AI-Driven Model for Fostering International Cooperation in Open Education: A Roadmap for Innovation and Equity

Synthesis: Patiño, Martínez-Pérez, Ramírez-Montoya, and Miranda (2026) ask how AI can support international cooperation in open education, and answer with an instrumental case study of the UNESCO/ICDE 2025 Residency rather than an experiment. The case is a two-week residency of 110 participants from 13 countries in 10 interdisciplinary teams across 52 organizations, documented through team logs, a 22-item questionnaire that drew 72 responses, and a co-authored manifesto, all coded qualitatively. The result is a framework, not an effect: ten fundable project proposals spread across the five UNESCO action areas, and three interdependent domains, Enabling Conditions, Cooperative Processes, and Sustainability Mechanisms, with AI as transversal coordinator. It can show that a policy-aligned residency produced these artifacts and self-reported competency gains. With no comparison group, no pre-measure, and no learner data, it cannot show that cooperation or equity improved.

Key Findings

1. The residency convened 110 academics from 13 countries (Argentina, Chile, Colombia, Costa Rica, Germany, Guatemala, Spain, the United Arab Emirates, the United States, Italy, Mexico, Peru, and the United Kingdom) and organized them into 10 interdisciplinary teams.
2. Those teams produced 10 research, innovation, and development projects written to attract funding, plus a co-authored Manifesto Chapter Book released as an open access resource.
3. Country participation, mapped against the five UNESCO action areas, was broadest for Mexico, Peru, and Colombia, which appear across all five, while Spain, Chile, and the United Kingdom are described as connectors between thematic clusters.
4. The analysis condensed cooperation into three domains: Enabling Conditions (policy alignment and institutional readiness), Cooperative Processes (team formation, participatory design, shared authorship), and Sustainability Mechanisms (stakeholder engagement, contextual anchoring, AI-enabled dissemination).
5. The post-residency survey drew 72 responses from 54 institutions in 13 countries, reporting open education, formative practices, and open science as the dominant themes and research, digital fluency, and collaboration as the competencies most strengthened.
6. AI appears in the model in three roles rather than one: an Artificial Intelligence Digital Advisor offering 24/7 support, analytics that helped self-organized teams form around shared interests, and curation and audience segmentation for dissemination.
7. The authors anchor the problem in the UNESCO implementation report of 2023, where 78 of 194 Member States replied, only 49 (63% of respondents) reported signed cooperation agreements, 22 (28%) reported OER-specific cooperation clauses, and 49 reported peer networks sharing OER.

A residency built to enact a declaration

The design follows Stake's instrumental case logic: the residency is interesting not for itself but as a bounded instance through which the mechanics of cooperation can be observed. Its structure was mapped onto the five strategic components of the Dubai Declaration, and those components then served as sensitizing concepts for the coding, which is the paper's central methodological choice. Networks, engagement with the open community and licensing lawyers, intersectoral research cooperation, repository and policy frameworks, and an AI-supported co-creation laboratory each became both a design element and an analytic category.

That alignment is a strength and a circularity at once. Because the intervention was built from the declaration's components, finding that the residency enacted them is close to a restatement of the design. What the case adds is operational detail: interdisciplinary teams formed without pre-assignment, facilitators pushed for disciplinary and geographic mix, and shared authorship was treated as the mechanism that converted participation into ownership. The concrete outputs, ten proposals and a manifesto, are then read as evidence that the model works.

Data came from three sources: the team registry and mobilization log kept by organizers, the 22-item questionnaire administered after the residency, and the manifesto chapters. Content analysis followed L'Écuyer's three phases in QDA Miner 5, with a second researcher coding independently and disagreements resolved by consensus. The paper reports no reliability coefficient for that agreement.

Three domains, held together by AI

The model's claim is that the three domains are interrelated rather than sequential, and the implementation guidance attached to each is the most transferable part of the paper. Policy alignment comes first, on the argument that institutional agendas must match global frameworks before operational steps make sense. Institutional readiness follows, through capacity mapping and team formation. Sustainability is designed in from the start rather than added at the end, with institutional leaders recruited as co-owners and local validation built into project design.

AI's position in this structure is deliberately modest in the text even where it is prominent in the framing. It coordinates and synthesizes, it tracks decisions, it curates dissemination, and it offers always-available advice. The paper's stronger claim is that AI has so far been applied to open education platforms, content creation, and learner support, but not to modeling cooperation itself, and that the roadmap fills that gap. Whether a model that leans on a digital advisor is different in kind from a project management platform with a chatbot attached is a question the case does not test.

The heterogeneity argument deserves separate notice. Fifty-two organizations and thirteen countries are treated as a strategic resource rather than a nuisance, on the grounds that cross-sectoral mix produced the cross-pollination the projects needed. That is a plausible mechanism, and it is asserted from the organizers' vantage point rather than measured.

The survey carries less than the conclusions ask of it

The questionnaire is the only source that reports participant perception, and its numbers are thin for the weight placed on them. Seventy-two responses from 110 participants are reported without a response rate, and the analysis is inductive coding of themes and competencies, presented in two figures rather than as counts. There is no pre-residency baseline, so "most strengthened" is a retrospective judgment. There is no comparison with a residency that lacked AI, so no attribution to the digital advisor is possible.

Two small internal discrepancies suggest the reporting was assembled from separate project documents. The method section counts 52 organizations among participants; the results section counts 54 institutions among survey respondents. Both may be accurate for their own source, but the paper does not reconcile them, which matters for a claim about institutional reach.

The equity argument is where the gap between evidence and aspiration is widest. Equity is named in the title and threaded through the framing, yet the study measures no access, cost, completion, or outcome variable that would show whether any learner was better served. The country list is dominated by Latin America and Europe, with the United Arab Emirates the only participant from outside those regions, and the residency is run by a Latin American chair. That is a legitimate regional base, and it is not evidence of global distribution.

What this means for practice

Instructors reading this should take the process design rather than the model's authority, and the most usable element is the ordering: settle policy alignment before recruiting teams, recruit for disciplinary and geographic mix, and treat shared authorship as a deliverable rather than a courtesy. The co-creation laboratory described here, with AI synthesizing discussion and logging decisions, is reproducible at department scale without a UNESCO mandate.

For faculty developers and international office staff, the paper's contribution is a planning checklist that names the failure modes it saw: teams that never co-write anything, dissemination planned after the fact, and institutional sponsors brought in when the project is already designed. For administrators, the honest reading is that the roadmap is a hypothesis about how cooperation is sustained, supported by one well-documented residency and by participant self-report, and that any institution adopting it should collect the outcome data this study did not.

Limitations

• The design is a single instrumental case with no comparison condition, so the three domains are an interpretation of one residency rather than a tested model, and the paper's own conclusion names the single-case design and self-reported data as the binding constraints.
• Participant perception rests on 72 questionnaire responses from 110 participants with no stated response rate and no baseline, leaving pre-post claims about competency growth unsupported.
• The residency was designed from the same Dubai Declaration components later used to code it, so the fit between framework and data is partly built in rather than discovered.
• No learner-level data were collected at all: equity, access, and learning outcomes are framed as goals of the model but never measured, and the country distribution is concentrated in Latin America and Europe.

Citation

Patiño, A., Martínez-Pérez, S., Ramírez-Montoya, M.S., & Miranda, J. (2026). AI-Driven Model for Fostering International Cooperation in Open Education: A Roadmap for Innovation and Equity. Journal of Interactive Media in Education, 2026(1), 11, pp. 1-19.