Women At The Table / A+ Alliance for Inclusive Algorithms / AI & Equality Initiative

The Dialogue would succeed if it establishes that AI governance requires a broader foundation than the one currently available — and if it creates the conditions for that foundation to deepen over time. Every governance framework embeds assumptions about what a flourishing life looks like, whose needs are prioritized, what counts as harm. A successful Dialogue would make those assumptions explicit and contestable — not by adding more voices to an existing conversation, but by changing what the conversation is about. The communities with the deepest expertise in the domains AI is transforming — health, labor, climate, development, human rights — should be supported to articulate what they want AI to make possible, not only what they want to prevent. A successful Dialogue would also demonstrate that gender is a structural dimension of AI governance, not a parallel track. The impacts of AI on work, health, justice, care and social protection are profoundly gendered. Governance that does not account for this will not serve the majority of the world's population. Finally, a successful first Dialogue would not try to resolve everything in two days. It would launch a sustained process that activates substantive AI conversations within existing intergovernmental convenings upstream of future milestones — so that domain communities arrive at subsequent dialogues with visions already articulated, not starting from scratch. The first Dialogue succeeds if it is understood as a beginning, not a conclusion.

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

Social, economic, ethical, cultural, linguistic and technical implications of AI is where the conversation most urgently needs to be widened. AI governance discussions are rich in technical and regulatory expertise but thin on the domain knowledge of the communities most affected - the health practitioners, labor negotiators, climate scientists, and development economists who understand the systems AI is reshaping from the inside. Their knowledge is not a "stakeholder perspective" to be consulted. It is expertise without which governance frameworks will be built on incomplete assumptions. Protection and promotion of human rights matters not only as a safeguard but as a generative starting point. The question is not only how to prevent AI from violating rights, but what a future where rights are genuinely realized would require of the systems we build. Beginning from that aspiration produces different governance than beginning from risk alone. Transparency, accountability, and human oversight are governance preconditions - but they must be defined with input from the communities most affected, not in exclusively technical terms. What meaningful oversight looks like in a health system differs from what it looks like in a labor market or a justice system. Domain expertise shapes what these principles actually require in practice. Open-source software, open data and open AI models determine who can participate in building AI systems, who can scrutinize them, and whose visions can become more than aspirational. Openness is not only a technical question - it is a question about whether the plural, inclusive development that legitimate governance requires is structurally possible.

The most significant gap is the relationship between the themes themselves. The listed topics are presented as parallel tracks, but the most consequential governance challenges sit at their intersections - where labor meets development, where health meets climate, where rights cut across all of them. AI does not respect sectoral boundaries, and neither do its trade-offs. A worker displaced by automation is also a person navigating a health system, a community member affected by climate stress, a rights-holder. Governance that treats these dimensions in isolation will miss the systemic dynamics that determine whether frameworks work in practice. Gender is the clearest illustration of why cross-cutting architecture matters. It appears nowhere as a standalone theme and everywhere as a determining factor. Women constitute the majority of health, care, and informal economy workers globally. They are disproportionately affected by algorithmic decision-making in social protection, credit, and justice. Technology-facilitated gender-based violence is already excluding women from digital participation and from shaping the systems of the future. These are not women's issues - they are governance issues that become invisible when gender is treated as a separate concern rather than a lens applied across every theme. An emerging issue worth naming: the growing fragmentation of the AI governance community itself into specialized tribes - existential risk, AI safety, participatory AI, algorithmic accountability - each developing its own language and red lines. The connective tissue between these communities, and between them and the broader multilateral system, is thinning. The Dialogue has an opportunity to rebuild it.

Women at the Table works at the intersection of AI governance, gender equity, and inclusive international policy. What we see on the ground is a widening gap between the pace of AI deployment and the capacity of affected communities to shape its trajectory. In health systems, AI tools are being introduced into contexts where women deliver the vast majority of care — often in informal, underpaid, or unprotected roles — without their experience informing the design, deployment, or governance of those tools. In labor markets, the workers most exposed to AI-driven disruption — gig workers, informal economy workers, care workers — are disproportionately women and disproportionately absent from the rooms where transition frameworks are being designed. In justice and social protection systems, algorithmic decision-making is producing outcomes that entrench existing inequalities, with limited recourse for those affected. Technology-facilitated gender-based violence represents an acute and growing threat — not only to individual safety, but to women's participation in the digital spaces and governance processes where AI's future is being decided. Exclusion compounds: those most affected by AI's harms are least present in shaping its governance. The most significant opportunity we see is that the multilateral system already convenes the domain communities whose expertise is missing from AI governance — through the World Health Assembly, the ILC, the Human Rights Council, CSTD, and others. These processes have mandates, constituencies, and authority. What is missing is the activation of the AI governance question within them — on their terms, led by their expertise.

The Dialogue's greatest potential contribution is not to produce another governance framework. It is to become the place where frameworks connect. International cooperation on AI governance is currently fragmented — across institutions, across regions, across the specialized communities that have emerged around distinct aspects of the challenge. Existential risk, AI safety, algorithmic accountability, digital rights, development equity — each is doing important work, but increasingly in its own language, with its own assumptions, and with limited connective tissue to the others. Meanwhile, the broader multilateral system — where the world's health, labor, climate, human rights, and development communities convene — remains largely disconnected from the AI governance conversation, despite holding the domain expertise most relevant to its consequences. The Dialogue can serve as that connective infrastructure. Not by absorbing all of these conversations, but by creating a recurring space where they encounter each other — where the trade-offs between different visions, different priorities, and different regions become visible and negotiable rather than buried in parallel processes. To do this credibly, the Dialogue must practice genuine multi-stakeholder engagement — not as a performance of inclusion, but as a recognition that legitimate governance requires the participation of those whose lives are most affected. That includes ensuring gender is present as a structural dimension across every theme, and that affected communities are shaping questions, not only responding to them. If the Dialogue becomes the place where the AI governance conversation reconnects with the multilateral system and with the breadth of human expertise that system represents, it will have advanced international cooperation in a way no single framework can.

The most underleveraged assets for AI governance are not AI-specific initiatives — they are the existing intergovernmental processes where the world's domain expertise already convenes. The World Health Assembly, the International Labour Conference, the Human Rights Council, the Commission on Science and Technology for Development, the IPCC, the Commission on the Status of Women — these bodies hold mandates, constituencies, and deep knowledge of the systems AI is transforming. Yet they remain largely disconnected from the AI governance conversation, and it from them. The Dialogue's added value lies in becoming the bridge. Not by duplicating what these bodies do, but by catalyzing AI governance conversations within them — supporting health, labor, climate, rights, and development communities to engage with AI on their own terms — and then providing a space where those distinct perspectives meet, intersect, and inform governance architecture. The Dialogue should also build on the Global Digital Compact and the WSIS process, which provide a broader framework for digital cooperation but have not yet generated the depth of sectoral engagement that AI governance demands. Regional AI strategies and national governance initiatives offer important context, but the Dialogue's unique contribution is at the multilateral level — connecting what is otherwise fragmenting. Partnerships with civil society, academia, and affected communities are essential — not as consultation but as structural participation in shaping the agenda. Women's rights organizations and gender-focused networks should be engaged across all thematic areas, not siloed into dedicated gender sessions.

The most important structural decision is sequencing: who speaks first shapes everything that follows. If the Dialogue begins with technical presentations and invites other stakeholders to respond, it will reproduce the dynamic that has defined AI governance so far — domain communities positioned to react rather than to lead. An alternative: begin each thematic session with the communities most affected. Let a health practitioner define the unsolved problems in health systems before a technologist describes what AI could offer. Let a labor organizer articulate what meaningful work looks like before the conversation turns to automation. Let a rights defender describe what dignity requires before governance mechanisms are proposed. This is not a symbolic gesture — it produces substantively different conversations because it changes what counts as the starting question. Governments contribute institutional authority, mandate, and the capacity to translate outcomes into policy. They should be present throughout, but the Dialogue's value is diminished if government statements dominate the floor. Civil society and academia contribute domain expertise, lived experience, and the ability to hold governance accountable to the communities it claims to serve. Their role should be structural — shaping the agenda, not only populating panels. The private sector contributes technical knowledge and implementation capacity. Their participation is essential but should not set the frame. The Dialogue is most valuable when technical possibility responds to human aspiration, not the reverse. A written contribution mechanism — open before and after the Dialogue — would ensure that communities unable to attend in person can still shape the conversation and hold it accountable.

The most consequential absence is not a single community — it is an entire category of expertise. The AI governance conversation is rich in technology specialists, digital rights advocates, and trade negotiators. It is remarkably thin on the domain experts who understand the systems AI is transforming from the inside: public health professionals, climate scientists, labor economists, development practitioners, educators, social workers, disability rights advocates, indigenous knowledge holders. These are not "affected communities" waiting to be consulted. They hold expertise — built over decades, grounded in specific contexts — without which governance frameworks will be built on incomplete assumptions about how the world actually works. Women are underrepresented not only in numbers but in influence over the frame. When gender is present, it is typically siloed — a dedicated session rather than a lens applied across all themes. The result is that the gendered dimensions of labor, health, justice, and development remain invisible in the main governance conversation. Global South voices are present in multilateral settings but often positioned to respond to agendas set elsewhere. The question is not only whether they have a seat, but whether their visions for their own regions and citizens are shaping the questions the Dialogue asks. Inclusion requires more than invitation. It requires upstream engagement — supporting underrepresented communities to develop and articulate their positions before the Dialogue, so they arrive ready to lead rather than to react. It requires funding for participation. And it requires format choices that distribute authority over the conversation, not only access to the room.

The most transformative format innovation would happen before the Dialogue itself: curated preparatory conversations embedded within existing intergovernmental convenings in the months leading up to July. If the World Health Assembly, the ILC, the Human Rights Council, and CSTD each hosted a structured AI dialogue within or on the sidelines of their own programmes — designed by their own communities, starting from their own priorities — participants would arrive at the Global Dialogue with positions already developed and visions already articulated. The Dialogue then becomes a convergence point, not a starting point. Within the Dialogue itself, several format principles would strengthen engagement: Start sessions from domain expertise, not technology. Structure conversations so that the human question — what do we want, what is broken, what does flourishing look like — comes first, and technical possibility responds to it. Create structured encounters between communities that do not normally meet. A climate scientist and a labor negotiator in the same conversation will surface trade-offs and possibilities that neither would reach alone. These collisions are where the most generative thinking happens — and where the intersectional governance challenges become visible. Use small, curated, facilitated groups rather than large plenary statements. Meaningful engagement requires that people respond to each other, not perform for an audience. Plenary sessions can synthesize, but the real work happens in rooms of twenty, not two hundred. Ensure that the engagement formats themselves model the governance approach the Dialogue advocates — inclusive, deliberative, built on the conviction that legitimacy comes from genuine participation. If the format contradicts the message, the message will not land.

Some of the most instructive governance precedents for AI come from outside the technology sector entirely - from moments when societies faced complex, systemic challenges and built governance worthy of the complexity. The Belmont Report (1978) emerged from a bioethics crisis and established principles - respect for persons, beneficence, justice - that have governed human subjects research for nearly fifty years. It succeeded because it brought together not only scientists and regulators but ethicists, theologians, and community representatives, and because it started from values rather than from the technology it sought to govern. AI governance faces an analogous moment and has not yet produced an analogous process. Elinor Ostrom's work on governing the commons demonstrated that the most durable governance systems are built by the communities closest to the resource - not imposed from above. Her principles of collective management, graduated sanctions, and local accountability offer a direct challenge to the assumption that AI governance must be centralized and expert-driven in the narrow technical sense. More recently, the IPCC model - translating complex science into policy-relevant assessment through structured dialogue between scientists and governments - offers a template for how domain expertise can be made legible to governance processes without being flattened. AI governance lacks an equivalent mechanism for translating the expertise of health, labor, and development communities into governance architecture. What these examples share: governance that begins from the knowledge of affected communities, holds competing values in productive tension rather than resolving them prematurely, and builds legitimacy through participation rather than prescription. These are not AI-specific practices. They are governance practices the AI moment urgently needs.