Day 1 closed with a fireside chat that explored how AI is driving scientific breakthroughs and expanding opportunities in Africa. A key focus was Google DeepMind’s development of Alpha Fold, which was tasked to solve a decades-long “protein folding problem.” By predicting the 3D structures of proteins from amino acid sequences, AlphaFold has transformed research in drug discovery and agricultural resilience. Its open-access database of over 200 million protein structures has accelerated global research, including work on antimicrobial resistance and malaria, two challenges central to Africa’s health landscape.
The conversation also addressed how Google DeepMind is working to increase AI’s positive social impact through its Impact Accelerator. This initiative emphasizes two core areas: talent development and technological accessibility. Examples by which AlphaFold is brought to Africa include a Master’s program in AI for Science at the African Institute for Mathematical Sciences and supporting capacity-building workshops through BioStruct-Africa. DeepMind has also developed models like Gemma, which operate on standard laptops, enabling AI research in low-resource environments.
The session highlighted further tools such as MedLM, Google’s large language model trained on medical data, and CoDoC, which evaluates AI’s accuracy in diagnosing medical conditions. These technologies augment human expertise and can support healthcare systems and scale service provision, especially in under-resourced regions. There are numerous examples of how AI can address Africa’s pressing healthcare challenges, from AI-driven research on bacterial biofilms to new antimalarial drug targets.
There are also lessons-learned for African researchers seeking to use AI. First, researchers need to be clear on the problem they are seeking to solve. Second, solving a problem with AI requires vast amounts of data, including data on the desired outcomes. A major reason why AlphaFold succeeded is that it could use a database of 170,000 known protein structures in the Protein Data Bank. This existing data also enabled researchers to validate AI’s performance.
African research institutions can prepare to take advantage of AI-driven scientific advancements through a number of ways, such as building multidisciplinary teams, defining bold, impact-driven research goals, and promoting open collaboration on AI applications. Securing vast datasets of high-quality local data is critical for success.