Insights from AfricAI Conference 2023: Driving the discourse on ethical AI in Africa

“Advancing Responsible and Open AI Ecosystems in Africa”

Against the backdrop of Rwanda, theLand of a Thousand Hills, the AfricAI Conference 2023 “Advancing responsible and open AI ecosystems in Africa” provided an invaluable forum for knowledge exchange, learning and networking. This gathering of key artificial intelligence (AI) stakeholders, including AI4D’s Africa program and GIZ’s FAIRForward initiative, consortia partners, African AI researchers and grantees, AI practitioners, and policymakers, strengthened understanding of the potential and challenges related to responsible and open AI on the continent.

The conference acknowledged global shifts in digital transformation driven by AI, which create opportunities to contribute to achieving the global sustainable development goals through innovations that address barriers to human development. However, the use of AI comes with risks, including reinforcement of inequalities and bias, that need to be managed and mitigated. As the adoption of AI technology across Africa gains momentum, the vibrant ecosystems of entrepreneurs, researchers, and policymakers converged to channel the benefits of AI technology to citizens in the fields of agriculture, education, energy, health, finance and beyond.

Eight emerging AI researchers supported with funding from Meta, including myself, conducted a panel discussion in which we presented short papers based on our proposed research areas and interacted with other participants on issues around AI ethics, human rights, and policy/governance in African research.

My presentation, Ethics by design:AI governance imperatives within the African context, noted that AI has excellent prospects for improving the human condition. Although AI shows potential to leapfrog some of Africa’s most pervasive socioeconomic problems, such as improving access to quality healthcare and education, this technological evolution tests ethical boundaries. Existing risks to privacy and data protection, relationality, fairness as well as discriminatory effects are heightened in this era of perceived data colonialism and surveillance capitalism[1]. SeveralAfrican governments (including South Africa, Kenya, Uganda, Ghana, and Tunisia)have begun to incorporate ethical principles into their AI governance strategies, but tensions still arise when implementing these principles in indigenous African contexts[2]. A critical knowledge gap within the AI development domain is the absence of an empirically established method to translate ethical principles into practice[3]. The conference drew attention to the need for interdisciplinary research to explore existing approaches and new mechanisms for resolving ethical tensions within the African context to minimise impact on policy and practice in the design and use of AI systems.

Other sessions accentuated and complemented ideas around ethical AI in Africa. One that stood out wasDeveloping ethical AI policies in context: Co-designing an AI policy playbook.The session highlighted that developing AI strategies and regulatory frameworks in Africa requires careful consideration to ensure policy approaches are appropriately tailored to national and regional priorities and realities.GIZ-backed FAIR Forward is developing one such framework and an AI policy playbook based on countries’ experiences in crafting AI policies across Africa and Asia. The playbook enhances the international community’s understanding of how to ensure local priorities effectively drive AI policy development.

The session discussing Commercialisation of AI solutions pathway presented AI commercialisation initiatives supported by funding such as the AI4D and GIZ open AI initiatives.Although these initiatives already have had an impact on the sector, few use cases are available. This might be due to most funding going into academic research institutes and little into the private sector; hence, scaled-up AI commercialisation is required across the continent for a more significant socioeconomic impact.

The final session I’ll highlight is Building AI capacities in Africa. AI capacity building is essential forAfrica as the continent can benefit significantly from adopting and implementing AI technologies. Although some progress has been made, challenges including inadequate infrastructure and datasets and a shortage of skilled professionals persist and must be addressed. Several AI capacity-building initiatives and programs have been developed, including:

·      community building to share knowledge and expertise

·      AI training to equip students with the necessary skills to work inAI-related fields

·      innovation to support startups and entrepreneurs in the developmentof AI solutions

·      partnership and collaboration to share infrastructure and datasets

The session indicated that building AI capacity in Africa requires a long-term and sustained effort and a multifaceted approach from multiple stakeholders, including governments, academia, the private sector, and civil society. By adopting these approaches and strategies, Africa can build sustained AI capacity and realise the full potential of AI technologies in transforming various economic sectors, from healthcare to agriculture to financial services to transportation.

Overall, the conference enabled participants from Africa’s AI communities and ecosystems to share knowledge and best practices from diverse contexts and perspectives. It provided a platform for exchange between the research community and the public and private sectors to build responsible and open AI ecosystems through new alliances and opportunities for future collaboration.

[1] Artificial Intelligence for Africa: AnOpportunity for Growth, Development, and Democratisation. University ofPretoria, South Africa.;Mhlambi, Sabelo. 2020. From Rationality to Relationality: Ubuntu as an Ethical and Human Rights Framework for Artificial Intelligence Governance. Carr Center for Human Rights Policy, Harvard Kennedy School, Harvard University.

[2] Whittlestone, J., Nyrup, R., Alexandrova, A.and Cave, S., 2019, January. The role and limits of principles in AI ethics:towards a focus on tensions. In Proceedings of the 2019 AAAI/ACM Conference onAI, Ethics, and Society (pp. 195-200).

[3] Mittelstadt, B., 2019. Principles alonecannot guarantee ethical AI. Nature Machine Intelligence, pp.1-7.