AI: Africa’s turn with planning and disaster response tools, centre

pangu

A staff member of Huawei introduces the Pangu AI weather models during the World Artificial Intelligence Conference in Shanghai, China on July 7, 2023. PHOTO | AFP

In September, an artificial intelligence (AI) tool known as Flood Hub successfully predicted the flooding in South Africa’s Western Cape.

The Flood Hub app, developed at Google’s Africa research centre in Ghana, is designed to offer advanced flood warnings. It currently covers 23 African countries and countless others worldwide.

“On the app, you can easily zoom in on specific areas of interest to assess water levels and flood risks,” explained a researcher at the centre, Florence Ofori.

“Partner organisations such as Red Cross and GiveDirectly help disseminate this crucial information. Additionally, push notifications are sent to residents in flood-prone areas based on predictions generated by our AI models.”

On its part, Open Buildings Dataset, leverages machine learning and satellite imagery to train an AI model to detect buildings, with the aim to bridge the gap in scarce building data in African countries. The project tracks the changes in buildings, providing insights for urban development and infrastructure management.

The project, launched in the Ghana centre, in 2021, has so far captured 60 percent of African buildings.

“This initiative supports various applications such as population census planning, mapping economic distribution, and enhancing public health planning and disaster response. The impact on humanitarian response is notable, as the dataset supports automatic building damage assessment and aids organisations including the UN in disaster response and immunisation planning,” explained research software engineer, Abigail Anka.

Expanding field

AI, although still in its nascent stages, is a rapidly expanding field that holds immense potential to transform the way problems are solved, says Google’s vice president, engineering and research, Yossi Matias, at a media tour of Google’s AI research centre in Ghana on December 12, where engineers and machine learning experts from Africa presented AI capabilities in weather forecasting, food security and healthcare, among others.

Google established its inaugural AI centre in Accra, Ghana for the design and development of AI technologies as supportive tools to address the unique challenges faced by the continent.

In its quest to combat hunger and food insecurity, the team has embarked on multifaceted AI programmes that harness the potential of AI in weather analysis with an eye on agricultural practices, crop management, and food production. Relatedly, another AI programme under development in Nairobi, is focusing on identifying and monitoring regions susceptible to food insecurity, by tracking hunger crises.

Limited weather expertise

Africa’s inadequate weather forecasting capabilities and limited weather observation network with only 37 stations double as challenges.

To overcome these limitations, the Ghana centre combines diverse data sources, including global precipitation measurements by Nasa and data from the European Weather Agency, to generate accurate spatial and temporal weather forecasts. This approach has shown promising results, outperforming traditional methods and potentially leading to significant advancements in AI-powered meteorology.

But computer scientists identified data shortages as a significant challenge for AI technologies in Africa. The success of AI initiatives relies on the quality and representativeness of the training data, reflecting the demographic variables of the targeted population.

Other obstacles are a shortage of technical skills, lack of structured data ecosystem, absence of government policies, ethical concerns, and uncertain perceptions.

To promote the adoption of AI technologies, the AI experts say stakeholders, particularly governments, need to improve knowledge base and understanding on how these technologies can be put to use for maximum impact.