We have built elaborate institutions to educate people, while human learning extends into activities those institutions do not control. A child learns language through relationships. A newcomer acquires judgment through work with experienced colleagues. Neighbors learn by comparing explanations for a shared problem. Schools contribute knowledge and skilled teaching to these processes, but learning also develops through care, curiosity, disagreement, experimentation, and participation (Lave & Wenger, 1991; National Academies of Sciences, Engineering, and Medicine, 2018).
Those opportunities depend on conditions people may have little authority to change. Consider a novice whose employer automates the tasks through which earlier recruits practiced judgment, a reader whose information comes through a platform with opaque selection rules, or a carer whose work schedule leaves no time to join a community project. These illustrative situations connect an educational question with an institutional one: who protects the conditions under which people can learn?
Unprepared: Education Beyond 2030 examines preparation without power, the prospect that people become capable of adapting while retaining little influence over the systems defining their options (Moravec, 2026b). This essay develops that problem through three connected propositions. Institutional decisions shape the development, distribution, and exercise of human capability by organizing practice, access, recognition, and authority. Cognitive automation can improve immediate performance while disrupting the reproduction of expertise if it removes experiences through which newcomers develop judgment. Educational sovereignty requires enforceable authority over learning conditions, supported by resources, representation, and viable alternatives.
These propositions connect learning science with an analytical and normative account of institutional power. The workshop scenario examines how increased output could narrow routes into expertise, without treating that outcome as inevitable. The six-gap framework identifies barriers to learning and action, while the sovereignty argument asks what authority people need to challenge and reshape those arrangements.
Understanding a system does not confer authority to change it. Institutions can possess extensive information about people while restricting scrutiny of their own decisions. Protecting personal inquiry and making consequential institutional decisions examinable address this imbalance; enforceable rights give people means to act on what they discover.
Learning happens across society
Cognitive research explains how people attend, remember, construct understanding, and regulate their learning. Sociocultural research explains how they enter practices, use tools, and develop relationships through which knowledge acquires meaning. An institutional account asks who controls access to those experiences. The three levels connect: decisions about time, membership, tools, and responsibility affect opportunities to participate, which affect the cognitive activity through which people develop expertise.
The National Academies of Sciences, Engineering, and Medicine’s (2018) How People Learn II examines learning through interactions among prior knowledge, memory, attention, culture, and self-regulation. People interpret new information through what they already know. That knowledge can support understanding or sustain misconceptions. Attention affects what receives processing, while self-regulation supports choices about goals, strategies, and further practice. Exposure to information therefore provides an incomplete account of learning.
Motivation and identity connect these processes with participation. The report’s chapter on motivation examines how expectations of competence, valued goals, and social identities influence engagement. A learning opportunity has a different meaning for someone who expects to contribute than for someone who expects exclusion. This connects the organization of a setting with the learner’s willingness to attempt a task, persist, and seek feedback.
Retrieval research provides a concrete distinction between doing something once and retaining the knowledge to do it again. In a vocabulary experiment, Karpicke and Roediger (2008) found that repeated retrieval after an initial correct response improved delayed recall, whereas repeated study under that condition did not. A correct answer during practice and recall after a delay reveal different aspects of capability. When institutions judge learning through completed work, they need to consider whether their measures capture understanding that persists beyond the support available during the task.
Lave and Wenger’s (1991) account of legitimate peripheral participation develops the social side of this problem. Newcomers learn through access to a community’s practice and movement toward fuller participation. Membership, responsibility, and relationships matter alongside instructional material. A novice may have a manual and a training account yet lack opportunities to observe consequential decisions, ask questions, or undertake work with support.
Consider an illustrative repair workshop. A beginner watches an experienced worker diagnose a fault, attempts part of the repair, and discusses the result. Over successive assignments, the beginner encounters different faults, tests explanations, and takes responsibility for decisions. The cognitive activity includes recalling relevant knowledge and interpreting feedback; participation gives that activity a purpose and places it within relationships of trust. Paid time, equipment, task allocation, and permission to make mistakes make this route possible.
Now suppose the employer introduces a system that supplies diagnoses and automates routine repairs. It may reduce errors and let novices complete work they could not undertake alone. The developmental question is what beginners still get to observe, attempt, explain, and revise. If experts handle every exception while novices follow instructions, the workshop could improve its output while weakening the route through which newcomers develop judgment. If it preserves supported investigation and discussion, assistance could create new routes into expertise. These alternatives concern the organization of practice, not a necessary consequence of automation.
Moll et al. (1992) extend attention to knowledge developed in households through work and everyday activity. Their funds of knowledge approach connects that knowledge with classroom teaching. It challenges an account that identifies knowledge only through what institutions teach or certify. A household’s capabilities become intelligible through the activities in which people use them and the relationships through which they share them.
The argument also concerns learning whose purpose lies in cultural participation, relationships, or inquiry itself. Through sharing stories, practicing a craft, or comparing experiences across generations, people can sustain knowledge while questioning and revising what they inherit. Independent reading and experimentation can extend understanding without a professional goal or a demand for recognition. These examples connect the cultural and everyday dimensions of learning described by Moll et al. (1992) and the National Academies of Sciences, Engineering, and Medicine (2018). They place the development of human capabilities across society within the argument, including activities whose value participants define for themselves.
Table 1 brings these connections together. The institutional level is the essay’s synthesis, linking learning processes with decisions about access and participation. It places schools within a wider account of how knowledge develops, while retaining their contributions to foundational and disciplinary understanding.
Table 1
Learning Processes and the Conditions That Support Them
| Level | What it explains | Illustrative activity | Conditions at stake |
|---|---|---|---|
| Cognitive | Attention, prior knowledge, retrieval, understanding, feedback, and self-regulation. | A novice tests a diagnosis, explains an error, and recalls the reasoning later. | Time for practice, useful feedback, and tasks that require thought. |
| Social | Participation, motivation, identity, membership, and the use of shared tools. | A newcomer joins consequential work and assumes responsibility with support. | Relationships, belonging, mentorship, and access to meaningful activity. |
| Political and institutional | Control over resources, participation, information, and consequential decisions. | Workers secure mentorship time; members govern a shared service. | Rights, resources, representation, independent scrutiny, and enforceable remedies. |
Note. Author’s synthesis of Karpicke and Roediger (2008), Lave and Wenger (1991), Moll et al. (1992), and National Academies of Sciences, Engineering, and Medicine (2018). Activities illustrate the relationships; they are not reported study cases.
The institutional paradox
Formal education offers governments identifiable organizations whose budgets, curricula, staffing, and qualifications they can regulate. Learning across society depends on decisions distributed among employers, households, public services, technology providers, and civic groups. This administrative difference helps explain the attraction of school-centered reform: an education ministry can change a curriculum while having little direct authority over a worker’s schedule or a platform’s contract. The contrast concerns the distribution of authority: responsibility for instruction sits within identifiable institutions, while control over participation crosses their boundaries.
The resulting paradox concerns the fit between the scope of learning and the scope of responsibility. Schools can strengthen teaching while learners lose access to practice elsewhere. A free course can exist while care obligations prevent participation. A community can possess relevant knowledge while its members lack the credentials institutions accept. Improving provision within one organization leaves these gaps unresolved when the barrier arises through decisions made by another.
Recognition introduces a distinction between learning, evidence of competence, and institutional acceptance of that evidence. An applicant may possess knowledge without holding an accepted qualification; a qualification may establish competence for specified tasks without demonstrating judgment in every setting. Recognition can open opportunities when assessment examines relevant capabilities. It can restrict them when fees, documentation, or narrow criteria exclude people whose knowledge developed through work or care. Assessment therefore needs to examine relevant competence while making its criteria and access requirements contestable.
Sen’s (1999) capability approach provides a normative basis for examining whether people can use the opportunities available to them. Applied here, it directs attention to the practical freedom to participate. Time, transport, accessible formats, income, and care obligations can determine what a person can do with a resource. Table 2 proposes an analytical framework distinguishing opportunity gaps from problems of evidence, recognition, agency, resilience, and sovereignty.
Three connected processes make the institutional relationship clearer. Development of capability concerns the experiences through which people acquire knowledge, judgment, and expertise. Distribution of capability concerns who can enter those experiences and obtain recognition for what they develop. Exercise of capability concerns the resources and authority through which people can act on what they know. These distinctions connect the learning accounts above with Sen’s normative approach.
In the workshop, task allocation affects development by determining who practices diagnosis. Selection for mentorship affects distribution: paid time and access may reach some recruits while others remain assigned to repetitive work. Assessment affects whose judgment the organization recognizes. Decision rights affect exercise: a competent worker may detect a problem yet lack authority to question the automated recommendation or stop a repair. Power can therefore shape the formation, allocation, and use of expertise. These processes can also reinforce one another, as recognized experts gain access to further consequential work while excluded novices lose opportunities to demonstrate what they could learn.
Table 2
The Six Gaps in Human Learning
| Gap | Illustrative situation | What a response must address |
|---|---|---|
| Opportunity | A free course exists, but a carer cannot attend. | The practical conditions of participation, including time and access. |
| Evidence | An AI-assisted answer is correct, but the user cannot explain it. | Retained knowledge, understanding, and capability under relevant conditions. |
| Recognition | An experienced applicant lacks an accepted qualification. | Fair routes to demonstrate competence for a defined purpose. |
| Agency | A worker understands a scheduling decision but cannot challenge it. | A route from voice to influence and enforceable rights, with reasons and remedies. |
| Resilience | A flood closes a shared space and disperses its participants. | Continuity of relationships, access, and supported practice. |
| Sovereignty | A community depends on infrastructure whose terms it cannot influence. | Continuing decision rights, protected inspection, usable alternatives, and resources for enforcement. |
Note. Author’s proposed analytical framework, informed by Sen (1999) and Moravec (2026a, 2026b). Situations are illustrative.
These six gaps locate different barriers between an opportunity to learn and the authority to sustain it. Agency concerns the capacity to act and contest a decision; sovereignty concerns sustained authority over the arrangements determining those opportunities. Responding to both requires connecting responsibilities where learning crosses institutional boundaries. Schools remain responsible for teaching and access to knowledge. Labor bodies govern employment rights; public authorities govern housing and emergency response. Shared duties become necessary when their decisions combine to determine participation. Educational sovereignty asks whether affected people can shape those decisions and obtain a response when responsibilities overlap.
When learning environments become instruments of power
Some threats arise through deliberate action, others through incentives or dependency. Treating all three as a single problem obscures both who benefits and what a remedy must change. A deceptive actor, a provider optimizing a commercial metric, and an organization controlling indispensable infrastructure may restrict agency through different mechanisms.
Deliberate manipulation uses deception or coercion to influence what people believe or can question. The Federal Trade Commission (FTC, 2019) found that Cambridge Analytica used deceptive practices to harvest Facebook users’ personal information for voter profiling and targeting. The finding establishes deception in data collection; it does not establish the effectiveness of the resulting political persuasion. The case matters here because an actor acquired information about people under misleading terms and used it for attempts to influence them. Protecting inquiry requires accountability for those practices as well as opportunities to examine the claims people encounter.
Incentive-driven harm can arise when an organization rewards outcomes that conflict with learning. Consider a hypothetical platform that selects material to maximize continued engagement. If provocative or misleading material serves that metric, a system may reward its circulation without anyone adopting an explicit goal of reducing understanding. The question concerns how a design objective shapes attention and evaluation. In a meta-analysis of 20 experiments with US participants, Pennycook and Rand (2022) found that accuracy prompts improved the quality of intended sharing. Their result supports the relevance of attention to accuracy; it does not measure the effects of an entire platform business model.
Structural dependency concerns the ability to contest or leave an arrangement. A service can be useful while placing its users under terms they cannot influence. In its 2025 UK cloud investigation, the Competition and Markets Authority (CMA) found that concentration and barriers to entry supported significant market power for Microsoft and Amazon Web Services. It also identified commercial and technical barriers to switching and using multiple providers, including data-transfer costs and difficulties substituting services (Competition and Markets Authority, 2025). These are findings about cloud markets. They illustrate why a contractual right to leave may offer limited practical choice when organizations lack the resources to move.
For learning institutions, the corresponding question is who controls access to records, tools, and knowledge, and what maintaining an alternative would require. A library cannot secure independence through a promise of portability alone if migration costs exceed its budget. A community may understand a system’s operation while lacking an institution capable of bargaining over its terms. Here, knowledge of dependency must connect with resources and collective authority before it can support action.
Table 3 separates these mechanisms from wider pressures on learning conditions. The safeguards are proposals, whose design should follow the mechanism and the institution exercising authority.
Table 3
Pressures on Learning, Mechanisms of Harm, and Protective Arrangements
| Pressure or mechanism | What is at stake | Evidence or scenario | Proposed protection |
|---|---|---|---|
| Deliberate manipulation | Trust, inquiry, and informed judgment. | FTC findings document deceptive collection for voter profiling and targeting. | Investigate deception and compel justified correction while protecting personal inquiry. |
| Incentive-driven harm | Attention and careful evaluation. | Accuracy-prompt experiments support an attention mechanism; the engagement example is hypothetical. | Examine design objectives and test effects on information quality and understanding. |
| Structural dependency | Control over tools, access, and alternatives. | CMA findings document barriers to switching cloud providers. | Combine portability and interoperability rights with pooled purchasing, migration funding, and maintained alternatives. |
| Cognitive automation | Practice, feedback, and expertise development. | An AI mathematics trial found design-dependent effects on unaided performance; workplace consequences are scenarios. | Preserve supported practice and assess retained capability alongside output. |
| Economic and climate disruption | Time, continuity, and relationships. | Financial-concern studies examine cognitive performance; UNICEF documents school disruption. | Address participation barriers and maintain shared spaces and networks during disruption. |
| Capture and coercion | Representation and freedom to question. | OECD analyzes capture risks; DSA enforcement against X documents resistance to transparency duties. | Protect independent inspection and remedies; scrutinize public and civic bodies as well as suppliers. |
Note. Sources: Bastani et al. (2025), Competition and Markets Authority (2025), European Commission (2025), FTC (2019), Mani et al. (2013), Organisation for Economic Co-operation and Development (2017), Pennycook and Rand (2022), and United Nations Children’s Fund (2025). Protective arrangements are the author’s proposals.
The danger beyond 2030
The 2030s introduce no fixed threshold at which learning changes character. They provide a horizon for examining how developments already underway could interact. Three matter here: automation of cognitive tasks, disruption to the social and material resources supporting participation, and concentration of digital infrastructure. Their convergence could separate successful performance from expertise and deepen dependence at the moment people most need alternatives.
AI makes the distinction between output and independent capability concrete. Bastani et al. (2025) studied close to a thousand mathematics students at one high school in Turkey. Both GPT-4 interfaces improved assisted practice performance. After access was removed, students using a general-assistance interface performed below the control group. A tutor with teacher-designed hints avoided that decline but did not establish superior unaided exam performance. The findings concern one school, subject, model, and implementation. They show that the design of assistance can change the relationship between completing a task and developing the capability to perform it.
Four dimensions help specify what changes. Task performance concerns what a person can accomplish with available assistance. Independent capability concerns developed knowledge, understanding, and judgment. Individual agency concerns choices and actions the person can exercise, including the use of assistance that enables participation. Collective authority concerns people’s capacity to shape the systems defining those choices. An AI service can improve performance and widen individual options while concentrating control over its rules, data, and continued availability. Gains in one dimension therefore require examination alongside changes in the others.
Return to the workshop as a futures scenario. If routine diagnostic work disappears and established experts retain the difficult cases, beginners may complete more repairs while practicing less judgment. When experienced workers retire, the organization may discover that its former pathway into expertise has narrowed. This is a question about the reproduction of expertise across generations, not a forecast of inevitable skill loss. The mathematics trial establishes a distinction between assisted performance and learning in its study setting; the workplace scenario draws its developmental mechanism from that distinction and the account of access to practice in Lave and Wenger (1991).
A different scenario preserves explanation, guided practice, feedback, and mentorship alongside automation. Institutions could use AI to widen access through translation and assistive functions while giving novices consequential tasks with support. The relevant variable is how assistance changes cognitive activity and participation. Removing an assistive tool during assessment can introduce an access barrier; evaluation must specify the capability of interest and the conditions under which people need to exercise it.
Unprepared identifies a wider problem as human lag: technological and social change can outpace law, curriculum, credentials, and collective understanding (Moravec, 2026b). In the 2030s, such a mismatch could affect the meaning of competence as well as the means of governing technology. A workshop might assess recruits against yesterday’s tasks while assigning today’s work through a changing system. Training could prepare people to follow recommendations without preparing them to investigate unfamiliar failures. Professional communities would then need to revise both routes into expertise and the criteria by which they recognize it.
The relevant timescales differ. A provider can change a service before a workplace completes a training cycle or a professional body revises its standards. In a plausible scenario, novices depend on a tool whose assumptions change between supervised practice and independent responsibility. Families and communities could face a related mismatch when synthetic media and automated persuasion change the information they must evaluate faster than shared practices of verification develop. These are futures questions, extending the report’s account of lag; they do not establish that technological change follows a single rate or that institutions must fail.
Lag becomes more consequential when disruption removes the time, relationships, and alternatives needed to respond. An organization under financial pressure might defer mentorship while buying automation to restore output. A displaced community might depend on a remote service whose terms it cannot negotiate. The developmental consequence would depend on whether those responses preserve opportunities to practice, compare judgments, and question decisions. Examining these interactions shifts the futures inquiry from whether technology advances to how societies sustain and renew human capability as conditions change.
Disruption can amplify either trajectory. Mani et al. (2013) found evidence that financial concerns can burden cognitive performance, including in comparisons of Indian sugarcane farmers before and after harvest. The United Nations Children’s Fund (UNICEF, 2025) estimates that climate hazards disrupted schooling for at least 242 million students in 85 countries or territories in 2024. Its measure concerns schools. Extending the concern to displaced households, libraries, and community networks is an inference about the relationships and places supporting participation.
Consider a further scenario: a community loses a shared learning space during a flood and turns to a remote platform for continuity. The platform restores access, but a funding shortfall prevents the community from rebuilding its space or maintaining another service. Over time, a temporary response becomes dependency. If the provider changes prices or access rules, people have fewer alternatives and fewer local relationships through which to organize a response. Climate disruption, economic constraints, and infrastructure control interact through this sequence.
Political and informational pressures can reinforce these dependencies. In the flood scenario, disputed information about service failures could divide participants, while an authority dependent on the same supplier might resist demands for scrutiny. Polarization could weaken the coalition needed to fund an alternative; loss of that alternative could make refusal more costly. These are plausible mechanisms within a scenario, not documented outcomes for a particular community. They extend Unprepared’s concern with preparation without power: adaptation can restore activity while leaving the rules of participation beyond the reach of those adapting (Moravec, 2026b).
These interacting pressures make the distribution of authority consequential for future learning. Preserving practice, relationships, and viable alternatives requires people who can influence work design, continuity provision, and technical choices. The question of how such influence becomes durable leads from individual agency to collective sovereignty.
From individual agency to collective sovereignty
Individual agency concerns a person’s capacity to understand, choose, and act. Collective agency concerns people’s capacity to organize around shared purposes. Educational sovereignty concerns effective authority over consequential decisions shaping learning conditions. It is a collective and institutional capacity supported by rights, resources, representation, and public obligations, subject to protections for individual freedom.
This differs from institutional autonomy. A university or platform can govern its affairs while giving learners little influence over its decisions. Autonomy concerns an organization’s relationship with external authority. Democratic governance provides the broader principles of participation, representation, and accountability. Educational sovereignty applies those principles to the conditions supporting inquiry, practice, expertise, and the exercise of developed capabilities. Its contribution is to connect the distribution of authority with an account of how people learn, extending the digital enclosure argument in Moravec (2026a) beyond formal education.
Educational sovereignty should also support the creation of new learning possibilities. People need freedom to pursue unfamiliar questions, form new associations, and develop practices that existing institutions have yet to recognize. This extends the normative concern with agency and shared futures in Moravec (2026a, 2026b): collective capability should give people resources to experiment, while individual agency protects their choice to participate, question, or depart. Epistemic integrity requires arrangements that let participants examine evidence, challenge claims, and revise judgments without coercion or deceptive interference. On this account, authority over learning conditions includes the capacity to create opportunities whose purposes and forms remain open to inquiry.
A worker may understand an algorithmic schedule yet lack a procedure for contesting it. A library may understand a restrictive contract yet lack resources to change suppliers. Learning contributes knowledge; organization connects that knowledge with means of action. The missing middle proposed in Unprepared consists of civic intermediaries capable of assembling shared capacity across institutions and communities (Moravec, 2026b). Their value depends on what participants can authorize them to do and what resources give an opposing party reason to respond.
Workplace organization connects learning opportunities with authority over task allocation and technological change. Doellgast et al. (2025) examine social dialogue on AI and algorithmic management across five continents, including employment and skills. Worker representatives sought influence over decisions held by employers through workplace, company, sectoral, and wider arrangements. The comparative findings suggest greater effectiveness where employer exit faces constraints, collective worker voice receives support, and organizing includes groups vulnerable to exclusion. The learning condition at stake is access to skills development within changing work. The evidence concerns technological governance; effects on novices’ expertise require a further inquiry into practice, mentorship, and responsibility.
Protected time makes representation usable. The Advisory, Conciliation and Arbitration Service (2010) sets out reasonable paid time off for eligible union learning representatives in recognized unions to perform learning functions. This gives people resources to consult colleagues and arrange learning support. Public funding can extend that capacity, while exposing it to political decisions: the Department for Education (2020) announced the end of Union Learning Fund support after March 31, 2021. Sustaining learning arrangements therefore requires organization capable of defending resources as well as expertise in using them.
The UC–Elsevier dispute concerned access to scholarly knowledge and the terms under which research would be published. UC held purchasing authority, while Elsevier controlled access to its journal services. Faculty-supported negotiations, institutional refusal, and library arrangements for alternative access enabled UC to sustain a shared position as some content became unavailable. UC announced an agreement with open-access provisions in 2021 (University of California Office of Scholarly Communication, 2019; University of California Office of the President, 2021). These negotiating-party accounts document a changed agreement and the costs of refusal, without isolating the contribution of each tactic or establishing learning effects. Smaller organizations need resources and alliances to sustain comparable opposition.
From voice to enforceable authority
Consultation gives people a voice. Influence means their intervention changes a decision. Enforceable decision rights give them a continuing basis to demand review, obtain a remedy, or alter specified decisions even when an institution resists. These forms of participation can coëxist, but moving from one to another requires a change in the distribution of authority. Educational sovereignty concerns whether people can sustain that authority over learning conditions, rather than depend on the goodwill of a particular official or supplier.
Cohen’s (2019) account of informational capitalism treats legal institutions as participants in struggles over economic and political power. Law can constrain dominance while also authorizing the arrangements through which it develops. This helps explain why a legal right needs examination alongside access to evidence, representation, and enforcement. Ostrom’s (2010) analysis of polycentric governance provides a related basis for examining how multiple centers of authority coördinate and resolve conflict. Neither account warrants assuming that decentralization or a new charter will give an underfunded group leverage over essential infrastructure.
Five mechanisms can give collective organization force. Their relevance depends on which decision shapes a learning opportunity and which actor has the authority to change it.
Legal rights and remedies establish duties that affected people can invoke against an unwilling institution. To challenge exclusion from practice or opaque assessment, people need access to evidence and a body able to order a response. The EU’s Digital Services Act (DSA) illustrates enforceable scrutiny of regulated platforms. In December 2025, the Commission fined X €120 million for breaches concerning account verification, advertising transparency, and researcher access to public data (European Commission, 2025). X Internet and X Holdings brought an action challenging the decision in February 2026 (X Internet and X Holdings v. Commission, 2026). In July, the Commission accepted an action plan addressing advertising transparency and researcher access, with six months for implementation followed by an audit (European Commission, 2026). These developments distinguish a finding of non-compliance from a challenge to that finding and commitments to correction. They do not establish lasting compliance or learning gains. For learning, the proposed connection is that people and independent investigators need usable evidence about decisions shaping what they encounter and can question.
Collective bargaining and organization assemble resources around a shared position. Workers can link demands for mentorship and training to negotiation over work organization; purchasing consortia can combine expenditure with common conditions for access. The ILO comparison and UC dispute identify sources of leverage alongside constraints (Doellgast et al., 2025; University of California Office of Scholarly Communication, 2019). Where bargaining rights are unavailable, a civic intermediary needs another basis for action, such as delegated purchasing authority, legal representation, or a coalition seeking changes to law. Sustaining refusal also requires support for those who bear its costs: alternative access, funded representative time, and shared technical assistance are proposed ways to make participation feasible.
Public procurement and conditional funding connect standards with resources providers seek. A learning consortium could seek inspection rights, continuity provisions, and usable records in its service contracts, within applicable procurement law (Public Procurement Directive, 2014, art. 70). These conditions would help protect access to shared knowledge and relationships when a supplier changes its service. Their force depends on buyers who can monitor performance and respond to breach. In the workshop scenario, purchasing decisions could also require support for supervised practice instead of measuring a system’s value through output alone.
Competition and interoperability requirements can help learning networks preserve relationships and knowledge when changing providers. The EU Digital Markets Act creates portability and specified messaging-interoperability duties for designated services, with privacy and security safeguards (Digital Markets Act, 2022, arts. 6(9), 7). Such duties have a defined scope. Making alternatives usable also requires technical staff, migration funding, and accessible services. The educational purpose is continuity of participation and access to accumulated knowledge when dependence on one provider restricts choice.
Public and coöperative infrastructure can place the tools supporting inquiry under public or member-oriented governance. A 2021 agreement involving Decidim’s free-software association, Barcelona City Council, and the Government of Catalonia documents shared support for participation infrastructure, including code governance and financial contributions (Decidim, 2021). It offers an organizational example, rather than evidence of learning gains. A learning network adopting this approach would need maintenance funding, technical expertise, representative control, and remedies for exclusion. Its educational purpose would be to sustain shared inquiry and intergenerational participation while giving members influence over the tools they use.
These mechanisms can reinforce one another. In a proposed library arrangement, portability obligations would support migration, pooled procurement would give buyers a shared negotiating position, and public funding would help smaller members maintain alternative access. Audit rights would supply evidence of a breach, while legal remedies or contractual sanctions would give the consortium a means to act. The resulting capacity would come from the relationship among rights, resources, and alternatives. Each also creates dependencies, including reliance on regulators, funders, technical staff, and the internal distribution of votes.
Platforms operating across jurisdictions complicate that relationship. EU requirements apply within their defined scope; they do not give every community worldwide equivalent rights. A consortium can coördinate contracts and legal expertise, but cannot create jurisdiction or regulatory capacity by agreement among its members. Where authorities lack resources, refuse to intervene, or become captured, participants may retain knowledge of a harm while losing the means to remedy it. The political task then includes building coalitions for enforcement capacity, defending independent review, and developing alternatives that reduce exposure to unaccountable authority.
Capture also concerns public and civic institutions. The Organisation for Economic Co-operation and Development (OECD, 2017) identifies transparency, integrity, accountability, and participation by groups with different interests as elements of prevention. Applied here, those principles require scrutiny of who appoints reviewers, controls funding, and can challenge a regulator or intermediary. A community-controlled body may suppress dissent or impose monitoring. Member rights, independent review, and access to remedies beyond the body are proposed safeguards. Collective authority gains legitimacy through the opportunities individuals retain to question its exercise.
Private inquiry and institutional scrutiny
Exercising these rights requires information about the decisions people seek to challenge. A person cannot contest a restriction whose basis remains inaccessible, while a platform may possess detailed records of that person’s behavior. This imbalance links privacy with authority. Asymmetric transparency identifies the principle applied here: people should retain substantial privacy in personal learning, while institutions exercising consequential power over learning conditions should face proportionate obligations to explain, justify, and permit independent scrutiny of their decisions.
This application draws on established scholarship. Nissenbaum’s (2004) contextual integrity ties privacy to appropriate information flows within a social context; privacy requires judgments about purposes and relationships, rather than a choice between total secrecy and disclosure. Cohen (2019) examines the legal organization of informational power, while Kroll et al. (2017) connect algorithmic accountability with technical and institutional means of verification. Asymmetric transparency applies these established concerns to the governance of learning conditions.
Three forms of information require different treatment. Personal learning activity includes reading, conversations, experimentation, and private development. Its governance should emphasize consent where appropriate, data minimization, and freedom from compulsory profiling. Institutional decision-making includes access rules, procurement, algorithmic systems, and assessment policies. It calls for reasons, accountability, and routes to contest decisions. System-level impacts include patterns of exclusion, bias, interruption, and institutional harm. They require independent evaluation with safeguards suited to the information and the risks of disclosure.
These distinctions set different burdens. A library user should be able to inquire without documenting progress. The library should explain its access rules and allow scrutiny of consequential procurement decisions. Researchers could evaluate exclusion through proportionate sampling and protected access to relevant data. Public reporting might use aggregates, but small groups can remain identifiable, so aggregation needs examination alongside access controls and disclosure risks. The proposed principle protects people while making institutional responsibility examinable.
Voluntary privacy commitments contribute to this arrangement. The American Library Association’s (2025) policy supports opt-in choices for nonessential collection and anonymous or aggregated analysis. Institutional promises, however, leave the terms of scrutiny with the institution when it can refuse access. Independent oversight needs authority to obtain evidence, examine confidentiality claims, and require correction. A learning intermediary should therefore secure inspection and reporting duties before it depends on a supplier, and connect those duties with a remedy for obstruction.
EU regulation illustrates how scrutiny and privacy can coëxist. The DSA establishes independent audits and routes for regulator and vetted-researcher access to data from designated services; confidentiality and personal-data protections constrain that access (Digital Services Act, 2022, arts. 37, 40). Commission guidance requires providers to justify redactions rather than withhold an entire risk assessment (European Commission, n.d.). Applied to learning systems, the proposed arrangement would combine protected inspection with public findings that people can use to question consequential decisions.
The General Data Protection Regulation provides another example through high-risk processing assessments and safeguards for certain consequential automated decisions (General Data Protection Regulation, 2016, arts. 22, 35). These obligations have legal limits. The educational proposal is to connect understandable reasons and independent review with remedies through applicable law and enforceable agreements. Its purpose is to make exclusion from practice, recognition, or responsibility contestable while protecting the learner’s personal inquiry.
In the workshop, beginners should be able to discuss uncertainty and test explanations without turning every provisional judgment into a permanent performance profile. The employer should still permit scrutiny of how its system allocates tasks, selects workers for mentorship, and evaluates competence. Researchers could study developmental effects with consent and suitable safeguards, while authorized reviewers examine consequential institutional decisions. This distinction preserves room to learn through mistakes while making the organization answerable for the opportunities it allocates.
Proprietary secrecy and technical complexity can still obstruct accountability. Source code may help an inspection, but a system’s effects also depend on data, configuration, interfaces, deployment, and organizational practices. Kroll et al. (2017) develop accountability approaches that extend beyond publishing code, connecting technical verification with legal and policy objectives. For learning, an audit should test providers’ claims against the decisions their systems support and the effects observed in use.
Inspection should follow a service through deployment and consequential changes, with relevant evidence available to reviewers and failures connected to correction. Continuing inspection duties require a legal or contractual basis; reviewers require resources, competence, and protection from conflicts of interest. Due process also protects those whose decisions are challenged. The point for learning is to discover whether changing tools and organizational practices preserve routes into expertise, distribute opportunities fairly, and let people act on developed judgment.
What must we protect?
The practical agenda follows from the connections among cognition, participation, and power. Institutions should protect access to meaningful practice, the continuity of relationships and shared spaces, freedom to inquire, and effective influence over consequential decisions. The resulting agenda concerns how to sustain developmental opportunities while making their allocation contestable.
For workplaces, this means treating mentorship and entry into consequential work as matters of work design and negotiation. For libraries and communities, it means funding access, shared space, and alternatives to restrictive infrastructure. For schools, it means sustaining foundational knowledge and skilled teaching while connecting with the activities through which learners extend their understanding. New York City community schools illustrate such coördination: a quasi-experimental evaluation estimated attendance improvements across three study years, while reading achievement estimates were not statistically significant (Johnston et al., 2020). The case shows why participation and learning need related but distinct measures.
Public and civic institutions should make those responsibilities enforceable through the mechanisms suited to their authority. Members need a role in directing shared resources, affected people need reasons and remedies, and reviewers need protected access to relevant evidence. Oversight should examine institutional decisions and effects while preserving private inquiry. Recognition of competence remains valuable where people seek it or a defined responsibility requires it; participation in everyday learning should carry no general duty to create a credential or permanent record.
Informal learning remains open to systematic study. With appropriate consent and research safeguards, investigators can examine retained knowledge, changes in practice, and participation. Institutional evaluation can examine access barriers, contract terms, and the response to complaints. The two inquiries connect when a change in governance alters opportunities to practice or participate, and research examines what people develop through those opportunities. Research can then connect institutional changes with the development, distribution, and exercise of capability.
Institutions shape the development of capability through opportunities to practice, its distribution through access and recognition, and its exercise through resources and decision rights. In the workshop, automation could improve immediate performance while weakening the route from beginner to responsible practitioner. Preserving the reproduction of expertise therefore requires attention to the relationships and experiences through which successive generations develop judgment, alongside the output a system produces.
Educational sovereignty concerns enforceable authority over these learning conditions. People need means to sustain and create opportunities, question their allocation, and change arrangements that restrict participation. Protecting individual inquiry and subjecting institutional decisions to scrutiny connect that collective authority with freedom and epistemic integrity.
Beyond 2030, preparation will depend partly on whether societies maintain routes into expertise while giving people authority to shape how their capabilities develop and are used. The proposed framework connects those responsibilities across institutions. Its normative proposition is that people should retain both opportunities to learn and the collective capacity to participate in creating their shared futures.
References
Advisory, Conciliation and Arbitration Service. (2010, January 1). Code of Practice on time off for trade union duties and activities including guidance on time off for union learning representatives. https://www.acas.org.uk/acas-code-of-practice-on-time-off-for-trade-union-duties-and-activities/html
American Library Association. (2025, June 29). Privacy: An interpretation of the Library Bill of Rights. https://www.ala.org/advocacy/intfreedom/librarybill/interpretations/privacy
Bastani, H., Bastani, O., Sungu, A., Ge, H., Kabakcı, Ö., & Mariman, R. (2025). Generative AI without guardrails can harm learning: Evidence from high school mathematics. Proceedings of the National Academy of Sciences, 122(26), Article e2422633122. https://doi.org/10.1073/pnas.2422633122
Cohen, J. E. (2019). Between truth and power: The legal constructions of informational capitalism. Oxford University Press. https://academic.oup.com/book/37371
Competition and Markets Authority. (2025, July 31). Cloud services market investigation: Summary of final decision. https:/
Decidim. (2021, February 9). The Government of Catalonia, Barcelona City Council and the Free Software Association Decidim sign a collaboration agreement to promote democratic participation with free and open technologies. https://decidim.org/blog/2021-02-09-the-government-of-catalonia-barcelona-city-council-and-the-free-software-association-decidim-sign-a-collaboration-agreement-to-promote-democratic-participation-with-free-and-open-technologies/
Department for Education. (2020, December 9). Union Learning Fund (Written question 123731) [Written answer]. UK Parliament. https://questions-statements.parliament.uk/written-questions/detail/2020-12-01/123731/
Digital Markets Act, Regulation (EU) 2022/1925, 2022 O.J. (L 265) 1. https://eur-lex.europa.eu/eli/reg/2022/1925/oj
Digital Services Act, Regulation (EU) 2022/2065, 2022 O.J. (L 277) 1. https://eur-lex.europa.eu/eli/reg/2022/2065/oj
Doellgast, V., Appalla, S., Ginzburg, D., Kim, J., & Thian, W. L. (2025). Global case studies of social dialogue on AI and algorithmic management (ILO Working Paper No. 144). International Labour Organization. https://doi.org/10.54394/VOQE4924
European Commission. (n.d.). Q&A on risk assessment reports, audit reports and audit implementation reports under DSA. https://digital-strategy.ec.europa.eu/en/faqs/qa-risk-assessment-reports-audit-reports-and-audit-implementation-reports-under-dsa
European Commission. (2025, December 5). Commission fines X €120 million under the Digital Services Act [Press release]. https://digital-strategy.ec.europa.eu/en/news/commission-fines-x-eu120-million-under-digital-services-act
European Commission. (2026, July 16). Commission accepts X’s action plan to comply with Digital Services Act [Press release]. https://digital-strategy.ec.europa.eu/en/news/commission-accepts-xs-action-plan-comply-digital-services-act
Federal Trade Commission. (2019, December 6). FTC issues opinion and order against Cambridge Analytica for deceiving consumers about the collection of Facebook data, compliance with EU-U.S. Privacy Shield [Press release]. https://www.ftc.gov/news-events/news/press-releases/2019/12/ftc-issues-opinion-order-against-cambridge-analytica-deceiving-consumers-about-collection-facebook
General Data Protection Regulation, Regulation (EU) 2016/679, 2016 O.J. (L 119) 1. https://eur-lex.europa.eu/eli/reg/2016/679/oj
Johnston, W. R., Engberg, J., Opper, I. M., Sontag-Padilla, L., & Xenakis, L. (2020). Illustrating the promise of community schools: An assessment of the impact of the New York City Community Schools Initiative (RR-3245-NYCCEO). RAND Corporation. https://doi.org/10.7249/RR3245
Karpicke, J. D., & Roediger, H. L., III. (2008). The critical importance of retrieval for learning. Science, 319(5865), 966–968. https://doi.org/10.1126/science.1152408
Kroll, J. A., Huey, J., Barocas, S., Felten, E. W., Reidenberg, J. R., Robinson, D. G., & Yu, H. (2017). Accountable algorithms. University of Pennsylvania Law Review, 165(3), 633–705. https://ir.lawnet.fordham.edu/faculty_scholarship/889/
Lave, J., & Wenger, E. (1991). Situated learning: Legitimate peripheral participation. Cambridge University Press. https://doi.org/10.1017/CBO9780511815355
Mani, A., Mullainathan, S., Shafir, E., & Zhao, J. (2013). Poverty impedes cognitive function. Science, 341(6149), 976–980. https://doi.org/10.1126/science.1238041
Moll, L. C., Amanti, C., Neff, D., & González, N. (1992). Funds of knowledge for teaching: Using a qualitative approach to connect homes and classrooms. Theory Into Practice, 31(2), 132–141. https://doi.org/10.1080/00405849209543534
Moravec, J. W. (2026a). Beyond the digital enclosure: AI, coloniality, and the pursuit of educational sovereignty. Learning Futures and Emerging Technologies. Advance online publication. https://doi.org/10.1108/LFET-01-2026-0004
Moravec, J. W. (2026b). Unprepared: Education beyond 2030. Education Futures. https://educationfutures.com/unprepared-education-beyond-2030/
National Academies of Sciences, Engineering, and Medicine. (2018). How people learn II: Learners, contexts, and cultures. The National Academies Press. https://doi.org/10.17226/24783
Nissenbaum, H. (2004). Privacy as contextual integrity. Washington Law Review, 79(1), 119–158. https://digitalcommons.law.uw.edu/wlr/vol79/iss1/10/
Organisation for Economic Co-operation and Development. (2017). Preventing policy capture: Integrity in public decision making. OECD Publishing. https://doi.org/10.1787/9789264065239-en
Ostrom, E. (2010). Beyond markets and states: Polycentric governance of complex economic systems. American Economic Review, 100(3), 641–672. https://doi.org/10.1257/aer.100.3.641
Pennycook, G., & Rand, D. G. (2022). Accuracy prompts are a replicable and generalizable approach for reducing the spread of misinformation. Nature Communications, 13, Article 2333. https://doi.org/10.1038/s41467-022-30073-5
Public Procurement Directive, Directive 2014/24/EU, 2014 O.J. (L 94) 65. https://eur-lex.europa.eu/eli/dir/2014/24/oj
Sen, A. (1999). Development as freedom. Alfred A. Knopf.
United Nations Children’s Fund. (2025). Learning interrupted: Global snapshot of climate-related school disruptions in 2024. https://www.unicef.org/reports/learning-interrupted-global-snapshot-2024
University of California Office of Scholarly Communication. (2019, March 1). UC and Elsevier: FAQs. https://osc.universityofcalifornia.edu/2019/03/uc-and-elsevier-faqs/
University of California Office of the President. (2021, March 16). UC secures landmark open access deal with world’s largest scientific publisher. https://www.universityofcalifornia.edu/press-room/uc-secures-landmark-open-access-deal-worlds-largest-scientific-publisher
X Internet and X Holdings v. Commission, Case T-114/26, 2026 O.J. (C/2026/2390) (action brought February 16, 2026). https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:62026TN0114
