatives. For institutional leaders, this requires them to create conditions for professional judgment, responsible experimentation, and pedagogical trust. Organizational models shape how institutions respond to technological change, so leaders need to examine those models as part of their work.
These conditions give institutional imagination a practical test: can teachers and learners help change the arrangements that govern their work? Their involvement connects the purpose of a policy with its effects on learning and trust.
For many institutions, this work can begin with decisions about assessment, professional time, pedagogical trust, and technological governance. The capacity to exercise judgment about artificial intelligence will depend on whether institutions support judgment, collective deliberation, and professional autonomy as technologies change.
The most critical part is identifying which aspects of human learning should remain at the center of educational experience.
I examine these questions in Build a Positive Rebellion: Create New Education Futures (Moravec, 2026b), through the relationships among agency, trust, and the conditions that allow people to shape their education.
Recommended readings
Benjamin, R. (2019). Race after technology: Abolitionist tools for the new Jim code. Polity.
Conrad, K., & Kamperman, S. (2025). Building critical AI literacy: An approach to generative AI. Thresholds in Education, 48(2), 142–158. https://eric.ed.gov/?id=EJ1483967
Couldry, N., & Mejías, U. A. (2019). The costs of connection: How data is colonizing human life and appropriating it for capitalism. Stanford University Press.
Kizilcec, R. F., Huber, E., Papanastasiou, E. C., Cram, A., Makridis, C. A., Smolansky, A., Zeivots, S., & Raduescu, C. (2024). Perceived impact of generative AI on assessments: Comparing educator and student perspectives in Australia, Cyprus, and the United States. Computers and Education: Artificial Intelligence, 7, 100269. https://doi.org/10.1016/j.caeai.2024.100269
Mills, K., Ruiz, P., Lee, K. W., Coenraad, M., Fusco, J., Roschelle, J., & Weisgrau, J. (2024). AI literacy: A framework to understand, evaluate, and use emerging technology. Digital Promise.
Moravec, J. W. (2026a). Beyond the digital enclosure: AI, coloniality, and the pursuit of educational sovereignty. Learning Futures and Emerging Technologies, 34(3). https://doi.org/10.1108/LFET-01-2026-0004
Moravec, J. W. (2026b). Build a positive rebellion: Create new education futures. Education Futures. https://positiverebellion.org/
Taeihagh, A. (2025). Governance of generative AI. Policy and Society, 44(1), 1–22. https://doi.org/10.1093/polsoc/puaf001
Tyack, D., & Cuban, L. (1995). Tinkering toward utopia: A century of public school reform. Harvard University Press.
van Dijck, J., Poell, T., & de Waal, M. (2018). The platform society: Public values in a connective world. Oxford University Press.
Williamson, B., Molnar, A., & Boninger, F. (2024). Time for a pause: Without effective public oversight, AI in schools will do more harm than good. National Education Policy Center. https://nepc.colorado.edu/publication/ai
Xia, Q., Weng, X., Ouyang, F., Lin, T. J., & Chiu, T. K. F. (2024). A scoping review on how generative artificial intelligence transforms assessment in higher education. International Journal of Educational Technology in Higher Education, 21(1), 40. https://doi.org/10.1186/s41239-024-00468-z
