Cheikh Anta Diop University Dakar-Senegal

A successful Global Dialogue on AI Governance would move beyond high-level principles and produce operational clarity that bridges global ambition with local realities. First, success would mean establishing a shared baseline of understanding across countries with very different levels of technological maturity. Today, AI governance discussions are often dominated by highly digitized economies, while many regions are still addressing fundamental gaps in infrastructure, data systems, and digital literacy. A meaningful dialogue must acknowledge these asymmetries. Second, it should deliver actionable pathways for implementation, not just frameworks. This includes: - context-sensitive governance models adaptable to low-resource environments - practical guidance for integrating AI into sectors such as agriculture, health, education, and environmental management - mechanisms to support countries in transitioning from AI consumers to AI contributors Third, success would require concrete commitments on capacity-building and knowledge transfer. Without this, governance risks becoming exclusionary, reinforcing existing global inequalities. Fourth, the dialogue should create interfaces between policy, academia, and field realities. Many governance models fail not because they are conceptually weak, but because they are disconnected from how systems function in practice. Finally, a successful outcome would embed accountability and long-term coordination mechanisms, ensuring that discussions translate into measurable progress. Ultimately, success is not defined by the strength of the declarations produced, but by the extent to which they are understood, adapted, and implemented across diverse contexts.

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

These priorities reflect the need to address AI governance not only as a technological issue, but as a systemic transformation affecting societies at multiple levels. AI capacity-building is foundational. Without local technical, institutional, and educational capacity, many countries remain passive users of externally developed systems. This limits both sovereignty and the ability to adapt AI to local needs. The social, economic, and cultural implications of AI are particularly critical in contexts where informal economies, diverse knowledge systems, and cultural practices shape daily life. AI systems designed without these considerations risk being ineffective or even disruptive. Transparency, accountability, and human oversight are essential to build trust. In many regions, limited regulatory enforcement capacity already challenges governance in other sectors. AI systems that operate as "black boxes" can exacerbate these challenges and weaken institutional credibility. Finally, protection and promotion of human rights must remain central. In contexts with existing inequalities, AI can unintentionally reinforce exclusion, for example through biased data, unequal access, or misaligned deployment priorities. Together, these priorities aim to ensure that AI governance is inclusive, context-aware, and implementable, rather than purely normative.

One critical cross-cutting issue is the growing disconnect between AI system design and local realities. Many AI tools are developed in environments with robust data ecosystems, infrastructure, and institutional capacity. When transferred to contexts where these conditions do not exist, they often require significant adaptation, or fail to deliver intended outcomes. This creates inefficiencies and limits impact. Another emerging issue is the risk of structural dependency. As countries adopt externally developed AI systems without building internal capacity, they may become increasingly dependent on external technologies, data, and expertise. This raises questions of digital sovereignty and long-term resilience. The informal economy also remains largely overlooked in AI governance discussions. In many regions, a significant share of economic activity occurs outside formal systems, yet AI deployment strategies rarely account for these dynamics. Additionally, infrastructure inequality including access to electricity, connectivity, and computing resources continues to shape who benefits from AI and who is excluded. Finally, there is a need to better integrate environmental considerations into AI governance. The energy consumption of AI systems, data infrastructure, and digital expansion must be aligned with sustainability goals, particularly in regions already facing resource constraints. Addressing these issues requires moving from a purely technological lens to a systems-based approach, where governance frameworks are co-developed with those who will ultimately use and be affected by AI.

In my context (West Africa), AI governance gaps are not only regulatory, they are systemic and deeply tied to structural inequalities. One of the most significant challenges is the asymmetry between AI adoption and institutional readiness. AI tools are increasingly accessible (often through global platforms), yet: - regulatory frameworks remain fragmented or non-operational, - data governance infrastructures are weak or absent, - and public institutions often lack the technical capacity to assess or supervise AI deployment. This creates a "use without oversight" reality, where AI is consumed faster than it is understood or governed. A second critical challenge lies in capacity gaps across the ecosystem. While a small segment of the population engages with AI at an advanced level, the majority remains: - under-equipped (limited access to devices and stable internet), - under-trained (education systems not aligned with AI realities), - and under-protected (limited awareness of risks such as bias, misinformation, or data exploitation). This widens existing socio-economic inequalities. However, these gaps also create significant opportunities. First, there is an opportunity to build governance frameworks from the ground up, integrating: - local realities, - informal economies, - and indigenous knowledge systems often overlooked in global AI discussions. Second, AI can serve as a leapfrogging tool in sectors such as: - agriculture (predictive climate tools), - environmental monitoring, - and access to decentralized services. Finally, there is a strategic opportunity to redefine capacity-building, not only as technical training, but as: - critical thinking, - contextual application, - and responsible use. Ultimately, the key issue is not access to AI alone, but alignment between technology, systems, and society.

The Global Dialogue on AI Governance can play a critical bridging role between global standards and local realities. Today, most AI governance frameworks are designed at a global or regional level, yet their implementation challenges are deeply local. This creates a structural gap between principles and practice, particularly in regions where infrastructure, institutional capacity, and socio-economic conditions differ significantly from those in which these frameworks are developed. The Dialogue can help address this gap in three key ways: 1. Translating global principles into operational realities It can facilitate the adaptation of high-level governance principles into context-sensitive implementation pathways, especially for low- and middle-income countries. 2. Elevating underrepresented perspectives Many regions remain underrepresented in AI governance discussions, despite being significantly impacted by AI systems. The Dialogue can amplify voices from the Global South, ensuring governance approaches are inclusive and grounded in diverse realities. 3. Aligning governance with development priorities AI governance should not be disconnected from broader development goals. The Dialogue can help align AI strategies with sectors such as education, agriculture, health, and environmental management, where the stakes are both immediate and structural. Ultimately, the Dialogue can move international cooperation from norm-setting alone to implementation-oriented collaboration, where knowledge exchange, capacity-building, and co-designed solutions become central.

The AI Dialogue should build upon existing global and regional initiatives while addressing their current limitations in implementation and inclusivity. Relevant initiatives include: - UNESCO Recommendation on the Ethics of AI - OECD AI Principles - Global Partnership on AI - Regional digital transformation and AI strategies across Africa and other emerging regions These frameworks provide strong normative foundations. However, several gaps persist: - Limited operationalization in diverse local contexts - Uneven participation from developing regions - Weak alignment with sector-specific realities (education, labor markets, environment) The added value of the AI Dialogue lies in its ability to: 1. Connect fragmented efforts Rather than creating new frameworks, it can act as a coordination platform linking existing initiatives, reducing duplication and increasing coherence. 2. Focus on implementation ecosystems By prioritizing how governance is applied in real-world contexts, the Dialogue can support practical tools, pilot initiatives, and locally adapted governance models. 3. Strengthen South - South and triangular cooperation Many countries face similar structural challenges. The Dialogue can facilitate knowledge exchange between them, rather than relying solely on North - South knowledge flows. 4. Integrate cross-sectoral perspectives AI governance should not remain siloed. Linking it to environmental systems, labor dynamics, and education reforms can improve its relevance and impact. In this sense, the Dialogue's value is not only in shaping principles, but in enabling context-aware, actionable governance pathways.

Meaningful AI governance requires moving beyond symbolic participation toward structured, role-based contributions from all stakeholders. - Governments should provide regulatory direction, align AI with national development priorities, and ensure policy coherence across sectors (education, labor, environment). - Private sector actors must contribute technical expertise, share data responsibly, and commit to transparency in AI systems design and deployment. - Academia and researchers should bridge theory and practice by documenting real-world impacts, especially in underrepresented regions. - Civil society organizations play a critical role in highlighting social risks, ethical concerns, and community-level implications. - Local communities and practitioners must be engaged not as beneficiaries, but as co-designers, bringing lived experience into governance frameworks. For effectiveness, the AI Dialogue should be structured around: Multi-level engagement: - global principles, - regional contextualization and - local validation Thematic working groups tied to real use cases (e.g. AI in agriculture, health, education) Feedback loops ensuring that insights from the ground inform global decision-makin. Without this structure, participation risks remaining performative rather than transformative.

Several critical voices remain underrepresented in global AI governance discussions: - Informal sector workers (who form a large part of African economies) - Rural and low-connectivity communities - Educators and trainers operating in resource-constrained environments - Local technologists and small-scale innovators - Women and youth in non-urban contexts These groups are not only underrepresented, they are often most affected by AI disruptions. Inclusion requires more than invitations: 1. Contextual access mechanisms Participation formats must account for language barriers, connectivity limitations, and time constraints. 2. Intermediated representation Local institutions (universities, NGOs, community leaders) can act as bridges to structure and convey grassroots perspectives. 3. Compensated participation Engagement should recognize time and expertise, especially for non-institutional actors. 4. Decentralized consultations Regional and national dialogues should precede global ones, ensuring that contributions are grounded in reality. True inclusion is not about diversity optics, it is about improving the quality and relevance of governance decisions.

To move from discussion to impact, the AI Dialogue should adopt hybrid, practice-oriented engagement formats: 1. Scenario based simulations Stakeholders work through real-life cases (e.g., AI in public services, agriculture, education) to identify risks, trade-offs, and governance needs. 2. "Policy to practice labs" Small, multidisciplinary groups test how global principles translate into local implementation contexts. 3. Reverse panels Instead of experts speaking to communities, local practitioners present challenges, and policymakers respond. 4. Evidence clinics Researchers and practitioners present field data, followed by structured feedback from policymakers and industry. 5. Micro & macro dialogue sessions Dedicated spaces where local realities are explicitly connected to global frameworks, addressing the common disconnect. 6. Digital & offline hybrid participation Combining virtual engagement with locally hosted discussion hubs to ensure broader inclusion. The objective is to shift from static dialogue to iterative, grounded, and action-oriented exchange. Because effective AI governance will not emerge from consensus alone, but from tested, context-aware solutions.

Effective AI governance requires approaches that go beyond high-level principles and actively bridge the gap between policy design and real-world implementation. One emerging good practice is contextualized governance frameworks. Rather than adopting global models as-is, some countries and institutions are adapting AI guidelines to local realities; taking into account infrastructure constraints, data availability, and socio-economic priorities. This reduces the risk of "policy misalignment," where well-designed frameworks fail in practice. Another promising approach is multi-stakeholder co-design. Initiatives that involve academia, civil society, local communities, and the private sector from the outset tend to produce more inclusive and applicable governance models. This is particularly important in regions where AI systems may affect informal economies or vulnerable populations that are often overlooked in formal policy processes. Capacity-building ecosystems also stand out as critical. Programs that combine technical training with ethical, regulatory, and application-oriented learning help shift stakeholders from passive users of AI to informed actors capable of shaping its deployment. In addition, use-case-driven governance is gaining relevance. Instead of regulating AI in the abstract, some frameworks focus on high-impact sectors such as health, agriculture, and environmental management. This allows for more targeted, measurable, and adaptable policies. Finally, transparency and accountability mechanisms such as algorithmic audits, impact assessments, and open reporting are essential to build trust and ensure responsible use, especially in low-resource settings where oversight capacities may be limited. Overall, the most effective approaches are those that integrate technical robustness with social relevance, ensuring that AI governance is not only compliant, but also equitable, practical, and locally meaningful.