Opinions expressed by invited speakers or program participants do not necessarily reflect the opinion of the U.S. government, its affiliates, or the Atlantic Council GeoTech Center.

The sixteenth AI Connect Webinar, hosted by the U.S. Department of State and the Atlantic Council’s GeoTech Center on August 31, 2023, focused on artificial intelligence (AI) technical developments and AI policy advancements in Africa.

The first panel, moderated by GeoTech Center Program Assistant Emily Sespico, highlighted use cases and policy perspectives on AI, open data, emerging technologies in Africa. Chief Technology Officer of data354 Fabrice Zapfack presented his work on open data and data privacy in Cote d’Ivoire, where his data consulting company established an open data portal. This data portal provides hundreds of datasets with the goal of accessibility for everyone in Cote d’Ivoire, regardless of the language they speak or their ability to type on a computer. To accomplish this, Zapfack has included large language models (LLMs) on the website, supporting chat functions where users can ask for AI-generated data visualizations. Zapfack hopes to include speak-to-chat functions to further increase accessibility of data. Throughout his panel, Zapfack underscored the importance of data accessibility and data privacy, as well as the importance of LLM accuracy.

In another use case, Carnegie Mellon University Africa student Muhammad Danso presented his research findings on deep learning and robotics in AI. Danso hopes to discover how to make robots “pay attention” to stimuli in their environments. To do this, Danso helped create eyeglasses that track human eye movement and apply this “focus” to robots. Danso also presented preliminary findings from another project on memory and cognition. In this project, Danso is exploring memory systems and cognitive architecture that will hopefully enable AI to solve novel problems, problems that the models have not been trained to answer.

In the first panel’s final discussion, digital consultant Nashilongo Gervasius discussed some of the policy, infrastructure, and financial challenges that have contributed to Africa’s slow uptake of AI. Gervasius highlighted the concern that only a few countries in Africa have made great strides in the development of AI, particularly Nigeria and South Africa, and most AI companies working in Africa were founded abroad. As such, Gervasius raised the concern that AI from Western companies may not be suited to solve unique problems that affect Africa and can perpetuate economic colonialism. Gervasius recommended that AI building and development includes active participants from Africa.

The webinar’s second panel, moderated by GeoTech Center Resident Fellow Trisha Ray, investigated guardrails to ensure the effective rollout of AI in Africa. Participants included Chief Counsel of Africa-Canada Artificial Intelligence and Data Innovation Consortium (ACADIC) Jake Okechukwu Effoduh, Director of IBM Research Africa Charity Wayua, and Advisor for Data Policy and Regulation at the African Union, Abraham Kuuku Sam. Effoduh discussed the lack of government AI readiness across Africa because of low investment and lack of data accessibility. Despite this, Effoduh expressed optimism for the rapid uptake and development of Africa-based AI for Africa-specific problems. Effoduh highlighted the importance of an Africa-wide cohesive effort to the development and regulation of emerging AI technologies. Building from this, Wayua discussed the importance of scaling and the use AI in healthcare. Wayua urged further investment in the private sector development of AI, and Wayua emphasized the importance of recognizing each countries’ context and comparative advantage in regulating AI. The last panelist, Abraham Kuuku Sam, discussed some of the challenges and opportunities that come with continent-wide African AI regulatory frameworks, particularly the lack of data infrastructure in Africa. Final comments from the panel focused on the importance of recognizing when AI is a necessary tool and the importance of language-inclusive AI.