The missing piece in the global race for public-sector AI

Artificial intelligence took centre stage at the UN General Assembly this week, with world leaders urgently calling for new regulatory guardrails. 

Yet, as governments rush to deploy machine learning across public services, they risk overlooking a critical preliminary step: the underlying data architecture. A government's digital records function merely as a map of society, and that administrative map is frequently outdated, fragmented, or fundamentally misaligned with the reality on the ground. 

In a new paper by Genesis AI solutions architect Bryson Mwamburi Mwakuwona, he argues the  gap between digital representation and physical reality poses a significant systemic risk.

When AI systems operate on disconnected or inaccurate public data, they do not inherently correct these blind spots. Instead, they industrialise the errors, scaling foundational flaws across national infrastructure rather than resolving them. 

Before determining what artificial intelligence can automate, infer, or recommend, nations must invest in robust Digital Public Infrastructure (DPI) to cultivate an information environment that accurately represents their populace. 

A model can be imported, but the trusted context it needs to understand a country cannot.   

For a comprehensive analysis on building trustworthy data ecosystems for machine reasoning, read the full paper: The Map Is Not the Terrain Digital Public Infrastructure and the Data Foundations for AI-Enabled Government.