National Institute of Development Administration, Thailand

In my opinion, the first Global Dialogue on AI Governance would be successful if it produces practical, inclusive, and lasting outcomes rather than only broad statements of principle. First, success would mean establishing a shared baseline for trustworthy AI governance: transparency, accountability, human oversight, safety, and respect for human rights. Countries may adopt different legal and regulatory models, but the Dialogue should help identify a minimum set of common expectations that can guide interoperable governance across jurisdictions. Second, the Dialogue should produce concrete recommendations for real-world use. As an academic and frequent user of generative AI, I have experienced clear benefits: AI can accelerate drafting, support multilingual communication, assist idea development, and widen access to knowledge tools. At the same time, regular use also reveals persistent risks, including factual inaccuracies, fabricated citations, embedded bias, opaque system behavior, and user overreliance. A successful Dialogue should therefore move beyond abstract ethics and identify practical safeguards such as provenance and disclosure standards, documentation requirements, risk-tiered oversight, auditability, and accessible mechanisms for contestation and redress. Third, success would require meaningful inclusion of voices from developing countries, educators, researchers, civil society, and public-interest institutions. AI governance should not be shaped only by the most technologically advanced states or large companies. Capacity-building, equitable access, and institutional readiness must be treated as governance priorities. Finally, success would mean follow-through: a roadmap, clear workstreams, and a mechanism for continued reporting and cooperation. The Dialogue should not end as a symbolic event. It should become a durable platform that helps the international community translate shared concerns into coordinated action.

• Open-source software, open data and open AI models
• Transparency, accountability, and human oversight
• Safe, secure and trustworthy AI
• Social, economic, ethical, cultural, linguistic and technical implications of AI

In my opinion, First, safe, secure and trustworthy AI is a core priority because trust is the foundation for any meaningful social adoption of AI. In practice, users need systems that are reliable, resilient, and designed to reduce harm, especially in education, research, public communication, and knowledge production. Second, the social, economic, ethical, cultural, linguistic and technical implications of AI are highly relevant because AI does not affect all communities equally. From my perspective, urgent governance must pay closer attention to unequal access, linguistic inclusion, cultural representation, and the risk that globally deployed models may reproduce biases that marginalize non-dominant languages and contexts. Third, transparency, accountability, and human oversight are essential because AI systems increasingly shape writing, decision support, and information flows. Users and institutions need to understand the limits of these systems, identify who is responsible when harm occurs, and ensure that meaningful human judgment remains in place. Fourth, open-source software, open data and open AI models matter because openness can support innovation, research, capacity-building, and more equitable participation, especially for developing countries and public-interest institutions. At the same time, open ecosystems should be accompanied by appropriate safeguards, documentation, and risk management.

One is "information integrity and content provenance". As generative AI becomes embedded in communication, education, and public discourse, governance should address not only model safety but also the ability to verify whether content is synthetic, edited, or authentic. This is especially urgent for academic work, journalism, public communication, and democratic deliberation, where misleading outputs can circulate quickly and at scale. A second issue is "AI literacy and institutional readiness". The Global Dialogue is intended as an inclusive forum for governments and stakeholders, but meaningful participation depends on whether institutions and users have the capacity to evaluate, govern, and use AI responsibly. In practice, many of the risks of AI do not arise only from the technology itself, but from uneven human understanding, weak internal policies, and limited governance capacity across countries and sectors. A third issue is "power concentration across the AI value chain", including compute, cloud infrastructure, foundation models, and platform access. This matters because formal openness alone does not guarantee equitable participation if technical and financial control remains concentrated in a small number of actors. Finally, I would emphasize "environmental sustainability" as a cross-cutting issue. Discussions of safety, transparency, and open models should also consider energy use, resource consumption, and the long-term sustainability of AI deployment. This is particularly important if international governance aims to align AI development with broader UN sustainable development objectives. The UN's current AI framework already links the Dialogue to international cooperation, open-source and open AI models, and broader digital public goods; sustainability should be made equally visible within that agenda.

In Thailand, the governance gaps in safe, trustworthy AI, transparency and human oversight, and open models/open data are felt as a problem of uneven implementation rather than lack of policy attention. Thailand has already built important foundations through ETDA's AI Governance Center, national governance guidelines for organizations, and a 2025 public consultation on draft AI law principles focused on high-risk AI. At the same time, UNESCO's recent AI Readiness Assessment for Thailand identified gaps in policy coherence, institutional coordination, and capacity development, which means that governance still moves faster in some sectors than in others. The most significant challenge is the gap between AI ambition and broad-based readiness. The World Bank reports that Thailand's AI adoption remains low: only 6 percent of internet users were accessing generative AI tools as of March 2024, the lowest rate in ASEAN, with barriers including low digital literacy, limited awareness, weak adoption incentives for small firms, language limitations, and shortages of professionals able to fine-tune open-source models for local needs. In parallel, Thai businesses are clearly moving toward AI use, but governance maturity is uneven: PwC reports that 73.3 percent of organizations had used GenAI to support operations over the previous 12 months, while AI use in compliance processes remained much lower, and firms highlighted the need for stronger internal controls and data protection to prevent leaks and misuse. The opportunity is substantial. Thailand already has strong consumer-side digital engagement, active open-government-data infrastructure, and policy momentum around AI for agriculture, health, and education. This creates a realistic pathway for Thailand to become a regional model for human-centered, Thai-language, sector-specific AI governance if it can now convert scattered initiatives into interoperable rules, stronger oversight capacity, workforce development, and wider access to trusted open data and public-interest AI tools.

I think it can play its most valuable role by becoming the place where international AI governance moves from fragmented debate to practical coordination. The UN has already framed it as an inclusive platform for governments and stakeholders to discuss international cooperation, share best practices and lessons learned, and support open, transparent, and inclusive discussions on AI governance. In practical terms, that means four things. First, it can help build a common baseline across jurisdictions: not identical laws, but shared understandings on safety, transparency, accountability, and human oversight. Second, it can connect science to policy by translating the Independent International Scientific Panel's evidence into governance priorities; the UN has explicitly said the Panel's first report is intended to inform the July 2026 Dialogue. Third, the Dialogue can reduce governance inequality by ensuring that developing countries are not merely rule-takers. UN discussions around the process have repeatedly stressed the need for the meaningful participation and full inclusion of developing countries in shaping AI governance. Fourth, it can catalyze cooperation beyond principles by encouraging concrete joint work on capacity-building, standards interoperability, open and trusted public-interest AI resources, and mechanisms for ongoing follow-up.

1. UNESCO Recommendation on the Ethics of AI gives a widely recognized ethical foundation, and the AI Dialogue could add value by turning shared principles into more coordinated international practice. 2. OECD AI Principles already provide a strong policy baseline for trustworthy AI, and the AI Dialogue could broaden that foundation into a more globally inclusive conversation. 3. GPAI should be connected because it already links governments with expert communities, while the AI Dialogue can bring that work into a broader UN-based multilateral setting. 4. The G7 Hiroshima AI Process offers useful guidance on advanced AI, and the AI Dialogue could add value by opening that agenda to wider participation beyond major economies. 5. ITU's AI for Good is already building work on standards, skills, and practical AI solutions, and the AI Dialogue can strengthen its impact by connecting those efforts more directly to intergovernmental governance discussions.

A useful approach I think that it is to treat AI use in research as normal, but require honest disclosure. Recent journal guidance such as ICMJE does not simply ban AI use; instead, it asks authors to explain how AI was used and makes clear that humans remain fully responsible for the accuracy, integrity, and citations of the work. Another strong example I think it is human-centred governance in education and research. UNESCO emphasizes capacity-building, ethical use, and institutional readiness, which is important because good governance is not only about controlling AI, but also about helping people use it well. Finally, I think labour policy should focus on skills transition, not only job loss. Recent ILO analysis suggests that generative AI is more likely to transform many jobs than simply eliminate them, so governance should support reskilling, new competencies, and the creation of new forms of work.