Research Article
From Connectivity Toward Shared Governance: Rethinking Transnational Faculty Development in the GenAI Era
Synthesis: Tian, Sun and Wang (2026) argue that the internationalization of faculty development rests on an unexamined premise: that connecting teachers across borders produces collaboration. Connectivity, they write, is "episodic by default" — a webinar distributes information without establishing a shared agenda, a platform hosts resources without agreement on how to interpret them, and an international event generates visibility without assigning responsibility for what happens next. Their claim is that faculty development in the generative AI era is as much an organizational coordination problem as a professional-learning one, and they set out five priorities for transnational programs: translating broad ambitions into actionable agendas, building temporal continuity and organizational memory, keeping pedagogical and ethical judgment alongside technical experimentation, preserving distributed participation and local adaptation, and converting events into reusable institutional infrastructure. They use "shared governance" as a direction of travel rather than an achieved state — enough coordination to sustain common work, enough autonomy for partners to adapt it. The Digital Intelligence International Development Alliance and the International Education Leadership Program are offered as an illustrative setting, not as evidence, and the authors say so more than once.
No primary data: this is a conceptual commentary built from policy documents, published scholarship, and the authors' own institutional experience. The alliance described below is presented by staff of the university that convenes it as an illustration rather than a case study, and no faculty development activity in it is evaluated.
Key Findings
- Connectivity is not collaboration, and the gap is institutional rather than technical. Webinars, online communities, shared platforms and hybrid conferences have changed the scale at which faculty meet ideas, at low cost and high speed. The authors do not treat that access as an outcome. "The problem is not a failure of technology, but a gap between access and institutionalization": a webinar can disseminate without setting an agenda, and a platform can host without producing agreement on use. Durable professional learning needs continuity, common reference points, institutional support and room for local adaptation, none of which a connection supplies by itself. Hence the doubled framing — instructional task and organizational coordination task — since only the second produces arrangements that outlast an event.
- GenAI intensifies the coordination problem in four ways, none of them unique to it. Competency expectations are unusually unstable, so faculty agendas must move faster than multi-year training cycles. Normative and regulatory divergence is visible across systems — privacy, platform access, assessment, intellectual property, disclosure — so an institution's acceptable practice cannot be assumed portable, a point they ground in comparative work on university AI guidance (Wu et al., 2024) and on assigning responsibility when AI enters professional judgment (McGrath et al., 2023). Capacity is uneven, and a network organized around its best-resourced members reproduces center–periphery dynamics. And technological time outruns organizational time: tools change in weeks, committees and budgets in years.
- Broad ambitions have to be translated into something a faculty member can do in a session. "Responsible AI" becomes work on assessment redesign, acceptable use, data protection, bias, disclosure and disciplinary judgment; AI literacy is layered for novices, experienced adopters, instructional leaders and policy staff. Translation is not standardization — it creates an operational starting point that permits comparison across contexts, not a definition imposed on partners. That distinction locates the coordinating work in the design of agendas rather than the enforcement of vocabulary, and it gives partners something concrete to disagree with.
- Continuity requires organizational memory, and memory requires infrastructure. One-off events buy visibility and entry; a rhythm of recurring workshops, follow-ups and annual cycles is what lets activities accumulate. The authors pair the rhythm with an archive — agendas, guidance notes, case examples, recordings, reusable resources — that serves as a reference point as technology and policy move. Without it, each cohort starts from zero and each new tool reopens questions the network already answered. Their reading of digital transformation as the management of persistent tensions rather than their resolution (Toh & Looi, 2024) is why they want recurring spaces for recurring questions, not a curriculum finished once and delivered thereafter.
- Divergence is a resource only if it is curated; otherwise it is friction. The aim is not to eliminate differences among institutional policies but to make them discussable and usable as learning resources: comparing how two universities draw a boundary around disclosure reveals the local values behind each line more instructively than a harmonized statement that hides them. This is why every technical conversation is paired with pedagogical questions — what problem the tool addresses, what risks it creates, which assessments remain valid, what should be disclosed, what stays under human control. Tool-centered training divorces technology from purpose, and the fix is not less technical content but judgment taught beside it.
- Shared governance means coordination plus distributed ownership, judged by whether its conditions are strengthening. Over-centralization weakens ownership, and a network is fragile when partners only receive content produced elsewhere; so a hub supplies continuity and common structure while partners keep authority over substance, contributing cases, leading sessions, challenging assumptions. Shared governance, they insist, is a direction of travel rather than a label: recurring meetings or a coordinating institution do not by themselves demonstrate shared decision-making. The test is whether continuity, reciprocal influence, distributed contributions, transparent coordination and capacity for local adaptation are improving — and the same test should replace attendance as the measure of an alliance.
What this means for practice
- Administrators: fund the secretariat, not the summit. A network that spends its budget on an annual event and none on coordination has bought visibility. The authors' five priorities imply a standing function — someone whose job is the rhythm, the archive, and the follow-up — and that function is what makes the event worth holding.
- Faculty developers: write the agenda at session granularity. Replace "responsible AI" on the program with the specific questions a teacher faces next week: what to permit in this assignment, what to disclose, what to do about a detection claim, which disciplinary judgment the tool cannot make. Abstract ambition is not a curriculum.
- Coordinating hubs: give every partner a deliverable, not an invitation. Ask each institution for a case, a policy comparison, or a session it leads, and put the contribution in the program. Distributed participation is produced by the design of the work, not by a statement of intent.
- Whoever owns the archive should date-stamp it. Shared resources need an owner, a review date and a visible revision history, or they become a record of what the network believed in its first year. Build the revision cycle into the calendar, not into good intentions.
- Evaluate the network by reuse, not by attendance. Count repeat participants, local adaptations, and resources that were revised by someone other than their author. Those numbers describe whether the collaboration has become infrastructure.
Limitations
- There is no case, only an illustration. DI-IDEA and IELP are described from publicly available alliance materials, with no interviews, no document analysis and no outcome for any activity. The authors state this plainly, but the five priorities remain a synthesis of experience rather than a tested design.
- The authors are describing their own institution's alliance. All three work at Peking University, in the Office of International Relations, the Office of the Provost and the DI-IDEA Secretariat. The paper's caution about center–periphery dynamics applies to its own authorship.
- "Shared governance" is defined by contrast and never operationalized. It is not the faculty-senate model, not uniformity and not an achieved state. What remains is a direction and a list of conditions, with no threshold, no scorable indicator and no account of who decides when coordination has tipped into over-centralization.
- The four pressures are asserted as intensifiers, not shown to be new. Regulatory divergence, uneven capacity and slow committee cycles long predate GenAI; the paper does not establish which of them GenAI worsens, or by how much, and the center–periphery risk is named without documentation.
- Cost, staffing and faculty time are absent. A hub, a recurring calendar, an archive and a comparative policy library imply permanent labor. The paper does not say who pays, how an institution without an international office would join, or what evidence would show the investment paid off.
- No failure cases. Every example is of a network doing the right thing; there is no account of a collaboration that collapsed, or of governance that suppressed the local adaptation the authors value, which leaves the priorities hard to falsify.
Citation
Tian, M., Sun, H., & Wang, X. (2026). From Connectivity Toward Shared Governance: Rethinking Transnational Faculty Development in the GenAI Era. Frontiers of Digital Education, 4(1), article 10.