Public Open University Cakovec

A successful first Global Dialogue on AI Governance should produce outcomes that are practical, inclusive, and credible enough to guide future cooperation. First, success would mean establishing a shared understanding that AI governance is not only about regulating advanced technologies, but also about protecting human dignity, supporting inclusion, and expanding human development. The Dialogue should show that AI governance must include the perspectives of governments, researchers, industry, civil society, educators, and users from different regions of the world. Second, the Dialogue should identify a limited number of clear common priorities for international cooperation. These could include transparency, accountability, privacy, child and learner protection, equitable access, and capacity-building for countries and communities with fewer resources. Even if full consensus is not possible, agreement on a basic set of priority areas would be an important step forward. Third, the Dialogue should create a strong foundation for future work by producing concise and accessible outputs: a summary of key recommendations, a roadmap for continued stakeholder engagement, and a commitment to include underrepresented voices in the next stages of the process. In my view, one particularly important sign of success would be recognition that AI on personal devices can widen access to learning and knowledge across continents. Learners of different ages and backgrounds can use AI tools to ask questions, develop skills, and strengthen curiosity and creativity. At the same time, this opportunity must be accompanied by critical thinking, fact-checking, privacy protection, and balanced use. Ultimately, the first Dialogue will be successful if it does not remain only a symbolic conversation, but becomes a trusted starting point for inclusive, practical, and human-centered global cooperation on AI governance.

• Safe, secure and trustworthy AI
• AI capacity-building
• Transparency, accountability, and human oversight
• Open-source software, open data and open AI models

I selected these priorities because, from my perspective as an educator and researcher, they are the most important for ensuring that AI develops in ways that are socially beneficial, globally inclusive, and usable in real learning and public contexts. Safe, secure and trustworthy AI is essential because trust is the foundation of meaningful adoption. If AI systems are unreliable, unsafe, or harmful, they will deepen public mistrust and inequality rather than support human development. AI capacity-building is a priority because access to AI is not enough on its own. People, institutions, and communities need the knowledge and skills to use AI critically, responsibly, and effectively. This is especially important in education and in regions or sectors with fewer resources. Transparency, accountability, and human oversight are central because users need to understand the limits of AI systems, question outputs, and retain human judgment in important decisions. AI should support human agency, not weaken it. In learning environments especially, transparency and oversight are necessary to protect learners and maintain educational integrity. I also selected open-source software, open data, and open AI models because openness can widen access, reduce dependency on a small number of dominant actors, and support innovation, adaptation, and learning across different contexts. For many educators, researchers, and smaller institutions, open approaches are one of the most realistic paths toward meaningful participation in the AI ecosystem. Taken together, these four priorities combine protection with opportunity. They support both responsible governance and broader democratic access to AI, including its potential to expand learning, creativity, and capacity-building across different regions and age groups.

Yes. One important cross-cutting issue is the relationship between AI governance, learning, and human capacity. Education is not just one sector affected by AI. It is also the main way societies prepare people to understand AI, question it, use it responsibly, and participate in shaping its future. Without broad AI literacy, critical thinking, and public understanding, even well-designed governance frameworks may remain weak in practice. A second issue is equitable access through personal devices. For many people across the world, smartphones and other personal devices are the main gateway to digital learning, information, and AI tools. This creates major opportunities for self-directed learning, creativity, and skills development, but also raises concerns about quality, misinformation, privacy, dependency, and unequal digital conditions. Governance discussions should pay more attention to how AI is actually experienced in everyday life, not only in institutional or industrial settings. A third emerging issue is learner well-being and balanced use. AI can support curiosity, inclusion, and personalized learning, but excessive or uncritical use may weaken reflection, attention, interpersonal interaction, or independent problem-solving. Governance should therefore consider not only access and innovation, but also healthy and developmentally appropriate use. Finally, I believe there should be stronger attention to public-interest and smaller-context participation. AI governance debates are often dominated by major states, large companies, and highly resourced institutions. More space is needed for educators, smaller countries, civil society actors, and communities that experience AI primarily as users rather than developers. These issues cut across safety, transparency, capacity-building, and openness, and they are important if AI governance is to remain human-centered, inclusive, and globally relevant.

In my sector, education, the governance gaps around AI are already having visible effects. The greatest challenge is that the use of AI is spreading faster than the development of clear guidance, institutional capacity, and shared standards. Teachers, learners, and educational institutions are often expected to make decisions about AI use without sufficient support on issues such as transparency, accountability, privacy, academic integrity, and age-appropriate use. In my country and region, this creates uneven adoption. Some educators and institutions are experimenting thoughtfully and productively, while others avoid AI entirely or use it without enough critical understanding. This risks widening inequalities between schools, teachers, and learners, especially where access to training, digital resources, or policy guidance is limited. At the same time, the opportunities are significant. AI tools on personal devices can widen access to learning, support self-directed education, and help learners of different ages and backgrounds explore ideas, strengthen skills, and build curiosity and creativity. For educators, AI can also support lesson preparation, differentiation, communication, and the creation of more adaptable learning materials. The most important governance challenge is therefore not only how to regulate AI systems, but how to ensure that people and institutions can use them safely, critically, and meaningfully. This is why capacity-building, human oversight, transparency, and trustworthy AI are so important in education. If these areas are addressed well, AI could help make learning more accessible, flexible, and inclusive. If they are not, the result may be confusion, dependency, mistrust, and deeper inequalities. For my sector, the key issue is to ensure that AI strengthens human agency, judgment, and educational opportunity rather than weakening them.

The AI Dialogue can play a valuable role by bringing people together who would not otherwise have much influence over global AI discussions. Right now, debates on AI governance are often shaped by the most powerful countries and the largest technology companies. The Dialogue can help open more space for smaller states, educators, researchers, civil society, and other stakeholders whose experiences also matter. It can also help countries learn from one another. Many governments and sectors are facing similar questions about safety, transparency, accountability, human oversight, and capacity-building, but they are approaching them from very different starting points. A UN-led dialogue can make it easier to share lessons, compare approaches, and identify what works in practice. Another important role is helping to build some common ground. Not every country will agree on the same rules or pace of regulation, but there can still be agreement on broad priorities and shared concerns. That kind of common understanding is useful because it creates a better basis for future cooperation. For me, one of the most important things is that the Dialogue should connect high-level governance debates with everyday life. AI is not only about major companies or advanced systems. It is also about how people learn, work, search for information, and use AI on their personal devices. International cooperation will be stronger if it pays attention to those real human experiences. In the end, the AI Dialogue will matter most if it becomes more than a formal meeting. Its real value will be in helping build trust, continuity, and a stronger sense that AI governance should be inclusive, practical, and shaped by the needs of people across different regions and sectors.

• The AI Dialogue should build on initiatives that already offer useful principles, experience, and policy groundwork, rather than starting from zero. A few stand out in particular: the UNESCO Recommendation on the Ethics of AI, which provides a global ethical framework
• the OECD AI Principles, which have helped shape discussion on trustworthy and human-centred AI
• and the Council of Europe Framework Convention on AI, which brings a stronger human rights, democracy, and rule of law perspective. Within the UN system itself, the Dialogue should also stay closely connected to the Independent International Scientific Panel on AI, so that political discussions are informed by credible scientific expertise. It should also connect with existing sector-based and regional efforts, including work in education, digital literacy, and public-interest technology. A lot of valuable experience is already being generated through national policies, research communities, and professional networks, but it is often fragmented and unevenly shared. One useful role for the Dialogue would be to help bring those scattered experiences into a more visible and inclusive global conversation. The added value of the AI Dialogue, in my view, is not that it replaces these initiatives. Its value is that it can connect them. The UN can provide a broader and more inclusive space where governments, researchers, civil society, educators, and other stakeholders can compare approaches, identify gaps, and learn from one another across regions and sectors. It can help translate existing principles into a more global conversation that includes countries and communities with less influence in current AI governance debates. If it works well, the Dialogue can act as a bridge: linking ethics, policy, human rights, scientific evidence, and everyday practice, while adding something that many existing mechanisms still lack — broader global legitimacy and more inclusive participation.

Different stakeholders can contribute best if the AI Dialogue is designed as a genuinely mixed space, not only as a formal exchange between governments. Governments can bring policy experience, national priorities, and lessons from regulation or public administration. Researchers and scientific experts can help clarify evidence, risks, and emerging developments. Civil society can highlight rights-based concerns, inclusion, and the lived experience of affected communities. Industry can contribute technical knowledge, implementation experience, and practical insight into how systems are built and deployed. Educators, youth, and public-interest actors can help connect AI governance to everyday realities such as learning, access, trust, and responsible use. For that reason, I would recommend a format that combines formal and practical elements. It should include plenary sessions for broad strategic discussion, but also smaller thematic working sessions where participants can speak more openly and in more detail. The structure should make room for both Member States and non-state stakeholders in a visible and meaningful way, rather than treating stakeholder input as secondary. It would also be useful to include short written inputs in advance, followed by moderated roundtables with balanced regional and sector representation. A good Dialogue should not be dominated by the loudest or best-resourced actors. It should create space for smaller countries, underrepresented regions, and sectors such as education, where AI governance is experienced in very practical ways. To make the process more useful, each Dialogue should end with a short public summary of key points, areas of convergence, open questions, and suggested next steps. In my view, the best structure is one that is inclusive, transparent, and continuous: not a one-off event, but part of an ongoing process where different stakeholders can contribute knowledge, raise concerns, and help shape practical international cooperation.

People with low incomes, limited formal education, and weaker digital access are still far too underrepresented in global discussions on AI governance, even though they are often among those most affected by technological change. In many parts of the world, AI is already shaping access to information, learning, work, and public services, but the people experiencing these changes most directly are rarely present in policy conversations. This also applies to rural communities, older adults with limited digital confidence, people with disabilities, migrants, linguistic minorities, and communities in smaller or less-resourced countries. Too often, global AI governance debates are shaped mainly by governments of powerful states, large technology companies, and highly educated experts. Those voices are important, but they do not represent the full reality of how AI is used and experienced in everyday life. If AI governance is meant to be inclusive, these groups need more than symbolic invitations. Their inclusion should be supported through practical measures: plain-language consultation materials, translation into more languages, hybrid and online participation formats, partnerships with local civil society organizations, and targeted outreach through schools, libraries, community centres, and adult education providers. It is also important to support participation financially, since many people and smaller organizations cannot contribute without travel support, compensation, or technical assistance. In addition, the Dialogue should actively include people not only as experts, but also as users of AI in everyday contexts. Listening to those with fewer resources, lower levels of formal education, or less institutional power would make the discussion more grounded, fair, and globally relevant. A credible global dialogue on AI governance should not only hear from those building AI or regulating it, but also from those living with its consequences every day.

Meaningful engagement will be stronger if the AI Dialogue moves beyond formal statements and includes formats that allow people to react, compare experiences, and solve problems together. One useful format would be small moderated roundtables with a balanced mix of governments, researchers, civil society, educators, industry, and underrepresented communities. These are often more productive than large plenary sessions because participants can speak more openly and respond directly to one another. A second format could be problem-focused labs built around real cases. For example, participants could discuss concrete scenarios linked to education, public services, disinformation, language inclusion, or access through personal devices. This would help move the discussion from abstract principles to practical governance challenges. It would also be valuable to include short stakeholder spotlight sessions, where participants from less-heard backgrounds have a few minutes to present lived experiences or practical concerns. This could help bring in the perspectives of poorer communities, people with limited formal education, smaller countries, and organizations with fewer resources. Another strong option is hybrid digital participation that allows people to contribute through short written inputs, video statements, polls, and moderated online discussion before and during the Dialogue. This would widen participation for those who cannot travel and make the process less dependent on physical presence and institutional power. Finally, the Dialogue could use facilitated synthesis sessions at the end of each thematic track. Instead of ending with disconnected comments, moderators could identify key points of agreement, open tensions, and practical next steps. That would make participation feel more meaningful and help turn discussion into usable outcomes. In my view, the most effective formats are those that are interactive, accessible, and grounded in real human experience, not only in formal diplomacy.

There are already some useful examples the AI Dialogue can learn from. The UNESCO Recommendation on the Ethics of AI is a strong one because it gives a broad global framework built around human rights, dignity, transparency, accountability, and human oversight. It matters because it speaks to shared values, not just technical rules. The OECD AI Principles are also helpful because they have shaped a lot of international discussion around trustworthy and human-centred AI. They show that it is possible to agree on common directions even when countries have different legal systems and policy approaches. Another important example is the Council of Europe's Framework Convention on AI. What makes it especially valuable is its clear connection to human rights, democracy, and the rule of law. That helps keep AI governance focused on people, not only on innovation or competition. On the more practical side, the NIST AI Risk Management Framework is a good example of how organizations can actually work with AI risks in a structured way. It is useful because it turns big governance ideas into something more concrete and operational. At the same time, good AI governance is not only about major frameworks or formal agreements. It also depends on everyday practices: being clear about what an AI system can and cannot do, documenting how it works, keeping humans involved in important decisions, and helping people build the skills to use AI critically and responsibly.