1. What did the collaboration of both academic and policy stakeholders at the Just AI Conference reveal about the strengths/shortfalls of multidisciplinary dialogue, and the collective framework-shaping process? And what procedural takeaways could be useful for the Global AI Dialogue, especially for incorporating multistakeholder voices and insights?
The conference demonstrated that multidisciplinary dialogue is most valuable when it moves beyond consultation towards genuine co-production of governance frameworks. The Just AI Framework of Inquiry (FoI) came to fruition from sustained engagement between researchers, policymakers, civil society, and technical experts, revealing that AI governance is not simply a technical or regulatory exercise, but one that must integrate economic, legal, social, and political perspectives.
In terms of the Global AI Governance Dialogue, we recognised that participation alone does not guarantee influence. There is a risk that multistakeholder engagement becomes symbolic unless it is accompanied by transparent decision-making and clear mechanisms that show how diverse inputs influence outcomes. This mirrors a broader challenge in AI governance, where inclusion is often measured by representation rather than by agenda-setting power. Therefore, the key procedural lesson for the Global AI Dialogue is that multistakeholder participation must be institutionalised throughout the policy cycle, from agenda setting and framework development to implementation and evaluation. Equally important are transparent processes for documenting how stakeholder inputs shape decisions and mechanisms for monitoring commitments over time. Such an approach would strengthen both the legitimacy and accountability of global AI governance while ensuring that Global Majority perspectives inform not only the conversation, but also its outcomes.
2. Which academic methodologies were presented at the conference, whose approaches you felt were well-suited to support contextually relevant African AI policy research?
The conference presentations brought diverse, useful and eye-opening methodologies that moved from purely technical assessments of AI, toward approaches that are participatory, interdisciplinary, historically grounded, and attentive to local contexts and institutions. One common methodological theme across the conference was the use of participatory and co-designed approaches that place affected communities at the centre of knowledge production. For example, one paper employed participatory engagement with teachers, students, farmers and entrepreneurs to understand how contextual knowledge and relational values influence technology adoption. Another paper demonstrated how community-validated mapping can correct biases in AI-generated geospatial datasets. These approaches are particularly relevant in African contexts where informal institutions, local knowledge systems, and diverse socio-economic realities are often poorly reflected in global datasets and imported AI models.
Another methodology that stood out was the use of institutional case studies to generate governance insights from existing African institutions. Rather than treating AI governance as an entirely new problem, this paper examined how community-based savings associations manage trust, accountability, collective decision-making, and uncertainty in practice. It then translated these governance principles into lessons for AI governance, which is valuable for African AI policy research because it builds from institutions and governance practices that already enjoy local legitimacy, rather than relying solely on imported governance models.
3. What institutional and cross-department challenges were presented in the policy-focused portion of the conference programme, which impact the efficacy of AI governance enforcement in Africa?
One of the insightful policy discussions was that the challenge of AI governance in Africa is increasingly institutional rather than technological. For example, the recurring concern was the growing tendency to position data protection authorities as de facto AI regulators. While these institutions play a critical role, their mandates are primarily focused on personal data, leaving important issues such as non-personal data, model behaviour, competition, labour impacts, consumer protection, and broader societal harms insufficiently addressed. This creates a risk of overburdening already constrained regulators, while simultaneously generating regulatory gaps and overlapping responsibilities across institutions.
Participants highlighted the fragmentation of governance structures. AI cuts across multiple ministries and regulators, yet institutional arrangements often remain siloed, making coordination difficult and weakening accountability. Questions of mandate clarity, therefore, become central to effective enforcement. In addition, policymakers must navigate competing objectives between innovation, economic growth, rights protection, and social justice. A particularly important insight was that building local innovation ecosystems and domestic AI capabilities should itself be viewed as a governance objective and a form of justice, ensuring that African countries participate not only as regulators of AI, but also as creators and beneficiaries of its value.
4. What role can academia and research organisations play in furthering AI governance efforts not only on the continent but within global AI processes, which are often hindered by limited capacity in Research and Innovation, Talent and Skills?
Academia and research organisations have a critical role to play in ensuring that AI governance is informed by evidence, responsive to local realities, and inclusive of voices that are often underrepresented in global debates. Reflections on integrating academia and research organisations into African and global AI processes show notable achievements in recent years, including the expansion of stakeholder participation and the increased visibility of Global Majority perspectives in global AI discussions, for example, within the India AI Impact Summit 2026. This helped shift the agenda toward issues such as AI democratisation and development priorities that are often overlooked in global forums. However, participation alone is insufficient. Despite broader engagement, civil society and academia still had limited influence over agenda-setting and decision-making processes. This points to an important role for research institutions, not only generating evidence, but also strengthening the accountability and legitimacy of global AI governance.
6. Which thematic and research areas that were brought forward in the conference but excluded from the AI dialogue’s thematic clusters should still be spotlighted in surrounding discussions?
While the four thematic clusters provide a strong foundation for global AI governance, the conference highlighted more areas that warrant greater visibility in the surrounding discussions. Most notably:
The political economy of AI received sustained attention but is not explicitly reflected within the Dialogue’s current architecture. Discussions repeatedly emphasised that AI governance should extend beyond risk management to address questions of value creation, market concentration, industrial development, labour transitions, and the equitable distribution of AI’s economic benefits. For many African countries, governance is inseparable from development; building domestic innovation ecosystems, digital infrastructure, and local research capacity is itself a governance objective.
The importance of institutional governance. While the Dialogue addresses regulatory interoperability, it places greater emphasis on the practical realities of implementation, including institutional mandates, regulatory coordination, public-sector capability, and mechanisms for monitoring commitments over time. Effective governance depends not only on good principles but on institutions capable of enforcing them.
Stronger emphasis on epistemic justice. Several papers demonstrated how African knowledge systems, community institutions, and participatory governance approaches can generate governance innovations rather than merely provide local case studies. Global Majority perspectives should not simply inform AI governance; they should actively shape its conceptual foundations.
