The promise of AI in journalism collapses quickly when it meets the reality of African data systems, where missing records and inconsistent documentation matter more than model sophistication.
That gap becomes clear in the experience of building Nubia, an African-focused data journalism and analytics platform designed to support reporting through structured data and computational tools. The core idea was not just to automate journalism, but to make African information systems more usable for analysis, reporting, and public accountability.
Over time, one challenge kept repeating itself: data scarcity and fragmentation. Many African institutions still publish information in formats that are not machine-readable, ranging from PDFs to scanned documents and unstandardised reports. Even when data exists, it is often uneven across sectors and countries, making comparison and large-scale analysis difficult without significant manual intervention.
For journalists and researchers, this creates a different kind of workflow. AI tools can assist with summarisation or pattern detection, but they struggle when inputs are incomplete or unreliable. In practice, a large portion of the work still goes into cleaning, verifying, and contextualising information before any automation becomes useful.
What this reveals is that African AI systems are shaped less by model limitations and more by infrastructure realities. The bottleneck is not intelligence, but the availability and structure of data itself. That shifts the focus away from building more advanced models and toward improving how information is collected, stored, and maintained across institutions.
The next phase of AI in African media will likely depend less on headline breakthroughs in machine learning and more on slow, structural improvements in data ecosystems.
Leave a Reply