KAIA Network

• Success requires three tangible outcomes: First, binding commitments, with clear accountability mechanisms, not aspirational statements. Governments must commit to human rights-centered AI regulation
• the private sector to transparency standards and risk assessment
• and civil society recognized for its role in supporting independent monitoring. Each stakeholder has competing interests, e.g., the private sector prioritizes innovation and profit, governments must balance growth with rights protection, and civil society holds both accountable. Success means designing processes that acknowledge these tensions while creating aligned incentives. Structural collaboration mechanisms that move beyond parallel advocacy will be needed. Currently, stakeholders speak to their constituents rather than with each other. True dialogue means governments and companies negotiating standards together. Third, concrete metrics to monitor success will be needed to hold all parties accountable. SDGs and MDGs have shown us the value of indicators, and these dialogues should have a similar approach to evaluation. What are the objectives, and how concretely will we assess whether they've been achieved?

• Social, economic, ethical, cultural, linguistic and technical implications of AI
• Protection and promotion of human rights
• AI capacity-building

AI capacity-building addresses the widening Global North-South divide. Developing regions lack computing infrastructure, talent pipelines, and research ecosystems. But capacity-building doesn't mean that every country must become a leader in each layer of AI (the full stack). Regional collaboration hubs, access to shared infrastructure, and targeted investment in domain applications (agriculture, health, education) are some options, but this will require dialogue and followed up with investments. Social, economic, and ethical implications demand prioritization because resources invested in AI-for-profit vastly exceed those for the public good. The asymmetry is severe: billions flow to generative AI while applications addressing global poverty remain underfunded. Governments must actively incentivize researchers and companies to develop solutions for pressing development challenges. Human rights protection is the last area I selected, it is essential because the UN is uniquely positioned to establish global norms. Without international frameworks, rights protections become voluntary and fragmented. AI's automation capacity threatens livelihoods; surveillance capabilities threaten privacy; algorithmic bias threatens equal protection. These require binding standards, not corporate self-regulation. Collectively, these themes address who benefits from AI (equity), what AI is built for (public versus private good), and how it's governed (rights-based standards). They're inseparable.

Yes, the 'human infrastructure' challenge. AI development remains concentrated among Global North elites (male-dominated, academically-credentialed, Western-trained). This homogeneity produces tools reflecting narrow values. Success requires deliberately connecting researchers, practitioners, and domain experts across geographies, from epidemiologists in East Africa, agricultural scientists in South Asia, to educators in Latin America. Use of AI/ML tools and techniques within coordinated networks focused on public challenges can unlock breakthroughs that were previously seen as lightyears away, but investment in people, not just machines, will determine whether AI serves broad development goals or reproduces existing inequalities. The Knowledge and AI Network for All (KAIA Network), a global network focused on making AI more inclusive, is one such initiative, but we need recognition of this gap in the AI stack at all levels and policy responses to address it.

Governance failures disproportionately affect the most vulnerable in society, but in our research sector, gaps in governance and investment affect social scientists working on issues that are not prioritized (e.g., barriers to gender equality), and those in the Global South, with limited access to frontier models and networks working on this cutting-edge technology. Female researchers, drastically underrepresented in AI/ML space (only 12% of ML researchers globally are women) are also affected. Moreover, investment and institutional support flow overwhelmingly toward commercial applications, starving development-focused research of resources. The disconnect between the technology and development sectors creates a fundamental misalignment. AI innovations optimized for profit rarely translate into solutions for problems that lack immediate commercial value. This leaves critical social justice challenges inadequately addressed by AI advances and governance vacuums prevent collaborative action. Fragmentation among researchers and stakeholders makes it impossible to establish shared safeguards against harm or coordinate efforts to maximize positive impact. The Path Forward: Strong governance must play a dual role: actively mitigating AI-related harms, e.g. algorithmic bias, surveillance risks, exploitative labor practices, while simultaneously incentivizing responsible innovation for public benefit. This requires moving beyond regulation-as-restriction to governance-as-catalyst. Without intervention, the status quo will persist: technological advancement serving narrow interests while profound social challenges remain unaddressed. Governance frameworks can rebalance incentive structures. We need targeted mechanisms, e.g., tax incentives, grant programs, that encourage AI-for-social-good projects.

The AI Dialogue can facilitate cross-sector collaboration. By creating neutral spaces and shared standards, we can bridge the technology-development divide, enabling researchers globally to access tools and expertise previously siloed in corporate environments.

Governance reform isn't peripheral to AI's future; it's fundamental to whether this technology becomes a tool for broad-based human flourishing or reinforces existing inequalities. Initiatives like the Knowledge and AI for All Network (KAIA Network) are a concrete example of how partnerships and collaborations across tech, academia, and civil society can be incentivized. It's a collaborative platform where technologists seeking purpose meet researchers, practitioners, and funders seeking impact. We need more of these initiatives, and the ones that exist need more support and funding. AI Dialogue can be a place where these initiatives are showcased, and other countries and stakeholders learn about them and how they can inform their own context. By working with KAIA Network and others, it can also function as a knowledge bridge, explicitly connecting technological innovation communities with development-focused researchers and practitioners.

• The AI Dialogue's strength depends on inclusive participation. Technologists and AI companies contribute expertise and innovation capacity
• academics bring research rigor and long-term perspective
• civil society and NGOs represent vulnerable communities and equity concerns
• policymakers translate dialogue into governance action
• and Global South researchers and practitioners ensure solutions address contextualized challenges rather than imposing external agendas. The AI Dialogue can be a place that truly brings all of these stakeholders together for genuine dialogue.

Women researchers, women entrepreneurs, and people of all genders interested in exploring and expanding the provision of AI for public good projects and initiatives.

Organize small working groups (15-20 people) bringing together technologists, practitioners, policymakers, and affected communities around specific challenges (e.g., "AI for health equity in low-resource settings", "ML methodologies to improve estimates of female poverty"). Rather than expert-to-audience knowledge transfer, create horizontal learning communities in which practitioners and researchers from different contexts share experiences in addressing similar challenges. Time-bounded intensive sprints where mixed teams prototypically develop governance solutions, e.g. "AI innovation for democracy, better public services and informed citizenship." Come up with formats and dialogue that show the UN is interested in fostering more uses of this powerful technology, in the public interest, countering head-on the perception that AI governance dialogue is only about curtailing AI or stifling innovation.

see www.kaia.network or reach out directly to azconasg@newschool.edu for examples of AI projects focused on AI uses/AI innovations in the interest of the public good.