Lagos, Nigeria — Tech giant Google has significantly expanded its digital coverage across Africa by mapping over 150 million previously undocumented buildings, in what it calls one of the most ambitious artificial intelligence (AI) infrastructure initiatives ever undertaken on the continent.

The breakthrough project, driven by Google Research and spearheaded by Abdoulaye Diack, Programme Manager at Google, uses advanced satellite imagery and machine learning to detect and document buildings in underserved urban and remote areas — many of which were previously invisible on digital maps.

“When we started, we noticed that many neighbourhoods in Africa weren’t even visible in Google Maps. Our project helped fix that,” Diack said in an interview with The PUNCH. “Almost half of the buildings you see on Google Maps in Africa came from our initiative.”

From Ghana to the Globe

Launched initially in Ghana, the project quickly scaled continent-wide and has since informed global applications. Using European Sentinel satellite data, which updates every five days, the AI was trained to recognize structures from lower-resolution images a feat made possible by innovative model efficiency techniques necessitated by Africa’s data limitations.

The algorithm was not only able to identify building footprints including informal settlements and scattered rural homes but also estimate building heights by analysing shadow lengths relative to the time of day, resulting in height data for over 1.8 billion buildings worldwide.

Transforming Urban Planning, Disaster Response, and Agriculture

The project has had tangible impact beyond navigation. The new maps are already aiding urban planners, emergency responders, and economists by offering insights into the physical structure of African cities, particularly where official data has long been scarce.

But Google’s AI work in Africa goes far beyond maps.

Through PlantVillage, one of its earliest Africa-centric projects in partnership with Uganda’s Makerere University, farmers now use smartphones to detect cassava diseases before symptoms become visible. The AI was trained on thousands of plant images collected directly from farms a process Diack described as “simple, but critical.”

“We send people to farms to take pictures of plants, and over time, the model learns by comparing patterns,” he explained.

Bridging the Language Gap with AI

Another critical area of Google’s focus is language inclusion. The company is building speech and text datasets for underrepresented African languages through partnerships with grassroots organisations like Masakhane and academic institutions across the continent.

Africans are asked to record themselves in their native languages and dialects, which helps the AI recognize regional accents and local syntax a step essential to making digital tools truly accessible.

“We must invest in datasets that reflect African languages and realities,” Diack stressed. “If AI tools can’t understand you, they can’t serve you.”

Investing in AI Talent Across Africa

Google is also focused on talent development, running programmes like the AI Residency and partnering with bodies such as the African Institute for Mathematical Sciences and AfricaToML to prepare early-career researchers and engineers for roles in machine learning.

“People with mathematics and physics backgrounds tend to do well,” said Diack, highlighting the need for long-term investment in grassroots training, community-building, and skills development.

The Bigger Picture: AI for Africa, by Africa

As tech adoption accelerates across Africa, Diack believes that localized AI innovation is key to unlocking the continent’s digital potential. For that to happen, he said, data accessibility, community inclusion, and homegrown expertise must remain front and center.

“Africa’s data scarcity pushed us to innovate. Now we’re using those innovations to solve global challenges,” he said.