Three Countries, Sixteen Experts in Wulai: Five Cards for Returning Knowledge, Cultural Authority, and Youth Livelihoods
Original Chinese title: 3國16位專家走進烏來：五張知識回流卡讓海外館藏接回文化主體與青年生計
A two-day Wulai forum brought together sixteen experts from Austria, Japan, and Taiwan to discuss overseas collections, digital sharing, traditional skills, youth livelihoods, and cultural economy.
Yuan Media AI Editorial Desk | AI-Driven Indigenous Township Economy and Policy Watch
The desk follows Indigenous cultural economies, knowledge governance, museum practice, youth livelihoods, data rights, and AI boundaries.
On October 22 and 23, 2026, the New Taipei City Indigenous Peoples Department convened an international forum in Wulai on Indigenous museums and cultural-economy industries. The official release reports participants from Austria, Japan, and Taiwan, including sixteen scholars, museum leaders, researchers, cultural workers, and local practitioners.
Those figures describe the scale of participation, not the completion of restitution, community authorization, or durable jobs. Responsible reporting must separate exchange that has taken place, research still under way, and cooperation that remains only a possibility.
The forum continued cooperation that followed the Wulai Atayal Museum's 2025 research visit to the Weltmuseum Wien. Museum specialists discussed historical archives, collections research, textile conservation, digital sharing, and long-term work with Indigenous source communities.
The theme, “Revolving,” draws on the motion of Atayal weaving: knowledge crosses distance and time and returns with new interpretations and relationships. Governance must still ask what returns, who receives it, who can decide its use, who performs the work, and how benefits flow back.
Knowledge can return before a physical object moves. High-resolution images, material records, collection histories, conservation observations, oral identifications, and making techniques can support local research and teaching. But digital availability does not automatically return authority.
A platform may preserve colonial-era labels, uncertain attributions, or restrictive institutional terms. If a community may view a record but cannot annotate it, correct it, limit sensitive access, or shape reuse, the flow remains one-directional.
Cultural economy must not be reduced to turning objects into products. Conservation, community learning, youth livelihoods, and tourism sales have different purposes. Some designs, names, narratives, or techniques may be public; others may be restricted by family, ceremony, gender, age, or relationship.
Short exhibitions, tours, workshops, and reproductions can create income and visibility without creating durable work. Evaluation should include year-round workload, fair pay, materials, workspace, care duties, social protection, learning time, intellectual rights, and the ability of young people to negotiate price and terms.
Five knowledge-return cards can turn the forum's vision into an auditable process. They do not decide cultural meaning for communities. They connect provenance, permission, co-research, translation, and return across research, digitization, exhibitions, education, and enterprise.
The first card records provenance and context: current and possible original names, collection date, collector, mode of acquisition, location, identifier, earlier classification, missing fields, and competing accounts. Unknown information stays unknown instead of being guessed.
Evidence is layered. Institutional archives, object study, publications, community testimony, family memory, and later research can coexist with dates and contributors. Disagreement is knowledge to preserve, not an error to erase. Every change retains the earlier version and a reason.
The second card governs permission and sensitivity. It states who may view, download, annotate, publish, or use material commercially; whether renewed consent is required; and how conditions can change or be withdrawn. Text, image, audio, three-dimensional models, and technique descriptions may need different access levels.
Permission cannot be a museum-only checkbox. Institutions explain legal and conservation duties, while source communities participate through a process that can change outcomes. When authority remains unclear, the system uses a conservative access level and provides a correction channel.
The third card structures co-research. It records who framed the question, who participates, how decisions are made, what translation and travel support exists, how time is paid, who reviews outputs, and how disputes are handled. Participation begins with problem setting, not a consultation at the end.
This card also recognizes maintenance and care labor. Weavers, elders, translators, young people, museum staff, and researchers all contribute time. Repeated identification, demonstration, or hosting requires support and reasonable compensation; cultural responsibility is not free labor.
The fourth card covers cultural translation and livelihood. Each exhibition, class, tour, reproduction, or product identifies its cultural purpose, audience, disclosure boundary, material source, labor time, cost, price, attribution, and benefit arrangement. Market access does not turn every design into a reusable asset.
Livelihood measurement combines revenue with conditions. It asks how much work is temporary or recurring, whether participants can negotiate, who bears equipment and venue costs, who owns remaining works and digital files, and how cancellation risk is shared. Dignified work and cultural authority matter alongside sales.
The fifth card governs return and correction. Every project specifies the material to be returned, file format, language, date, repository, and responsible person. After publication, community members can request correction, renaming, additional context, reduced access, or removal, and the institution reports the status.
Results extend beyond outputs. Indicators can include locally usable records, corrected errors, paid learning and work, teaching materials returned to schools and communities, sensitive content appropriately restricted, and reasons cooperation stalled. Unsuccessful outcomes remain part of institutional learning.
All five cards share identifiers, dates, and version history while separating access. Conservators see what preservation requires; researchers receive only approved scope; practitioners use authorized material; the public sees reviewed descriptions; designated community participants can propose annotations and access changes.
AI can support low-risk tasks: comparing catalogues, flagging uncertain fields, drafting translations, linking identical identifiers, summarizing version differences, transcribing authorized public audio, and reminding partners of return dates. Every output cites evidence and uncertainty and receives authorized human review.
AI must not infer ethnicity, family authority, or sacred status from appearance; treat historical labels as truth; invent missing provenance; present generated images as traditional objects; or send restricted images, voices, and techniques to unapproved models. Cultural sensitivity and permission remain human decisions with paths for objection and withdrawal.
A minimum pilot can begin with one group of textiles: create provenance, permission, and co-research records; choose one community-approved public output; document labor, attribution, pay, and return; and review accuracy, workload, and access after three months.
Public agencies and museums can publish de-identified quarterly progress on provenance checks, permissions, co-research, returns, and corrections without exposing sensitive content. External research and commercial partners should disclose purpose, data scope, benefit terms, and exit procedures before access.
Three countries, sixteen experts, a two-day forum, and cross-museum cooperation are an opening. The test comes afterward: whether knowledge returns in locally usable form, descriptions can be jointly corrected, cultural teaching receives time and pay, young people gain sustainable work, and authorized people retain control.
Five cards make “return” a set of responsibilities: provenance has context, use has permission, research has shared authority, translation has conditions, and outcomes return to the people concerned. Only then can cultural revitalization and cultural economy reinforce rather than consume one another.
Sources
AI use and content-safety disclosure
AIGC assisted in organizing public material from New Taipei City Government and the Wulai Atayal Museum. AI is limited to source comparison, missing-field prompts, translation drafts, version summaries, and deadline reminders; it does not replace community authorization, cultural-sensitivity decisions, collections rights, research review, or benefit arrangements.