Centro para el Desarrollo del Pensamiento Etico, Oficina de Etica Gubernamental

In my view, the success of the first Global Dialogue on AI Governance would be measured not only by the commitments made, but by the depth and inclusivity of the conversation that produces them. A successful outcome would begin with the direct acknowledgment of digital divides, particularly in the Global South, and clear recognition that the unchecked development of AI technologies risks further widening existing social and economic vulnerabilities. Any governance framework that does not start from this premise will fall short. From there, we will consider this Dialogue a success if it produces an articulated agenda for AI development that is genuinely aligned with the Sustainable Development Goals, with particular attention to the most vulnerable societies. This agenda must go beyond efficiency and innovation, and center the protection of citizens' rights, institutional transparency, and ethical responsibility in administrative decision-making. Equally important is reaching consensus on recognizing women, children, and the elderly as groups intrinsically and disproportionately affected by the digital divide. A successful Dialogue would not only acknowledge this reality but commit to a strategic work plan that concretely addresses these populations. Finally, I believe success requires establishing shared principles for how governments implement AI, not just whether they do so, grounded in legal, ethical, and institutional accountability. Artificial intelligence holds real potential to improve public services and strengthen the State. But that potential will only be realized if this Dialogue produces more than declarations. It must produce actionable frameworks, inclusive commitments, and a shared responsibility to ensure that AI governance works for everyone; especially those most at risk of being left behind.

• Transparency, accountability, and human oversight
• Protection and promotion of human rights
• Social, economic, ethical, cultural, linguistic and technical implications of AI

Technology is not simply science; it is the application of knowledge to reshape our environment and meet human needs. From this perspective, artificial intelligence is not merely a technical innovation. It is a tool capable of reconfiguring administrative processes, institutional structures, and decision-making dynamics within the State. For this reason, its implementation cannot be approached from a purely technological standpoint. It is not enough for AI systems to function; they must behave responsibly. The real challenge is not adoption but achieving reliable and sustainable institutional transformation. Many initiatives fail because they attempt to implement AI without first rethinking the administrative processes they seek to improve. AI cannot simply be used to automate inefficient bureaucracies. Incorporating AI into public administration raises fundamental questions about accountability. Any responsible implementation must include three essential elements: algorithmic transparency, oversight mechanisms, and periodic technical audits. Algorithms cannot become black boxes within the State. If a public decision affects citizens' rights, that decision must be explainable, reviewable, and open to challenge. Clear legal frameworks must define the scope, limits, and responsibilities of these systems. Finally, there is no universal solution. AI in public administration cannot follow a one-size-fits-all model. Each system must be designed around a fundamental question: what administrative problem are we trying to solve? This requires identifying which dimension of e-government is being addressed - government-to-citizen, government-to-government, or government-to-business - as each respond to distinct institutional dynamics and requires different technological approaches.

Artificial intelligence has the potential to significantly improve the efficiency of the State, institutional transparency, and the quality of public services. But that potential will only be realized if its implementation is carried out with legal, ethical, and institutional responsibility. Integrating AI into public administration requires a multidimensional approach that balances operational efficiency with robust ethical and social safeguards. This means, first and foremost, maintaining a human-centered focus; AI must serve as a tool in the service of people, respecting human dignity, individual autonomy, and core principles such as transparency, accountability, privacy, and non-discrimination. Equally critical is ensuring that implementation reduces, rather than widens, existing digital divides. This requires designing inclusive systems from the outset, accounting for inequalities related to gender, age, income, geographic access, and accessibility for people with disabilities. High-risk AI applications affecting employment, education, access to public services, or justice, must be subject to impact assessments, external audits, and stricter regulatory standards. Citizens must also know when they are interacting with AI systems and be able to understand and challenge automated decisions that affect them. For high-impact public decisions, meaningful human oversight remains indispensable. Effective AI governance also requires quality data, safe experimentation spaces such as regulatory sandboxes, and technological sovereignty strategies that allow States to build their own capacities and reduce dependence on external providers. At the Center for the Development of Ethical Thinking, we hold a central epistemic premise: ethical reflection is not an obstacle to technological innovation; it is its condition of possibility. Innovation that does not incorporate ethical frameworks lacks the foundations necessary to be legitimate, sustainable, and consistent with human dignity. In this regard, transdisciplinary education is essential to addressing the challenges posed by emerging technologies. Puerto Rico currently lacks a public policy framework that addresses these elements. As a result, the adoption of these technologies does not align with the intrinsic needs of a population that includes a significant proportion of older adults with limited digital literacy. Compounding this, deep-seated distrust in government institutions and a high public perception of corruption further intensify the already existing challenges — making responsible, inclusive, and transparent AI governance not just desirable, but urgent.

The first Global Dialogue on AI Governance arrives at a defining moment. Artificial intelligence is already reshaping public administration, institutional decision-making, and the delivery of public services — yet the governance frameworks needed to guide this transformation remain fragmented, uneven, and in many cases, absent. In this context, the Dialogue can play a foundational role — but only if it moves beyond declarations and toward binding, actionable commitments. First, the Dialogue must serve as a space for epistemic equity. The dominant narratives around AI governance have largely been shaped by technologically advanced nations and private sector actors. International cooperation must actively elevate the voices of the Global South, small island developing states, and communities disproportionately affected by digital divides — including women, older adults, and persons with disabilities. Second, the Dialogue can establish a shared baseline of principles: algorithmic transparency, meaningful human oversight, risk-based governance, and the protection of citizens' rights in automated decision-making. These cannot remain aspirational — they must form the foundation of interoperable regulatory frameworks that transcend national borders. Third, international cooperation must address the structural asymmetries that governance gaps create. Countries like Puerto Rico, lacking dedicated AI public policy, are particularly vulnerable. The Dialogue should catalyze mechanisms for technical assistance, capacity building, and the co-development of governance tools tailored to diverse institutional realities. Finally, the Dialogue must anchor AI governance explicitly within the Sustainable Development Goals — recognizing that unchecked AI development can deepen inequality just as readily as it can reduce it. At its best, this Dialogue can become the foundation of a global governance architecture that ensures AI serves humanity — not the other way around.

A credible and effective AI Dialogue cannot be built from scratch. It must deliberately connect with, and build upon, the body of work already developed by international institutions, civil society, and regional bodies — while adding the political weight and inclusive legitimacy that only a UN-convened process can provide. Several existing frameworks offer essential foundations. The UNESCO Recommendation on the Ethics of AI (2021) established a globally endorsed set of principles centered on human dignity, transparency, and inclusion — particularly its attention to gender equity and cultural diversity. The OECD AI Principles and the EU AI Act have advanced risk-based governance models and accountability mechanisms that can inform interoperable regulatory standards. The ITU's AI for Good platform and the UN Secretary-General's Roadmap for Digital Cooperation have further mapped the intersection of AI with sustainable development. The work of the High-level Advisory Body on AI provides a direct institutional bridge into this Dialogue. The Dialogue should also connect with regional initiatives — including those emerging from Latin America and the Caribbean — where governance gaps are acutely felt and locally driven solutions are being developed, often without adequate international visibility or support. The added value of the AI Dialogue lies precisely in what these existing mechanisms cannot fully provide: a universal, intergovernmental space where equity is a structural condition — not an afterthought. It can bridge the gap between norm-setting and implementation, between technologically advanced nations and those still building foundational digital infrastructure, and between global policy frameworks and the lived realities of the most vulnerable populations. For communities like Puerto Rico — navigating AI adoption without a dedicated public policy framework, with aging populations and institutional trust deficits — the Dialogue represents an opportunity to ensure that global governance architecture reflects local realities, not just the priorities of the most powerful actors. The Dialogue's ultimate added value is legitimacy through inclusion: ensuring that the rules governing AI are made by, and for, all of humanity.

The legitimacy of any global governance process is inseparable from the quality of participation it enables. For the AI Dialogue to produce outcomes that are both credible and actionable, its format and structure must reflect the same principles it seeks to advance: transparency, inclusion, and accountability. Who must have a seat at the table Meaningful participation requires deliberate structural design. Member States must engage not only through foreign affairs ministries, but through the agencies and institutions directly responsible for digital policy, public administration, and social protection. Civil society organizations — particularly those working at the intersection of technology, human rights, and social inequality — must be recognized as substantive contributors, not merely observers. Academic and research institutions bring the transdisciplinary perspective essential for evidence-based governance. The private sector has a role to play, provided that participation is governed by clear conflict-of-interest safeguards. Critically, the Dialogue must create dedicated channels for voices that are systematically underrepresented in global tech governance: communities from the Global South, Indigenous peoples, older adults, persons with disabilities, and women's rights organizations. Recommendations for format and structure The Dialogue should move beyond plenary declarations toward working formats that produce tangible outputs — thematic working groups organized around the four priority areas, with clear mandates, timelines, and accountability mechanisms. Preparatory regional consultations should feed directly into the global process, ensuring that local realities shape global frameworks rather than the reverse. Hybrid participation modalities are essential to ensure geographic and economic inclusion. Simultaneous interpretation, accessible formats, and advance publication of documents are baseline requirements — not optional accommodations. Finally, the Dialogue must establish a follow-up mechanism. A one-time convening, however well-structured, is insufficient. Sustained international cooperation on AI governance requires a permanent, inclusive, and transparent review process. The format must match the ambition. If the goal is governance for all of humanity, the structure must make that possible.

A recurring problem in global conversations on issues that impact humanity is the insufficient representation of civil society and academia. While dedicated dialogue spaces are created and contribute meaningfully to the conversation, these actors rarely hold decision-making power proportional to the quality of their contributions. Civil society, the third sector, and academic institutions can serve as vital bridges between the bureaucratic scaffolding of the State and the real needs of communities. Their participation must be structural, not ceremonial. Beyond this, entire populations remain systematically invisible in global AI governance discussions. Women, older adults, persons with disabilities, Indigenous communities, and low-income populations are rarely centered in the frameworks being designed — despite being among those most acutely affected by both the promises and the risks of AI. Puerto Rico represents a particularly illustrative case. Due to our complex political relationship with the United States, we are frequently absent from international dialogue tables, and excluded from the studies, statistics, and policy frameworks that shape global governance. Yet the challenges we face, an aging population with limited digital literacy, the absence of a dedicated AI public policy framework, and deep institutional trust deficits, are precisely the kinds of realities that global governance must account for. Being rendered statistically and politically invisible does not make our vulnerabilities any less urgent. Inclusion cannot be an afterthought. The AI Dialogue must establish guaranteed participation mechanisms for civil society and academia with substantive, not merely advisory, roles. It must also develop specific outreach strategies for territories and communities that fall outside conventional diplomatic representation, ensuring that governance frameworks are informed by the full spectrum of human experience, not only by those with the loudest institutional voice.

Effective engagement requires formats that are as thoughtfully designed as the governance frameworks they seek to produce. A dialogue that aspires to be truly global must move beyond traditional plenary sessions and adopt structures that generate concrete, regionally grounded, and actionable outcomes. We recommend anchoring the Dialogue around regional working tables — structured spaces where participants from shared geographic, cultural, and developmental contexts can identify common challenges, priorities, and solutions before bringing them into the broader global conversation. This bottom-up architecture ensures that global frameworks are shaped by local realities, not imposed upon them. These working tables should be organized around clearly defined thematic pillars — such as transparency and accountability, digital inclusion, human rights protection, and risk-based governance — with each pillar informing the design of strategic work plans tailored to the specific populations and regions they affect. Critically, this process must begin with a comprehensive needs assessment. Before designing solutions, the Dialogue requires a clear conceptual map of current realities: where the greatest challenges reside, which populations are most vulnerable, and what governance capacities already exist at the regional and national level. This evidence base is the foundation upon which meaningful engagement — and credible outcomes — must be built. Each working table should integrate both technical and political actors in deliberate balance. Technical experts provide the analytical depth necessary to design effective governance tools; political actors provide the institutional authority needed to translate those tools into binding commitments. Neither alone is sufficient. Finally, hybrid participation modalities, multilingual facilitation, and accessible formats are not logistical details — they are equity commitments that determine whose knowledge counts and whose experience shapes the outcome.