RESPECT - An NGO

In my view, the first Global Dialogue on AI Governance will be successful if it delivers three concrete outcomes that directly address the structural gaps preventing AI from benefiting people at the local level, especially in LMICs. First, the Dialogue should formally acknowledge the extensive evidence base documenting gaps in the three inseparable foundations of Local AI: solid and updated local datasets (fuel), community‑level digital literacy (ignition), and reliable, affordable connectivity (oxygen). These gaps are well‑established across global and national studies, yet they remain largely absent from high‑level AI governance discussions. Recognizing these gaps is essential for ensuring that AI governance frameworks are grounded in the lived realities of communities rather than only in global principles. Second, the Dialogue should commit to developing a coherent Local AI Design Framework that guides Member States in strengthening these foundations in an integrated manner. Such a framework should include standards for local data ecosystems, guidance on AI literacy for frontline workers and communities, and principles for ensuring last‑mile connectivity and offline‑capable AI tools. This would help ensure that AI systems are context‑aware, inclusive, and capable of delivering equitable benefits. Third, the Dialogue should establish mechanisms for interoperability, transparency, and human oversight that are practical for low‑capacity environments. This includes interoperable governance approaches, community‑level feedback loops, and safeguards that protect human rights in contexts where digital literacy and institutional capacity may be limited. If the Dialogue can achieve these outcomes—recognition of existing gaps, commitment to a Local AI Design Framework, and practical mechanisms for safe and inclusive implementation—it will set a strong foundation for global cooperation and ensure that AI becomes a tool that strengthens resilience, governance, and development outcomes for all communities.

• AI capacity-building
• Safe, secure and trustworthy AI
• Transparency, accountability, and human oversight
• Social, economic, ethical, cultural, linguistic and technical implications of AI

My selections reflect the urgent need to ensure that artificial intelligence delivers meaningful and equitable benefits at the local level, particularly in low- and middle-income countries. AI capacity-building is my top priority because numerous studies have already documented significant gaps in three inseparable foundations of Local AI: solid and updated local datasets, community-level digital literacy, and reliable connectivity. Without strengthening these foundations, AI systems cannot produce contextually relevant outputs or support local governance, disaster risk reduction, agriculture, health, or public services. I selected Safe, secure and trustworthy AI because LMICs require governance approaches that are practical for low-capacity environments. AI systems must be designed and deployed in ways that prevent misinterpretation of local contexts, avoid reinforcing inequalities, and ensure that communities can rely on AI tools during critical decision-making processes. Trustworthiness must be grounded in real-world conditions, not only in high-level principles. I selected Transparency, accountability, and human oversight because communities must be able to understand, question, and challenge AI outputs. This is especially important where digital literacy is limited and where AI may influence decisions affecting rights, access to services, or disaster preparedness. Human oversight must be accessible, participatory, and available in local languages to ensure that AI strengthens-not weakens-public trust. Finally, I selected Social, economic, ethical, cultural, linguistic and technical implications of AI because AI governance must reflect the diversity of local realities. This includes indigenous knowledge, linguistic diversity, cultural practices, and community-based decision-making. AI systems that ignore these dimensions risk producing outputs that are irrelevant or harmful. Together, these four priorities support a coherent approach to Local AI design, ensuring that AI governance frameworks are inclusive, context-aware, and capable of delivering equitable benefits to all communities.

Yes. In my view, a major cross-cutting issue not fully captured by the listed themes is the urgent need to develop a Local AI Architecture that addresses the foundational gaps preventing AI from benefiting people at the community level, particularly in low- and middle-income countries. While the themes address capacity-building, rights, safety, and governance, they do not explicitly recognize the structural interdependence of three essential components: solid and updated local datasets, community-level digital literacy, and reliable last-mile connectivity. These three elements are inseparable, and without all of them, AI systems cannot produce contextually relevant outputs or support local governance, disaster preparedness, agriculture, health, or public service delivery. This omission is significant because a wide range of studies has already documented these gaps across LMICs. Yet global AI governance discussions often focus on high-level principles without addressing the practical realities that determine whether AI can function meaningfully at the local level. The absence of a dedicated theme on Local AI Readiness risks reinforcing existing inequalities and limiting the ability of AI to strengthen resilience and development outcomes. A second emerging issue is the need to integrate indigenous knowledge, local languages, and cultural contexts into AI systems. These dimensions are not fully reflected in the current themes, yet they are essential for ensuring that AI tools are trusted, relevant, and aligned with community practices. Without this integration, AI risks producing outputs that are technically correct but socially or culturally misaligned. Finally, there is a need for offline-capable and edge-based AI solutions for communities with limited connectivity. This is not explicitly addressed in the listed themes but is critical for equitable access. Recognizing these cross-cutting issues would strengthen the Dialogue's ability to support inclusive, context-aware, and locally grounded AI governance.

The governance gaps in the selected thematic areas—AI capacity-building; safe, secure and trustworthy AI; transparency, accountability and human oversight; and the social, economic, ethical, cultural, linguistic and technical implications of AI—are already having significant impacts across LMICs, including in the regions and sectors where my work is focused. The most critical challenge is that AI capacity-building remains structurally weak at the local level. Many LMICs lack solid and updated local datasets, community-level digital literacy, and reliable last‑mile connectivity. These three components are inseparable, and their absence prevents AI systems from producing contextually relevant outputs for agriculture, disaster risk reduction, health, education, and local governance. As a result, AI tools often reflect global data patterns rather than local realities, limiting their usefulness and sometimes generating misleading or inaccurate recommendations. A second challenge is the limited applicability of global AI safety and trust frameworks in low‑capacity environments. Many LMIC institutions lack the technical, financial, or regulatory capacity to implement complex governance mechanisms. This creates a widening gap between global norms and local feasibility, increasing the risk of misuse, exclusion, or over‑reliance on unverified AI outputs. Transparency and human oversight gaps also affect LMICs disproportionately. In contexts with low digital literacy, communities may not understand how AI systems work, how to question outputs, or how to seek redress. This undermines trust and can reinforce existing inequalities. At the same time, there are significant opportunities. AI can strengthen early warning systems, improve local service delivery, support indigenous knowledge, and enhance community resilience—but only if governance frameworks prioritize local relevance, local data ecosystems, and community participation. Addressing these gaps through a coherent Local AI Design Framework would enable AI to become a practical tool for development, resilience, and inclusive governance across LMICs.

The AI Dialogue can play a pivotal role in advancing international cooperation by creating a shared, practical, and inclusive framework for AI governance that reflects the realities of low- and middle-income countries (LMICs). The Dialogue offers a unique opportunity to move beyond high‑level principles and establish a common global agenda that addresses the foundational gaps preventing AI from benefiting communities at the local level. First, the Dialogue can harmonize global understanding of Local AI readiness by recognizing that AI cannot function meaningfully without three inseparable components: solid and updated local datasets, community‑level digital literacy, and reliable last‑mile connectivity. These gaps are widely documented across LMICs, yet they remain underrepresented in global governance discussions. By elevating these issues, the Dialogue can help ensure that international cooperation focuses on strengthening the "fuel–ignition–oxygen" foundations required for equitable AI deployment. Second, the Dialogue can facilitate interoperable governance approaches that LMICs can adopt without excessive technical or financial burden. Many countries lack the capacity to implement complex regulatory frameworks. International cooperation can support the development of adaptable, resource‑sensitive models that promote safety, transparency, and accountability while remaining feasible for low‑capacity environments. Third, the Dialogue can promote shared standards for local data ecosystems, including open data, open AI models, and culturally relevant datasets. This would enable countries to collaborate on building local language corpora, indigenous knowledge repositories, and community‑level datasets essential for context‑aware AI. Fourth, the Dialogue can strengthen South–South and triangular cooperation, enabling LMICs to exchange practical solutions, local innovations, and community‑driven approaches to AI governance. Finally, the Dialogue can serve as a platform for inclusive participation, ensuring that local governments, civil society, indigenous communities, and frontline workers contribute to shaping global AI norms. Through these roles, the AI Dialogue can help build a globally coherent yet locally grounded governance ecosystem that ensures AI benefits all communities.

Several existing initiatives, partnerships, and mechanisms provide strong foundations that the AI Dialogue can build upon, but they remain fragmented and insufficiently connected to the realities of low- and middle-income countries (LMICs). The Dialogue can add significant value by integrating these efforts into a coherent, globally coordinated approach that prioritizes Local AI readiness. At the global level, initiatives such as the UN Secretary-General's High-Level Advisory Body on AI, UNESCO's Recommendation on the Ethics of AI, the Global Digital Compact, and the OECD AI Principles offer important normative guidance. Similarly, sectoral initiatives—such as WHO's work on AI in health, FAO's digital agriculture frameworks, and UNDRR's efforts on risk data and early warning systems—provide valuable technical foundations. Regional bodies, including the African Union, CARICOM, and ASEAN, are also advancing AI strategies that reflect local development priorities. However, these initiatives do not yet converge around the three inseparable components required for AI to function meaningfully at the local level: solid and updated local datasets, community-level digital literacy, and reliable last‑mile connectivity. This is where the AI Dialogue can bring unique added value. First, the Dialogue can connect global normative frameworks with local implementation realities, ensuring that principles on safety, rights, and accountability are feasible for low‑capacity environments. Second, it can promote shared standards for local data ecosystems, including open data, open AI models, and culturally relevant datasets—areas where current initiatives remain siloed. Third, the Dialogue can strengthen South–South and triangular cooperation, enabling LMICs to exchange practical solutions, indigenous knowledge approaches, and community-driven innovations. Finally, the Dialogue can serve as a coordination platform, aligning global, regional, and sectoral initiatives into a coherent Local AI Design Framework that ensures AI benefits reach communities, not only capitals.

The AI Dialogue can be most effective when it brings together governments, international organizations, civil society, academia, the private sector, and local communities in a structured, inclusive, and action‑oriented process. Each stakeholder group has a distinct role to play in shaping a governance ecosystem that is globally coherent yet grounded in local realities, particularly in low- and middle-income countries (LMICs). Governments can contribute by sharing national experiences, identifying capacity gaps, and aligning domestic policies with emerging global norms. International organizations can provide technical guidance, normative frameworks, and platforms for cross‑regional cooperation. Civil society and community-based organizations can ensure that governance discussions reflect the lived realities of marginalized groups, indigenous communities, and frontline workers. Academia and research institutions can contribute evidence, methodologies, and impact assessments. The private sector can support innovation, transparency, and responsible deployment practices. Local governments and community leaders—often overlooked—can provide insights into the practical challenges of implementing AI in low‑capacity environments. To enable meaningful participation, the AI Dialogue should adopt a multi-layered format: Global plenary sessions to establish shared principles and high-level priorities. Regional and subregional consultations to capture diverse contexts, especially from LMICs. Thematic working groups focused on key issues such as local data ecosystems, AI literacy, connectivity, safety, and human oversight. Community-level listening sessions to ensure that local voices—including indigenous knowledge holders—inform global governance. Technical roundtables involving researchers, practitioners, and industry to translate principles into implementable guidance. South–South and triangular cooperation forums to exchange practical solutions and local innovations. The Dialogue should prioritize inclusive participation, evidence-based decision-making, and practical outputs, such as a Local AI Design Framework. This structure will ensure that global cooperation leads to governance approaches that are feasible, context-aware, and capable of delivering equitable benefits to all communities.

Several critical voices, communities, and perspectives remain underrepresented in global discussions on AI governance, particularly those from low- and middle-income countries (LMICs) and from communities at the local or grassroots level. These groups are often the most affected by AI-driven transformations yet have the least influence over how AI systems are designed, governed, and deployed. Local governments, frontline workers, and community leaders—including village councils, municipal authorities, health workers, teachers, and disaster response volunteers—rarely participate in global AI governance processes. Their exclusion means that governance frameworks often overlook practical realities such as limited connectivity, weak local datasets, and low digital literacy. Indigenous communities and holders of traditional knowledge are also underrepresented. Their knowledge systems, languages, and cultural contexts are seldom reflected in AI datasets or model design, leading to outputs that may be technically correct but socially or culturally misaligned. Women, youth, and marginalized groups in LMICs face structural barriers to participation, including limited access to digital tools, training, and decision-making spaces. Their perspectives are essential for ensuring that AI systems do not reinforce existing inequalities. Small civil society organizations and community-based NGOs—particularly those working on digital inclusion, disaster risk reduction, agriculture, and local governance—are often absent from global dialogues despite their deep understanding of community needs. To include these voices, the AI Dialogue should adopt deliberate inclusion mechanisms, such as: Community-level listening sessions in LMICs Regional consultations that prioritize local governments and indigenous groups Dedicated participation pathways for grassroots organizations Support for local-language submissions Funding for travel, connectivity, and digital access to enable meaningful participation South–South and triangular cooperation platforms to amplify LMIC-led innovations Including these perspectives will ensure that AI governance frameworks are grounded in real-world conditions and capable of delivering equitable benefits to all communities.

To foster meaningful and dynamic engagement, the AI Dialogue should adopt innovative formats that bring diverse voices into the process—especially those from low- and middle-income countries (LMICs), local governments, indigenous communities, and frontline practitioners. Traditional consultation formats often privilege well‑resourced actors; therefore, new engagement models are needed to ensure that governance discussions reflect real-world conditions and community-level priorities. 1. Community-Level Listening Sessions Organizing structured listening sessions in rural and urban communities across LMICs would allow the Dialogue to capture perspectives from frontline workers, local leaders, women's groups, youth networks, and indigenous knowledge holders. These sessions should be conducted in local languages and facilitated by regional partners. 2. Regional and Subregional Dialogues Decentralized regional consultations—hosted by the African Union, ASEAN, CARICOM, SAARC, and others—would ensure that regional priorities and cultural contexts shape global governance norms. These dialogues can surface practical challenges related to local datasets, digital literacy, and connectivity. 3. Multi-Stakeholder Working Labs Instead of traditional panels, the Dialogue could convene "working labs" where governments, researchers, civil society, and private-sector actors co‑design solutions on topics such as local data ecosystems, AI literacy, safety, and human oversight. These labs should produce actionable outputs, not only discussions. 4. South–South and Triangular Cooperation Forums Dedicated forums for LMIC-to-LMIC exchange would enable countries to share practical innovations, indigenous knowledge approaches, and community-driven AI solutions that are often overlooked in global debates. 5. Offline and Low-Bandwidth Participation Channels To ensure inclusion of communities with limited connectivity, the Dialogue should allow submissions via SMS, WhatsApp, community radio, and offline-first digital tools. 6. Youth and Indigenous Knowledge Assemblies Dedicated assemblies would ensure that underrepresented groups contribute directly to shaping global AI norms. By adopting these formats, the AI Dialogue can become a genuinely inclusive, dynamic, and locally grounded process that strengthens global cooperation while ensuring that AI governance reflects the needs of all communities.

Several existing policies, practices, and platforms offer valuable lessons for effective AI governance and provide concrete solutions to emerging challenges. However, most remain fragmented and insufficiently adapted to the realities of low- and middle-income countries (LMICs). The AI Dialogue can build on these initiatives while addressing their gaps. 1. UNESCO's Recommendation on the Ethics of AI This is one of the most comprehensive global normative frameworks, emphasizing human rights, transparency, and accountability. Its value lies in its universal applicability, but it requires adaptation for low-capacity environments where digital literacy and local datasets are limited. 2. OECD AI Principles and the Global Partnership on AI (GPAI) These initiatives promote responsible AI, risk management, and international cooperation. Their strength is in establishing shared governance norms, but LMIC participation remains limited, and the frameworks often assume high institutional capacity. 3. Open-Source and Open Data Ecosystems Platforms such as OpenStreetMap, Humanitarian Data Exchange (HDX), and FAIR data principles demonstrate how open data can strengthen local decision-making. These models are particularly relevant for building the "fuel" of Local AI-solid, updated local datasets. 4. Sectoral Governance Models WHO's guidance on AI in health provides practical safeguards for clinical and public health settings. FAO's digital agriculture frameworks support responsible use of AI in farming and food systems. UNDRR's risk data and early warning initiatives show how AI can support disaster preparedness when local data ecosystems are strong. 5. National AI Strategies in LMICs Countries such as Rwanda, India, and Kenya have developed AI strategies emphasizing inclusion, local innovation, and public-sector applications. These offer scalable models for other LMICs. The AI Dialogue can add value by integrating these initiatives into a coherent Local AI Design Framework, ensuring that global norms translate into practical, context-aware solutions that strengthen local data ecosystems, digital literacy, and connectivity.