Pacific LCIPP Gathering in Fiji Moves from Climate Observations to Community-Led Solutions: A Resilience Action-Card Model for Taiwan’s 55 Indigenous Townships
Original Chinese title: 太平洋LCIPP正在斐濟交流社群氣候解方：55原鄉可把部落觀察、傳統知識與AI工具接成韌性行動卡
The UNFCCC LCIPP Regional Gathering for the Pacific is taking place in Nadi, Fiji, from August 18 to 21. On August 19, the agenda moves from long-term place-based observations toward regenerative, community-led solutions, stewardship, intergenerational knowledge transmission and enabling conditions. Taiwan’s 55 Indigenous townships can adapt this approach through governed resilience action cards that connect community observations, official climate data, permissions, local actions and review cycles before adding GIS, RAG or AIGC.
Yuan Media AI Editorial Desk
Covers official notices across Taiwan’s 55 Indigenous townships, Indigenous education, language technology, AIGC, Taitung agriculture, local economic resilience, traditional knowledge governance and digital public services.
Track 2 assessment: no person-focused article is recommended today. A review of public information from Legislator Chen Ying’s office and Legislative Yuan sources for the previous two days did not identify a new August 18–19 item that was both fully verifiable and substantial enough to support a standalone Indigenous economy or technology policy article. Today’s observation therefore uses a Track 1 international Indigenous climate-resilience case rather than forcing a person-focused story without sufficient new official material.
The UNFCCC Local Communities and Indigenous Peoples Platform, or LCIPP, is holding its Regional Gathering for the Pacific in Nadi, Fiji, from August 18 to 21, 2026. The official programme places knowledge, capacity for engagement, climate policy and action in one framework, with cultural continuity, lived experience and diverse knowledge systems at the centre of the 2026 gathering. August 18 uses Talanoa dialogue to surface long-term, place-based observations of climate change. On August 19, the programme moves from observed impacts toward regenerative, community-led solutions, stewardship of land and ocean, intergenerational knowledge transmission and the partnerships and enabling conditions needed for implementation. The afternoon also includes cultural immersion with Naidiri Village and a marine biodiversity park. See UNFCCC | LCIPP Regional Gathering for the Pacific.
For Taiwan’s 55 Indigenous townships, the useful lesson is the sequence: first make local observations visible and traceable; next document the solutions communities already use, along with permissions and conditions; only then decide whether GIS, RAG, AIGC or another digital tool is actually needed. This sequence is easier to maintain than beginning with a large AI platform and then trying to retrofit governance later. It also gives knowledge holders, township staff, farmers, fishers, forest users and young people a clearer role in deciding how climate information should be interpreted and used.
Layer 1: Turn climate experience into resilience action cards, not only story summaries
A township can begin with one concrete situation, such as road and water access after extreme rain, hillside crop timing, stream conditions, coastal weather and nearshore fishing, wild-food and gathering seasons, forest-fire risk, crop pests, or relationships between ceremonies and seasonal change. Each resilience action card can preserve an observation date, location precision, observer or source, the change that was noticed, an existing community response, disclosure level, whether a second review is required, related official information, a practical local-service action and the next review date. The aim is not to convert living knowledge into a single technical score. It is to make provenance, context and responsibility visible enough for local action.
This structure helps prevent traditional knowledge from being reduced to a short cultural anecdote detached from season, land, water, livelihood, governance and intergenerational learning. The IPCC recognizes that Indigenous knowledge and local knowledge can contribute important information for understanding climate risk and can enrich adaptation policy and practice. In appropriate settings, place-based long-term observations can complement formal meteorological and ocean services by showing micro-scale changes and practical consequences that national data alone may not capture. See IPCC | Climate Change 2022, Chapter 1 and IPCC | Climate Resilient Development Pathways.
Layer 2: Govern public data and traditional knowledge in tiers; AI checks permission before retrieval
If the cards are connected to RAG or a chatbot, the first design question should be who may use which information for what purpose. Public weather, roads, rivers, agriculture and government notices can remain in a public-service layer. Community observations can be classified according to community decisions as public, community-only, role-restricted or awaiting confirmation. Exact gathering locations, vulnerable-species sites, ceremonial places, family-specific knowledge, restricted marine areas and detailed traditional-territory information should not become public simply because an AI service is being built. A public web page, permission to retrieve within a controlled system and permission to train a model should also be treated as different permissions rather than one blanket consent.
The Global Indigenous Data Alliance | CARE Principles emphasize Collective Benefit, Authority to Control, Responsibility and Ethics alongside the technical value of data sharing. Local Contexts provides TK and BC Labels and related tools through which communities can communicate provenance, cultural authority, conditions of use and access expectations in digital systems. In implementation, these principles can become executable metadata such as `public / community / restricted / pending`, `retrieval allowed / retrieval denied`, `summary generation allowed / generation denied`, or `research allowed / additional consent required`. A RAG system should evaluate the user’s role and intended use first, determine the eligible data tier second, and retrieve content only after those checks pass.
Layer 3: Make local observation to community solution to policy interface a traceable workflow
The LCIPP Baku Workplan for 2025–2027 brings together exchanges among knowledge holders, regional participation, intergenerational engagement, cooperation with UNFCCC processes and stronger participation in climate policy. This suggests that township digital services do not need to place traditional knowledge and public-sector information in two disconnected worlds. They can be linked through a governed interface while preserving the distinct authority, provenance and appropriate uses of each source. See UNFCCC | Baku Workplan Activities.
In an Indigenous township in Taiwan, each resilience action card can therefore include a policy or service interface. Road and water observations can connect to disaster management, public works, fire services and township offices. Crop and seasonal observations can connect to agricultural research and extension, agricultural weather services, farmers’ organizations and technical support. Marine and fishing observations can connect to ocean conditions, fisheries services and local safety information. When Indigenous languages and cultural transmission are involved, the card can preserve approved language terms, the knowledge holder or community authority, permitted educational uses and conditions for revision or withdrawal. AIGC can help convert verified material into an accessible community version, a staff briefing or educational content, but it should not decide whether community knowledge is public and should never substitute for an official warning authority.
Layer 4: Start with a 90-day pilot so data governance matures first
During the first 30 days, one township can choose a single problem and build only 10 to 20 cards, completing provenance, permission, review and withdrawal fields before introducing advanced tools. During the next 30 days, approved public information and consented community observations can be connected to a simple GIS or table-based interface, with tests for dates, source visibility, location precision, expiry and correction. During the final 30 days, RAG or a chatbot can be added and tested with real questions from township staff, young people, farmers, fishers, knowledge holders and community organizations. When information is sensitive, sources conflict, records are stale or permissions are unclear, the system should stop and mark the item for review rather than fill the gap with a generated answer.
Success should not be measured mainly by how many questions an AI system answers. More useful indicators include the share of records reviewed on time, the time needed to remove stale information, completeness of community permissions, whether withdrawal actually removes data from retrieval and derived outputs, whether residents reach the correct service more quickly, and whether local observations return to real work in agriculture, forestry, fisheries, disaster preparedness, education or culture. This aligns with the August 19 LCIPP emphasis on community-led solutions and the conditions needed for implementation: technology is a support tool, while resilience depends on a community’s continuing ability to observe change, make collective decisions, adjust practices and transmit knowledge safely across generations.
For Yuan Media AI, the 55-township announcement index, Chatbot MVP and the Two-Eyed Seeing Lab, the most useful takeaway is not another label such as climate AI. It is a workable governance sequence: observations have sources, knowledge has permissions, actions have responsible service windows, and outcomes have review cycles before AI is asked to assist. This makes international Indigenous climate experience useful as a reference while preserving the cultural context, authority and decision-making of Indigenous communities in Taiwan.
Sources
- UNFCCC | LCIPP Regional Gathering for the Pacific
- UNFCCC | Baku Workplan Activities
- IPCC | Climate Change 2022: Impacts, Adaptation and Vulnerability, Chapter 1
- IPCC | Climate Resilient Development Pathways, Chapter 18
- Global Indigenous Data Alliance | CARE Principles
- Local Contexts | Grounding Indigenous Rights
AI use and content-safety disclosure
This article is based on public information from the UNFCCC Local Communities and Indigenous Peoples Platform, the Baku Workplan, IPCC, the Global Indigenous Data Alliance CARE Principles and Local Contexts. It translates international examples into public-service and data-governance ideas for Taiwan’s 55 Indigenous townships; it does not treat Indigenous knowledge as generic data that may be copied, disclosed or used by AI without community authority.