Indian Institute of Technology Ropar - Technology and Innovation Foundation

A successful first Global Dialogue on AI Governance, established under United Nations General Assembly Resolution A/RES/79/325, must deliver outcomes that reflect the operational realities of building and deploying AI in resource-constrained environments, particularly across Asia and the wider Global South. From the perspective of implementing AI applications on the ground, the most critical success metric is access. Today, the ability to develop and scale AI solutions is limited not by ideas, but by constrained access to compute, high-quality datasets, and affordable cloud infrastructure. The Dialogue should therefore result in a global framework for shared AI infrastructure, including regional compute hubs and public-interest cloud access for academia, startups, and public institutions. Second, affordability and pricing transparency must be addressed. AI deployment costs—from model access to APIs and GPUs—remain prohibitive for many institutions in Asia. A meaningful outcome would be the creation of global financing and subsidisation mechanisms, especially for public-good applications in agriculture, water, healthcare, and education. Third, the Dialogue must prioritise capacity building as a core governance pillar. There is a significant gap in technical, regulatory, and deployment capabilities across countries. Success would include commitments to establish regional centres of excellence, joint research platforms, and practitioner-level training ecosystems, enabling countries not just to adopt AI, but to build and govern it. Fourth, data equity and local context must be recognised. AI systems built on non-representative datasets risk excluding large populations. The Dialogue should promote locally governed data ecosystems and language models, particularly for underserved regions and sectors. Finally, success will depend on moving from intent to implementation architecture—clear timelines, funding pathways, and accountability mechanisms. For the Global South, AI governance must not only regulate risk but also unlock capability, inclusion, and development outcomes at scale.

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

From the perspective of IIT Ropar Technology and Innovation Foundation (AWaDH) working under the NM-ICPS mission, the priorities must align with ground-level AI deployment, ecosystem building, and equitable access. Selected Priorities (Top 4): 1. AI capacity-building This is the most urgent priority. Across India and the broader Global South, the key bottleneck is not intent but capability-in terms of skilled manpower, institutional readiness, and domain-level AI integration. Our work directly involves training, skilling, and enabling practitioners, startups, and public institutions, making this central to our engagement. 2. Open-source software, open data and open AI models Given the high cost and limited access to proprietary AI systems, open ecosystems are essential for democratisation. Our focus on developing and deploying AI solutions in agriculture, water, and rural ecosystems requires accessible, adaptable, and locally deployable AI frameworks. 3. Transparency, accountability, and human oversight As AI moves into critical sectors like agriculture advisories, climate intelligence, and public systems, ensuring trust, explainability, and responsible deployment becomes essential. Our applied AI work demands clear accountability frameworks and human-in-the-loop systems. 4. Social, economic, ethical, cultural, linguistic and technical implications of AI AI systems must reflect local realities, languages, and socio-economic contexts. Our focus on regional deployment highlights the importance of inclusive datasets, local language models, and culturally aligned AI solutions. Rationale (Overall) These priorities collectively address the core challenges of the Global South: Access → through open AI Capability → through capacity building Trust → through transparency Relevance → through contextual and inclusive AI This ensures that AI governance is not only about regulation, but about enabling meaningful, scalable, and inclusive impact.

Yes while the listed themes are comprehensive, a few critical cross-cutting issues remain underrepresented, particularly from a Global South, implementation-driven perspective. First, equitable access to AI infrastructure (compute, cloud, and chips) is not explicitly captured. Today, the ability to build and deploy AI is constrained by limited access to high-performance compute and affordable cloud resources. This creates structural dependency on a few global providers and restricts innovation in developing economies. AI governance must therefore address infrastructure democratisation and digital public infrastructure for AI as a distinct priority. Second, pricing and economic accessibility of AI systems is an emerging concern. Beyond technical access, the cost of APIs, model licensing, integration, and maintenance makes AI adoption prohibitive for public institutions, startups, and MSMEs. Without addressing fair pricing frameworks and financing mechanisms, AI risks widening global inequalities. Third, data equity and sovereign data ecosystems require stronger emphasis. Many regions lack access to high-quality, representative datasets, particularly in local languages and domain-specific applications like agriculture or climate. Governance frameworks should promote locally governed, trusted data ecosystems that balance innovation with privacy and rights. Fourth, implementation capacity and public-sector readiness go beyond general capacity-building. Countries need institutional capability in AI procurement, auditing, standard-setting, and lifecycle governance, especially for deploying AI in critical public services. Finally, AI for public-good applications in underserved sectors such as agriculture, water, and climate resilience should be recognised as a distinct priority. Governance must not only mitigate risks but also actively enable development-oriented deployment at scale. Together, these issues highlight that AI governance must evolve from a primarily regulatory lens to one that equally prioritises access, affordability, and implementation capacity.

From the perspective of India and similar Global South contexts, the governance gaps across AI capacity-building, open ecosystems, transparency, and contextual relevance are already shaping both constraints and opportunities in real deployments. The most significant challenge is capacity asymmetry. While demand for AI applications in sectors like agriculture, water, and climate resilience is high, there is a shortage of trained professionals who can design, deploy, audit, and maintain such systems. This creates dependence on external solutions and limits the ability of institutions to govern AI effectively. A second major gap is restricted access to high-quality datasets and affordable compute infrastructure. Many AI models are trained on non-representative data, leading to reduced accuracy and reliability in local contexts. At the same time, high costs of cloud services and proprietary models constrain experimentation and scale-up, particularly for startups and public-sector institutions. Third, limited transparency in AI systems especially black-box models poses challenges in critical applications. In sectors like agriculture advisories or climate forecasting, lack of explainability affects trust, adoption, and accountability, particularly when decisions directly impact livelihoods. However, these challenges also present strong opportunities. There is a growing push toward open-source AI, local language models, and domain-specific datasets, enabling more inclusive and context-aware solutions. India's digital public infrastructure approach provides a strong foundation to extend AI as a public good. Additionally, the need for capacity-building is driving the creation of innovation hubs, skilling ecosystems, and academia–industry collaborations, which can position the region as a leader in applied AI for development sectors. Overall, bridging these governance gaps can enable the Global South not only to adopt AI, but to shape scalable, inclusive, and globally relevant AI solutions.

The Global AI Dialogue can play a catalytic role by shifting international cooperation from fragmented discussions to coordinated, outcome-driven action, particularly for the Global South. First, it can serve as a platform to align priorities across countries not only on safety and ethics, but also on access, affordability, and development use cases. By bringing together governments, academia, industry, and multilateral institutions, the Dialogue can help create a shared understanding of AI as both a risk and an opportunity, especially for emerging economies. Second, the Dialogue can enable cooperation on shared infrastructure. Many countries lack access to compute, datasets, and advanced models. The Dialogue can facilitate partnerships for regional compute hubs, open data ecosystems, and public-interest AI platforms, reducing duplication and improving collective capability. Third, it can strengthen capacity-building collaborations. Through structured programmes, fellowships, and joint research initiatives, countries can exchange knowledge on AI governance, regulatory frameworks, and deployment strategies. This is critical for ensuring that countries are not just adopters, but active participants in shaping AI systems. Fourth, the Dialogue can support interoperability of governance approaches. As different regions develop their own regulatory models, there is a risk of fragmentation. The Dialogue can help identify common principles, standards, and best practices, enabling smoother cross-border collaboration and innovation. Finally, its most important role is to ensure inclusive participation in global rule-making. The Dialogue must amplify the voices of the Global South, ensuring that governance frameworks reflect diverse socio-economic realities and do not disproportionately favour technologically advanced economies. In essence, the AI Dialogue can become a bridge linking policy, technology, and development priorities and enabling a more balanced, cooperative, and inclusive global AI governance ecosystem.

The Global AI Dialogue should build upon and connect with a growing ecosystem of international and national initiatives that are already shaping AI governance, while adding coherence, inclusivity, and implementation focus. At the global level, initiatives such as those led by UNESCO (AI Ethics Recommendation), the OECD (AI Principles), and platforms like Center for AI and Digital Policy (AI Policy Clinics and global governance inputs) have laid strong normative and policy foundations. Similarly, organizations such as the Responsible AI Foundation are working on operationalizing responsible AI through tools, standards, and enterprise-level frameworks. At the national level, India's IndiaAI Mission and MANAV Principles signed by countries during India AI Summit held in February represents a comprehensive approach to building AI ecosystems through investments in compute infrastructure, datasets, skilling, and innovation. Such mission-mode programmes are critical examples of how countries in the Global South are translating policy into implementation. However, these efforts often operate in silos. The Global AI Dialogue can add value by acting as a convergence platform bringing together these diverse initiatives to enable: Interoperability of frameworks and standards, reducing fragmentation Cross-learning between policy and practice, especially from implementation-driven programmes like IndiaAI Scaling of best practices, particularly in capacity building, open ecosystems, and public-sector AI deployment Importantly, the Dialogue can bridge the gap between norm-setting and execution by creating mechanisms for: - joint pilot programmes - shared infrastructure initiatives - coordinated funding and capacity-building efforts For the Global South, this added value is critical. The Dialogue can ensure that existing initiatives are not duplicated, but aligned and amplified, while enabling more equitable participation in shaping global AI governance.

For the Global AI Dialogue to be effective, it must move beyond a traditional conference format and adopt a multi-stakeholder, action-oriented structure where each stakeholder group has a clearly defined role. Governments should lead on policy alignment, regulatory frameworks, and international cooperation mechanisms, while also committing to pilot implementations in priority sectors. Academia and research institutions such as those under mission-mode programmes like NM-ICPS, India AI Mission, National Quantum Mission should contribute through evidence-based research, model validation, and capacity-building programmes. Industry should provide technology insights, scalable solutions, and responsible deployment practices, while also supporting open innovation and fair access models. Civil society and global policy organizations (e.g., Center for AI and Digital Policy) can play a critical role in ensuring accountability, human rights alignment, and inclusive participation. International organizations such as UNESCO and OECD can anchor the Dialogue in global norms, standards, and cross-country coordination. Recommended Format and Structure 1. Thematic Working Tracks Organize the Dialogue into focused tracks (e.g., infrastructure access, capacity building, governance frameworks, public-sector AI), each co-led by a mix of stakeholders. 2. Regional Consultations (Pre-Dialogue) Conduct structured consultations across regions—especially the Global South—to ensure that outcomes reflect diverse realities. 3. Implementation Labs / Pilot Showcases Move beyond policy discussions by including live case studies and pilots (e.g., AI in agriculture, healthcare, climate), enabling practical learning. 4.Outcome-Oriented Commitments - Each track should deliver: - actionable recommendations - partnership announcements - funding or capacity-building commitments 5.Follow-up Mechanism Establish a permanent coordination platform or secretariat to track progress, share best practices, and sustain engagement beyond the Dialogue. In essence, the Dialogue should function not just as a forum, but as a collaborative execution platform linking policy, technology, and development outcomes.

Global AI governance discussions remain disproportionately shaped by advanced economies, large technology firms, and policy institutions, leaving several critical voices underrepresented. First, grassroots and last-mile users farmers, MSMEs, informal workers, and frontline public service providers are rarely part of governance conversations. Yet, they are among the largest users and most impacted by AI in sectors like agriculture, healthcare, and livelihoods. Their inclusion is essential to ensure that AI systems are practical, accessible, and contextually relevant. Second, regional innovators and startups from the Global South are underrepresented. While they are actively building solutions for local challenges, they often lack visibility and access to global platforms. Including them would bring implementation-driven insights into governance discussions. Third, local language communities and culturally diverse populations are often overlooked. AI systems that do not account for linguistic and cultural diversity risk exclusion and bias at scale. Their perspectives are critical for building inclusive and representative AI systems. Fourth, public sector practitioners and regulators from developing countries need stronger representation. Many countries face capacity constraints in AI procurement, auditing, and regulation, yet their practical governance challenges are not sufficiently reflected in global frameworks. Fifth, civil society organizations working at the intersection of technology and social impact, including groups focused on digital rights, gender inclusion, and community data governance, require greater voice to ensure equity and accountability. To address these gaps, the AI Dialogue should: - institutionalize regional and sectoral consultations, - create dedicated representation quotas or tracks for underrepresented groups, - support participation through funding and capacity-building, and - enable bottom-up inputs through pilot projects and field-level case studies. A truly inclusive AI governance framework must reflect not only those who build AI, but also those who live with its consequences.

To move beyond static panel discussions, the AI Dialogue should adopt interactive, outcome-driven formats that connect policy with real-world implementation. 1. Policy-to-Prototype Labs Multi-stakeholder teams (government, academia, startups, civil society) work together on a specific challenge such as AI in agriculture or healthcare and co-develop policy-backed solution prototypes within the Dialogue. This ensures immediate translation of ideas into actionable models. 2. Live Use-Case Demonstrations Curated showcases of field-tested AI applications especially from the Global South allow participants to engage with real deployment challenges, including data gaps, cost constraints, and user adoption. This grounds discussions in practical realities. 3. Scenario Simulation Exercises Interactive simulations where participants respond to scenarios such as AI system failure, bias in decision-making, or cross-border data conflicts. This helps policymakers and practitioners stress-test governance frameworks and decision-making processes. 4. Regional Roundtables with Reverse Dialogue Instead of top-down discussions, practitioners and community representatives present bottom-up insights, followed by responses from policymakers and global institutions. This ensures that lived experiences inform governance design. 5. Co-Creation Sprints on Standards and Frameworks Short, focused sessions where stakeholders collaboratively draft guidelines, principles, or model policies, which can be refined and adopted post-Dialogue. 6. Commitment and Partnership Marketplace A structured platform where organizations announce collaborations, funding commitments, and pilot projects, enabling direct matchmaking between solution providers and adopters. 7. Continuous Digital Engagement Platform An online platform running alongside and beyond the Dialogue to enable ongoing discussions, knowledge sharing, and tracking of commitments. These formats can transform the AI Dialogue into a living, participatory ecosystem, ensuring that engagement is not only dynamic, but also leads to measurable outcomes and sustained collaboration.

India offers several strong examples of policy, platforms, and mission-mode approaches that advance effective and inclusive AI governance. A key example is the IndiaAI Mission, which takes a comprehensive ecosystem approach focusing on compute infrastructure, datasets, skilling, startups, and responsible AI. By investing in shared national AI resources and enabling access for academia and startups, it directly addresses challenges of democratisation and affordability. India's broader Digital Public Infrastructure (DPI) including platforms like Aadhaar, UPI, and DigiLocker provides a replicable model for AI governance. These platforms demonstrate how interoperable, scalable, and inclusive digital systems can be built with strong governance frameworks, enabling AI-ready ecosystems in finance, identity, and service delivery. The Bhashini initiative is another important example, promoting AI for local languages. It addresses linguistic inclusion by enabling speech and language technologies across Indian languages, ensuring that AI systems are accessible and representative. In the research and innovation domain, initiatives under the NM-ICPS mission, such as IIT Ropar's AWaDH, demonstrate applied AI governance through testbeds, real-world deployments, and domain-specific solutions in agriculture and water. These hubs integrate technology development with policy, standards, and capacity building, ensuring responsible deployment. Additionally, India's approach to open networks, such as ONDC, reflects a shift toward open, interoperable ecosystems that reduce platform monopolies and enable fair participation an important principle for AI governance as well. Together, these examples highlight a distinct Indian approach: combining public digital infrastructure, open ecosystems, mission-mode innovation, and inclusive design to address both governance risks and development opportunities in AI.