Former UNHCR staff member

A successful Global Dialogue on AI Governance would be measured by clear, practical actions rather than statements. Key outcomes should include: agreed priority actions on AI safety, transparency, and accountability, plus voluntary pilot initiatives or regulatory sandboxes to test governance approaches in practice. It should also establish a concrete follow-up mechanism (working groups or task forces with timelines) to ensure continuity beyond the event. Finally, success means real commitments to capacity-building and knowledge-sharing, so all regions can participate in and benefit from AI governance.

AI capacity-building;

AI capacity-building is critical not only for institutions and countries, but also for individuals. Many people are still at a very early stage of understanding what AI is and how it works. There remains a significant knowledge gap in how to use AI appropriately-when to apply it, when not to, and how to use it responsibly in daily life and decision-making. Addressing this early-stage awareness is essential for inclusive and safe adoption. Human rights remain central in the AI era. Humanity fought for decades to establish values like equality, diversity, transparency, and inclusion. As AI systems increasingly shape decisions, opportunities, and narratives, we must actively defend and re-embed these values - otherwise, technology risks undoing hard-won progress.

One important gap is gender equality and meaningful inclusion in AI ecosystems. While human rights and social implications are covered broadly, the current AI landscape remains largely male-dominated, particularly in technical design and governance spaces. This can unintentionally shape systems in ways that overlook diverse perspectives. Ensuring greater participation of women is essential not only for equity, but also for improving the fairness and relevance of AI outcomes. Another emerging issue is the role of children and youth in AI governance and design. Children are increasingly exposed to AI through education, online platforms, and digital services, yet they are rarely included in discussions about how these systems are built or regulated. AI should be designed with them in mind and with their perspectives included, particularly in relation to safety, development, and digital literacy. A further cross-cutting concern is AI literacy at the individual level. While "AI capacity-building" is included, it is often framed at institutional or national levels. There is a growing need to ensure that individuals understand how AI works and how to use it responsibly, including when not to use it. These issues reinforce the need for a truly human-centered and inclusive approach to AI governance, ensuring that no group is left behind as AI systems continue to evolve and shape society.

In my context, the main governance gaps in AI are already visible in how quickly AI tools are being adopted compared to the slower development of regulatory, ethical, and capacity frameworks. One of the most significant challenges is the uneven level of AI literacy among individuals and institutions. Many users are adopting AI tools without fully understanding how to use them responsibly—when to rely on them, when not to, and how to critically assess outputs. This creates risks related to misinformation, over-reliance, and unintended misuse, particularly in education, work, and decision-making processes. A second major challenge is the lack of consistent governance and oversight frameworks across sectors and borders. While AI is being rapidly adopted, there is still uncertainty about accountability—many stakeholders continue to question who is responsible for the outputs and consequences of AI systems, whether it is the developer, the deployer, or the user. This ambiguity creates trust gaps and slows the establishment of clear standards for transparency and accountability. At the same time, with this ongoing uncertainty, there is also increasing discussion around automation, but it remains at a very early stage. At present, automation is largely limited to very specific tasks where human input is minimal or not critical. Broader claims about large-scale job replacement are still premature, as most real-world applications remain assistive rather than fully autonomous.

The AI Dialogue can play a key role in turning fragmented discussions into coordinated global action through a few practical steps. First, it can help align minimum global principles for AI governance—not as binding rules, but as shared reference points on safety, human rights, transparency, and accountability that countries can adapt nationally. Second, it can establish practical cooperation mechanisms, such as joint pilot projects, regulatory sandboxes, and testbeds where countries and stakeholders jointly experiment with governance approaches in real settings. Third, the Dialogue can support structured follow-up through thematic working groups with clear deliverables and timelines, ensuring continuity beyond annual meetings and avoiding a purely declaratory process. Finally, it can promote capacity-building partnerships, linking governments, private sector, and international organizations to close gaps in skills, literacy, and institutional readiness for AI governance. In short, the AI Dialogue should move beyond discussion and act as a coordination hub that connects principles, pilots, and partnerships into sustained international cooperation.

The AI Dialogue should build on and connect with existing initiatives such as the OECD AI Principles and Policy Observatory, the G7 Hiroshima AI Process, the Global Partnership on AI (GPAI), and UNESCO's work on AI ethics, which already provides a strong normative foundation. It should also take into account regional regulatory efforts and industry-led safety and governance commitments. Rather than duplicating these efforts, the added value of the AI Dialogue lies in its role as a global coordination and convergence platform under the UN system, bringing together fragmented initiatives into a more inclusive and coherent space. First, it can help bridge policy gaps between regions, ensuring that countries with lower technical and regulatory capacity are not left out of global governance discussions. Second, it can focus on translating principles into action, by promoting joint pilots, regulatory experimentation (sandboxes), and shared approaches to risk assessment and accountability. Third, it can strengthen interoperability between existing frameworks, identifying where alignment is possible and where flexibility is needed across different governance models. Finally, it can support practical capacity-building and knowledge exchange, ensuring that governance is not only discussed at global level but also operationalized at national and institutional levels. In short, the AI Dialogue's value is not to replace existing initiatives like UNESCO's, but to connect them, reduce fragmentation, and accelerate practical, inclusive implementation of AI governance globally.

Action-oriented Dialogue with pilot initiatives and follow-up mechanism The structure of the AI Dialogue should go beyond annual discussions and include practical implementation tracks, such as voluntary pilot projects, regulatory sandboxes, and cross-border experiments. In addition, a permanent light coordination mechanism should track progress between sessions, share lessons learned, and ensure continuity. This would help turn dialogue into sustained action and measurable impact.

One key gap is the Global South, particularly low- and middle-income countries. Many of these states face AI primarily as users rather than developers, yet they are often not sufficiently represented in shaping global rules. This can be addressed through dedicated seats in dialogue platforms, targeted capacity-building, and support for regional consultations feeding directly into global processes.

Problem-solving "policy labs" could be used instead of traditional panel discussions. These would bring together governments, experts, the private sector, and civil society to work on specific real-world AI governance challenges (e.g., accountability, bias, or AI in public services) and produce joint practical outputs such as draft frameworks or pilot proposals within the session.

OECD AI Principles and OECD AI Policy Observatory, UNESCO's Recommendation on the Ethics of Artificial Intelligence European Union AI Act Global Partnership on AI (GPAI) G7 Hiroshima AI Process AI safety commitments by major technology companies