UNESCO Rwanda National Commission

A successful first Global Dialogue on AI Governance would move beyond broad principles and deliver concrete, inclusive, and actionable outcomes. From my experience in AI training and engagement with UNESCO-related initiatives, three outcomes stand out. First, practical policy guidance tailored to different regions. Many countries, especially in the Global South, struggle to translate high-level frameworks like UNESCO's AI Ethics Recommendation into implementable policies. During AI trainings I've facilitated, participants often ask not what ethical AI is, but how to operationalize it within limited infrastructure and capacity. A successful dialogue would therefore produce adaptable toolkits, case studies, and policy templates that reflect diverse socio-economic realities. Second, capacity-building commitments. In my work delivering AI training programs, I've seen that governance gaps are often skills gaps. Policymakers, educators, and developers need shared literacy in AI systems, risks, and accountability mechanisms. The dialogue should result in funded partnerships, linking governments, academia, and organizations like UNESCO to scale training, particularly in underrepresented regions. Third, meaningful inclusion and representation. Too often, global AI conversations are dominated by a small group of countries and companies. In UNESCO-aligned forums I've participated in, the most impactful moments came when local voices, youth, civil society, and practitioners from emerging economies were given space to shape the agenda. Success would mean institutionalizing these voices in decision-making, not just consultation. Ultimately, success is not a declaration, but a shift: from principles to practice, from inclusion as a slogan to inclusion as structure, and from fragmented efforts to sustained global cooperation.

• AI capacity-building
• Social, economic, ethical, cultural, linguistic and technical implications of AI
• Open-source software, open data and open AI models
• Interoperability of governance approaches

AI capacity-building is my top priority because, in my experience delivering AI training, the biggest barrier to effective governance is not lack of principles but lack of skills. Policymakers, educators, and practitioners often need practical understanding of AI systems, risks, and oversight tools. Scaling accessible, context-aware training especially in underrepresented regions would directly strengthen governance outcomes. The broader implications of AI social, economic, ethical, cultural, linguistic, and technical are equally urgent. Through engagement with UNESCO aligned initiatives, I've seen how global frameworks must be grounded in local realities. Issues like language inclusion, cultural bias in datasets, and unequal economic impacts require interdisciplinary and region-specific responses, not one-size-fits-all solutions. Open-source software, open data, and open AI models are critical for equity and participation. In training environments, open tools have enabled wider experimentation, transparency, and local innovation, particularly where resources are limited. Supporting open ecosystems can democratize AI development while also enabling scrutiny and accountability. Finally, interoperability of governance approaches is essential to avoid fragmentation. Many countries are developing AI policies in parallel, often with limited coordination. From my perspective, aligning these approaches while respecting national contexts would help ensure consistency, facilitate cross-border collaboration, and reduce regulatory gaps. Practical bridges between frameworks, such as shared standards and mutual learning platforms, would be a key outcome. Together, these priorities reflect a shift from abstract discussions to actionable, inclusive, and globally coherent AI governance.

First, AI and environmental sustainability is often underrepresented. The growing energy and water demands of AI systems raise concerns about their environmental footprint. From a UNESCO perspective that links technology with sustainable development, governance frameworks should integrate environmental impact assessments and promote "green AI" practices, especially for countries already vulnerable to climate change. Second, data governance and data sovereignty require stronger emphasis. In many AI training contexts I've engaged in, participants highlight challenges around who owns, controls, and benefits from data. This is particularly critical for the Global South, where data extraction without equitable returns can reinforce existing inequalities. Aligning with UNESCO's principles, there is a need for fair data governance models that protect rights while enabling innovation. Third, AI in education systems is an emerging priority. UNESCO has been at the forefront of exploring AI's role in education, yet governance discussions often overlook how AI reshapes learning, assessment, and knowledge production. From my experience in AI training, there is a growing need for policies that guide responsible AI use in classrooms while preparing learners with critical AI literacy. Finally, trust, public awareness, and societal readiness cut across all themes. Beyond technical and policy solutions, successful AI governance depends on public understanding and trust. UNESCO's emphasis on human-centered AI highlights the importance of engaging citizens, not just experts, in shaping AI futures. Addressing these cross-cutting issues would strengthen the dialogue by ensuring AI governance is not only technically sound, but also socially responsible, sustainable, and inclusive.

In my context—working in AI training and engaging with initiatives aligned with UNESCO—governance gaps across the selected thematic areas are already shaping both challenges and opportunities in the region. A key challenge is limited institutional capacity. While interest in AI is growing rapidly, many policymakers and institutions lack the technical expertise to design, implement, and enforce effective AI regulations. This often leads to either delayed policy action or the adoption of frameworks that are not fully adapted to local realities. In my AI training work, this gap is clear: participants are eager to engage, but need practical tools and sustained learning support. Another major issue is data inequality and linguistic exclusion. Many AI systems underperform in local languages and contexts, reflecting broader gaps in open data and inclusive datasets. This reinforces digital inequality and limits the relevance of AI solutions in sectors like education and public services—areas that UNESCO prioritizes. At the same time, there are important opportunities. The growing availability of open-source AI tools is lowering barriers to entry, enabling local innovation ecosystems to emerge. In training environments, I've seen how access to open models allows young developers and researchers to experiment and build context-specific solutions. There is also momentum around regional and global alignment. Efforts to harmonize governance approaches—drawing on frameworks like UNESCO's AI Ethics Recommendation—offer a pathway to develop policies that are both locally grounded and internationally consistent. Overall, the governance gaps highlight an urgent need for capacity-building, inclusive data ecosystems, and stronger collaboration. If addressed strategically, these same gaps can become entry points for more equitable, human-centered AI development in the region.

The AI Dialogue can play a catalytic role in advancing international cooperation by turning shared principles into coordinated action, in line with the human-centered approach championed by UNESCO. First, it can serve as a bridge between global frameworks and national implementation. While instruments like UNESCO's AI Ethics Recommendation provide a strong normative foundation, many countries need support to operationalize them. The Dialogue can facilitate peer learning, policy exchange, and the co-creation of practical tools that reflect diverse regional contexts. Second, it can strengthen inclusive multistakeholder collaboration. Effective AI governance requires the involvement of governments, academia, private sector actors, and civil society. The Dialogue can institutionalize participation from underrepresented regions—particularly the Global South—ensuring that international cooperation is not only broad but also equitable. Third, it can promote interoperability and policy coherence. As countries develop their own AI strategies, there is a risk of fragmentation. The Dialogue can align standards, encourage mutual recognition of governance approaches, and foster common benchmarks for accountability, transparency, and safety, while respecting national sovereignty. Fourth, it can mobilize capacity-building partnerships. Drawing on UNESCO's experience in education and knowledge-sharing, the Dialogue can connect stakeholders to scale AI literacy, technical training, and institutional capacity—especially where governance gaps are most pronounced. Finally, it can act as a platform for trust-building and foresight. By enabling open exchange on emerging risks and innovations, the Dialogue can help anticipate challenges and coordinate responses proactively. In essence, the AI Dialogue can transform international cooperation from fragmented efforts into a sustained, inclusive, and action-oriented ecosystem for responsible AI governance.

The AI Dialogue should build on existing global and regional initiatives that already advance ethical, inclusive, and human-centered AI, particularly those aligned with UNESCO. A key foundation is UNESCO's AI Ethics Recommendation and its Readiness Assessment Methodology, which provide countries with tools to evaluate gaps and guide implementation. The Dialogue can amplify these by creating a space for countries to share lessons learned, compare outcomes, and co-develop practical implementation pathways. It should also connect with the work of OECD on AI principles and policy observatories, as well as the Global Partnership on Artificial Intelligence, which brings together governments and experts to advance responsible AI. These platforms already generate valuable research and policy guidance, but coordination across them remains limited. At the multilateral level, initiatives under the United Nations system—including digital cooperation efforts and capacity-building programs—offer additional entry points for alignment, especially in supporting developing countries. From my experience in AI training and UNESCO-related work, there is also significant value in regional and grassroots initiatives—such as local AI communities, academic networks, and civil society-led training programs—which are often overlooked but critical for implementation. The added value of the AI Dialogue lies in connecting these fragmented efforts into a coherent ecosystem. It can act as a convening and coordination platform that reduces duplication, aligns priorities, and ensures that global frameworks translate into local impact. It can also elevate underrepresented voices, link policy with capacity-building, and foster sustained collaboration rather than one-off engagements. In this way, the Dialogue would not replace existing initiatives, but strengthen and unify them into a more inclusive and action-oriented global AI governance landscape.

Different stakeholders can contribute to the AI Dialogue by leveraging their unique roles within a shared, human-centered framework aligned with UNESCO. Governments should lead in shaping policy commitments, sharing national experiences, and identifying regulatory needs, particularly from underrepresented regions. Academia and research institutions can provide evidence-based insights, foresight on emerging risks, and evaluation of policy impacts. Private sector actors should contribute technical expertise, transparency practices, and responsible innovation models. Civil society and local communities play a critical role in grounding discussions in lived realities—especially on issues of inclusion, human rights, and social impact. From my experience in AI training, educators and practitioners are essential bridges, translating governance principles into skills and real-world applications. To be effective, the AI Dialogue should adopt a hybrid, multi-layered structure: Thematic working groups aligned with priority areas (e.g., capacity-building, data governance, interoperability), tasked with producing concrete outputs such as toolkits or policy guidance. Regional tracks to ensure context-specific discussions and elevate Global South perspectives. Plenary forums for high-level exchange and political commitment, connected to technical discussions rather than isolated from them. Capacity-building labs or clinics, where participants engage in hands-on learning, reflecting UNESCO's emphasis on education and knowledge-sharing. Continuous engagement mechanisms, such as online platforms or communities of practice, to sustain collaboration beyond annual meetings. Importantly, the Dialogue should prioritize co-creation over consultation, ensuring stakeholders are involved in decision-making, not just feedback. Clear pathways from discussion to implementation—through partnerships, pilot projects, and follow-up mechanisms—will be key. Such a structure would make the AI Dialogue inclusive, action-oriented, and capable of translating global principles into meaningful local impact.

Global discussions on AI governance still overlook several critical voices, which risks reinforcing existing inequalities—an issue central to the work of UNESCO. Most notably, communities from the Global South remain underrepresented, despite being deeply affected by AI deployment. In my experience with AI training, participants from these regions often highlight that policies are shaped elsewhere and later applied to their contexts without adaptation. Inclusion requires dedicated regional tracks, funding for participation, and support for local research and policy development. Youth and students are another missing voice. As both current users and future leaders of AI systems, their perspectives on education, employment, and digital rights are essential. UNESCO's emphasis on future-ready education suggests integrating youth-led forums and co-creation spaces within governance processes. Linguistic and cultural minorities are also often excluded. AI systems frequently fail to reflect diverse languages and knowledge systems, limiting accessibility and fairness. Inclusion here means supporting multilingual datasets, culturally grounded AI design, and participation from indigenous and local knowledge holders. Civil society and grassroots practitioners, especially those working at the intersection of AI and social impact, are frequently consulted but not meaningfully involved in decision-making. From my experience, these actors bring practical insights on how AI affects communities in areas like education and public services. They should be integrated as equal partners in working groups and policy design processes. Finally, educators and capacity-builders are often overlooked, despite their role in translating AI governance into skills and practice. Strengthening their involvement can bridge the gap between policy and implementation. To include these perspectives, the AI Dialogue must move beyond symbolic representation toward structured participation, funding mechanisms, and co-decision models—ensuring governance is not only global in scope, but genuinely inclusive in practice.

To foster meaningful and dynamic engagement, the AI Dialogue should move beyond traditional panels and adopt participatory, practice-oriented formats aligned with the human-centered approach of UNESCO. One effective format is policy co-creation labs, where diverse stakeholders—governments, technologists, civil society, and educators—work together on real governance challenges. Rather than discussing principles abstractly, participants jointly develop draft policies, toolkits, or implementation roadmaps, reflecting UNESCO's emphasis on actionable outcomes. Capacity-building clinics are another valuable approach. Drawing from AI training experience, these hands-on sessions allow participants—especially from underrepresented regions—to engage directly with tools for risk assessment, ethical impact analysis, or data governance. This bridges the gap between knowledge and application. Scenario-based simulations can deepen engagement by placing participants in realistic governance dilemmas (e.g., regulating AI in education or public services). These exercises encourage negotiation, foresight, and interdisciplinary thinking, helping stakeholders understand trade-offs and align on shared responses. To ensure inclusivity, the Dialogue could integrate regional innovation hubs or parallel local dialogues connected to the global event. This allows voices that cannot travel to still shape outcomes, aligning with UNESCO's focus on equitable participation. Additionally, multilingual digital platforms can support continuous engagement before, during, and after the Dialogue. Features such as collaborative drafting spaces, open consultations, and community forums would enable sustained, global participation. Finally, youth-led sessions and community showcases can highlight grassroots innovation and lived experiences, shifting from top-down discussions to co-created narratives. These formats emphasize interaction, co-creation, and continuity—ensuring the AI Dialogue is not just a forum for exchange, but a space for building practical, inclusive, and lasting governance solutions.

everal existing policies and practices already offer concrete pathways for effective, human-centered AI governance, closely aligned with the work of UNESCO. A leading example is UNESCO's AI Ethics Recommendation, which provides a comprehensive normative framework grounded in human rights, inclusion, and sustainability. Its accompanying Readiness Assessment Methodology (RAM) is particularly practical, enabling countries to identify governance gaps and prioritize context-specific actions-an approach I've seen resonate strongly in AI training settings. At the regional level, the European Union's AI Act offers a risk-based regulatory model that classifies AI systems by their potential harm and applies proportionate obligations. This approach provides a useful template for balancing innovation with accountability, especially when adapted to local contexts. From a technical and operational perspective, algorithmic impact assessments (AIAs) are gaining traction as a governance tool. These frameworks-used in countries like Canada-require organizations to evaluate the social and ethical risks of AI systems before deployment, promoting transparency and accountability. Open and collaborative platforms also play a key role. Initiatives like Global Partnership on Artificial Intelligence foster knowledge-sharing, research, and policy experimentation across countries and sectors. Similarly, open-source ecosystems enable scrutiny, localization, and capacity-building, particularly in resource-constrained environments. In practice, capacity-building programs-including AI literacy training for policymakers, educators, and developers-are among the most impactful approaches I've encountered. They translate governance principles into actionable skills and empower stakeholders to engage meaningfully in policy and implementation. Together, these examples highlight that effective AI governance is not a single instrument, but a combination of normative frameworks, regulatory tools, technical practices, and continuous learning mechanisms-all of which must be adapted to diverse cultural and socio-economic contexts.