T.I.M.E

Success for the first Global Dialogue hinges on a fundamental shift from a Global North-centric regulatory paradigm to a truly universal one. As a policy professional specializing in education systems within the Global South, I believe success must be measured by how effectively the Dialogue integrates the "classroom realities" of large-population, under-resourced contexts. A key outcome would be the establishment of international benchmarks for "pedagogical appropriateness" in AI tools. Governance must ensure AI empowers human agency rather than automating teacher roles into obsolescence. Furthermore, a successful Dialogue will produce a clear roadmap for a "Global Fund for AI" that prioritizes linguistic sovereignty. This means ensuring AI development isn't just an English-first endeavor but supports the 22 official languages of India and other diverse regions. Finally, success requires a commitment to equitable resource allocation, ensuring the "AI divide" doesn't become a permanent feature of global education. By the end of this Dialogue, we should have a shared vision where governance isn't just about preventing harm, but about actively enabling inclusive growth in emerging markets.

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

My selection is driven by the urgent need for linguistic sovereignty and structural capacity-building in emerging markets. In India, AI cannot be inclusive if it does not function across 22 official languages; thus, cultural and linguistic implications are a primary concern for equitable access. Without transparency and accountability, AI systems risk reinforcing historical biases in sensitive areas like high-stakes education assessment. I prioritize "capacity-building" not merely as teacher training, but as a structural policy design that protects teacher autonomy and readiness. Furthermore, promoting Digital Public Infrastructure (DPI) through open-source software and data is critical to prevent "digital colonialism". By focusing on these areas, we ensure that AI tools remain public goods rather than proprietary black boxes, allowing countries in the Global South to build local, context-aware learning systems.

Two significant cross-cutting issues are not adequately captured by the existing thematic areas. First, AI governance for children and learners specifically. The existing themes address human rights generically. But children in educational contexts face distinct risks that require dedicated governance attention: algorithmic bias in assessment tools that systematically disadvantages certain demographic groups; data collection at scale without meaningful consent mechanisms appropriate for minors; and AI systems that optimise for measurable outcomes - test scores, completion rates - at the expense of holistic learning and developmental wellbeing. The UNESCO Recommendation on the Ethics of AI addresses education as one of many policy areas. The UNICEF Policy Guidance on AI for Children provides a more focused framework. Neither, however, has produced binding governance norms. A dedicated thematic area on AI in education, with specific attention to learner rights, would fill this gap. Second is the Infrastructure-Pedagogy Gap. Global policy often assumes reliable high-speed internet and high device-to-student ratios, yet the reality in many Global South schools involves "offline-first" or low-bandwidth environments. Governance must address how AI can be deployed safely and effectively in these under-resourced contexts without requiring constant cloud connectivity.

The governance gaps in AI for education are not abstract in the Indian context. They are visible, measurable, and widening. Three specific challenges are most significant. First, linguistic inequality in AI systems. India's 22 official languages and hundreds of regional dialects mean that AI tools trained predominantly on English-language data perform significantly less well for the majority of Indian students. No current international governance framework mandates linguistic equity testing as a condition of deployment. This is a gap that disproportionately harms students from already disadvantaged communities. Second, the gender digital divide. According to the GSMA Mobile Gender Gap Report 2025, women in India are 33% less likely than men to use mobile internet — and this gap is widening, with the smartphone gender gap growing from 32% in 2023 to 39% in 2024. Deploying AI-enabled learning tools in this context without equity safeguards risks deepening, not closing, the education gender gap. Third, inadequate data protection for learners. India's Digital Personal Data Protection Act 2023 is a meaningful step forward, but it was not designed with the school classroom in mind. Critical questions remain unanswered in policy: who owns a child's learning data? Who can monetise it? What happens when an EdTech vendor's contract ends? These questions need answers in governance frameworks — not just in vendor terms and conditions. The opportunity is equally significant. India's scale — 250 million students across a diversity of languages, income levels, and infrastructure conditions — makes it one of the world's most instructive governance laboratories. What works here transfers directly to other emerging economies. The Dialogue should treat this not as a problem to be managed but as an evidence base to be learned from.

The AI Dialogue's most distinctive potential contribution is one that no individual country or regional framework can replicate: legitimate convening power in the service of coherence. Multiple governance frameworks already exist. The OECD AI Principles have been adopted by 46+ countries. The UNESCO Recommendation on the Ethics of AI has 193 member state endorsement. The EU AI Act is the most detailed regulatory instrument currently in force. The G7 Hiroshima AI Process has produced an international code of conduct. These frameworks exist in parallel, are inconsistently referenced, and create compliance complexity for countries — particularly in the Global South — that lack the regulatory capacity to navigate multiple overlapping frameworks simultaneously. The Dialogue can play three specific roles in advancing coherence. First, it can map the existing framework landscape clearly — identifying where frameworks align, where they conflict, and where they are silent — particularly on education, children's rights, and Global South contexts. Second, it can create a legitimating function for implementation guidance. The OECD Principles and UNESCO Recommendation are strong on values and weak on implementation. The Dialogue can commission and validate practical guidance on how these principles apply in specific contexts — including large-scale, multilingual, low-resource education systems. Third, it can establish a shared, publicly accessible evidence base. Currently, Global South evidence in AI governance discussions is largely anecdotal. A structured, curated repository of case studies — including both successes and failures — would give all countries a common reference point and would shift the Dialogue from a talking shop to a knowledge institution.

The Dialogue should build directly upon the UNESCO Recommendation on the Ethics of Artificial Intelligence, which is currently the only global normative instrument covering education extensively. It should also align with the OECD AI Principles, which have been adopted by 46+ countries as a foundation for national strategies. In the Indian context, the principles within NEP 2020 regarding technology-integrated learning offer a valuable roadmap for emerging economies. The added value of the AI Dialogue is its multilateral legitimacy. While individual initiatives provide excellent guidelines, the Dialogue can synthesize these into a cohesive, globally-accepted framework. It can turn static recommendations into "living" policy tools that help ministries of education negotiate with global EdTech providers, ensuring that international standards for transparency and data privacy are met before tools are deployed at scale.

I recommend a "Bottom-Up" format that moves beyond high-level diplomatic representatives. The structure should include "Policy Hackathons" where frontline educators, local EdTech founders, and students from emerging markets work alongside regulators to co-design frameworks. This ensures that policies are grounded in the practical realities of the classroom. To reduce travel and financial barriers, the Dialogue should utilize decentralized Regional Policy Labs. These labs would allow local stakeholders to provide input in their own languages and contexts, with findings fed directly into the central UN sessions. By using a tiered structure—local, regional, and then global—the Dialogue can ensure that the final governance frameworks are not just inclusive in name, but truly representative of the global learner population.

Frontline educators in the Global South are systematically underrepresented. They are the "street-level bureaucrats" of AI policy, yet they are rarely invited to the table when global standards are drafted. Similarly, the perspectives of girls in emerging markets—who face unique structural barriers to digital access—are often missing. To include these voices, we must establish quota-based representation in working groups. We can also use "Citizen Assemblies" and AI-powered collective intelligence tools to gather real-time feedback from rural educators who cannot travel to international summits. Policy professionals must act as bridges, translating these ground-level insights into the technical and legal language required for UN-level governance. The inclusion mechanism is not complicated: structured pre-dialogue regional consultations with mandatory civil society input that is formally incorporated — not merely noted — in the main Dialogue agenda.

We should move beyond the traditional "prepared speech" format toward Policy Simulations and Digital Twins. By simulating the impact of a specific regulation on a virtual model of a rural school network, participants can visualize the consequences of their decisions before they are enacted. This makes abstract policy concepts concrete and fosters deeper, data-driven debate. Example: an AI-enabled student assessment tool is deployed across 50,000 rural schools in a country with significant urban-rural infrastructure inequality, multiple official languages, and no specific student data protection law. The exercise asks: which existing governance framework applies? Who is accountable when the tool produces systematically lower scores for girls in rural areas? What redress mechanism is available to affected students and families? Another effective format is the "Living Policy Lab," where specific AI governance experiments (such as a local data trust for schools) are showcased and critiqued in real-time. This encourages a "sandbox" approach to governance, where international stakeholders can learn from successful—and unsuccessful—local experiments in the Global South. This iterative process is better suited to the rapid pace of AI development than traditional, slow-moving diplomatic cycles. ________________________________________

India's DIKSHA platform is a world-leading example of "Digital Public Infrastructure" in education. By being open-source and scalable, it demonstrates how a national learning system can provide resources to millions while maintaining local control over content and data. It serves as a practical model for how countries can build "Open AI" for learning without becoming dependent on proprietary international platforms. Another critical benchmark is the UNESCO AI Competency Framework for Teachers. It provides a clear, actionable guide for what "Human Oversight" looks like in practice, moving from abstract ethics to specific professional skills. Policies that adopt such frameworks allow for "Responsible AI" integration that enhances, rather than diminishes, the vital human connection at the heart of the learning process.