However, we move beyond the official framework to discuss critical concerns around algorithmic sovereignty. Drawing on insights from Professor Benjamin Rosman, we examine why it is essential for South Africa to build its own “algorithmic refineries” to avoid becoming a perpetual exporter of raw data and a dependent importer of foreign-owned insights.

We are joined (via their research) by fellows from the African Institute for Data Science and AI (AfriDSAI) to discuss the challenges of decoloniality and AI Justice. Prof. Emma Ruttkamp-Bloem unpacks how a relational ethic can protect communities from the extractive practices of global Big Tech, while Prof. Chijioke Okorie introduces the Nwulite Obodo Open Data Licence (NOODL)—an innovative South African solution designed to ensure local datasets benefit the communities that created them.

We also highlight the explosive growth of local academic networks, such as the African Institute for Data Science and AI (AfriDSAI) at the University of Pretoria, the MIND Institute at Wits and the Data Science Law Lab, which are serving as the “intellectual infrastructure” for sovereign research capacity. We conclude the episode by reflecting on the core concept of Ubuntu AI—a guiding lens for technological progress that prioritises interdependence, community responsibility, and human dignity.

Key Topics Covered

• The 2026 Draft Policy — Understanding the ambitious institutional setup, including seven proposed oversight bodies
• Algorithmic Sovereignty — Moving beyond data privacy to local system design; why South Africa needs its own “algorithmic refineries”
• AfriDSAI Insights — Critiques on “bureaucratic ambition” and the need for decolonial frameworks
• Sovereign R&D — How institutes like MIND and communities like Masakhane are decolonising African NLP
• NOODL — The Nwulite Obodo Open Data Licence: a local solution for community data sovereignty
• Ubuntu AI — Framing AI ethics through a philosophy of shared responsibility and human dignity

Featured Voices & Research

• Prof. Benjamin Rosman — on algorithmic sovereignty and the “raw data exporter” problem
• Prof. Emma Ruttkamp-Bloem (AfriDSAI) — on relational ethics and protection from extractive Big Tech practices
• Prof. Chijioke Okorie (AfriDSAI) — on the NOODL open data licence
• Masakhane — community-driven decolonisation of African NLP

Key themes included:

• The transformative potential of AI in agriculture, healthcare, and education
• Challenges small businesses face in adopting AI technologies
• The need for collaboration between academia, industry, and government to enhance AI education and secure policy support
• Ethical and sustainability concerns in AI development
• The importance of tailored, inclusive AI solutions for the African context

🔗 Panel info: www.dsfsi.co.za/blog/AI-Workshop-industry-panel/

The conversation emphasised the unique challenges of building AI systems for languages that have limited computational and linguistic resources — including data scarcity, evaluation benchmarks, and the need for community involvement in dataset creation.

🔗 Panel info: www.dsfsi.co.za/blog/AI-Workshop-LLM-panel/

This episode examines how Twitter data can be used to analyse commuter sentiments, applying Multilingual Opinion Mining and NLP techniques to understand user experiences of public transport. The findings offer valuable insights for policymakers seeking to enhance urban mobility and sustainability across the region.

🔗 Dissertation page: www.dsfsi.co.za/blog/Rozina-dissertation/

The work highlights the importance of building language technology infrastructure for languages that are underrepresented in global AI systems, and demonstrates how community-driven annotation can produce high-quality training data.

📄 Paper: arxiv.org/abs/2306.07426