Despite what many think, AI is not just about chatbots and virtual assistants. It is a powerful technology that can help address some of the most pressing challenges in the world, such as health, education, agriculture, and the environment. Hence, Google is investing in the potential of AI to solve some of Africa’s pressing challenges. In 2018, the tech giant launched its first AI research centre on the continent in Accra, Ghana. In 2022, it expanded its presence by opening a product development centre in Nairobi, Kenya. The Accra Center, also known as Google Research Africa, is a hub for cutting-edge research and innovation, where a team of mostly African experts work on solving local and global challenges using AI.
Some of the projects that the centre has been working on include using AI to improve crop yield and food security, AI to reduce maternal mortality and neonatal complications, AI to enhance natural language processing and speech recognition, and AI to support environmental conservation and disaster management. The centre collaborates with various partners, such as universities, NGOs, and government agencies, to develop and deploy AI solutions that are relevant and impactful for Africa and beyond.
Yossi Matias, the head of Google Research Africa, said that the centre aims to change the way innovation happens in Africa. Instead of importing solutions, the centre is creating tech solutions that are tailored to Africa’s needs and opportunities, and that can also be applied globally. Here are five challenges in Africa that Google aims to address with AI.
AI for precise flood forecasting
Floods are a major threat to millions of people in Africa, especially in West and Central Africa, where they have caused massive displacement, injury, and death. According to the UN, floods affected 8.2 million people, displaced 2.9 million, injured 4,398, and killed 1,418 in these regions as of December 2022. In 2012, over 2.5 million Nigerians were affected by floods. the devastating. A post-disaster assessment survey revealed the flood must cost Nigeria an estimated N2.6 trillion. The situation is worse for the whole continent.
One of the best ways to prevent the devastating effects of floods is to have early warning systems that can alert people and authorities before the disaster strikes. However, many people and farmers in flood-prone areas lack access to such systems and are often caught unprepared. Google is trying to change that with its AI for flood forecasting, a technology that can predict the occurrence, location, and severity of floods up to seven days in advance. The AI uses two models to make these predictions: the Hydrologic Model, which estimates the water flow in a river, and the Inundation Model, which determines the areas that will be flooded and how high the water will rise.
Google publishes all its flood forecasts on its Flood Hub platform, which is updated daily and freely available to the public. Flood Hub covers river basins in 80 countries, including 23 in Africa, and provides flood forecasting services to 460 million people across 1,800 sites. Google also sends flood alerts to people through Google Maps, Google Search, Maps, and Android notifications, thanks to the work of Florence Ofori, a Google research resident. Additionally, Google partners with governments and NGOs, such as the UN, to increase the reach and impact of these alerts. By providing timely and accurate information about floods, Google enables people and authorities to take proactive measures to save lives and properties. This is how Google is using AI to tackle one of the most pressing challenges in Africa and beyond.
AI for weather forecasting
Africa faces frequent and severe weather events that affect the lives and livelihoods of millions of people, especially those who depend on agriculture. The World Bank’s 2021 Africa’s Pulse report shows that droughts and storms have increased significantly in the past decade, compared to the 1970s. These extreme weather fluctuations have a huge impact on the livelihood of the residents, as 70% of Africans depend on agriculture.
To cope with these weather challenges, accurate and timely forecasting is essential. Google is using AI to improve weather forecasting in Africa, through its tool called Nowcast. Nowcast uses machine learning models to estimate precipitation conditions with high precision. Emmanuel Brempong, the lead researcher for Google AI for weather forecasting, said that Nowcast is the foundation for long-term forecasting, as it provides reliable short-term forecasts. He also said that Nowcast has been used to provide early warnings and weather information across Africa.
With better weather forecasting, governments, industries, and individuals, especially farmers, can plan and prepare for weather events and optimize food production. This can reduce the vulnerability and uncertainty of the 57% of Africans who rely on agriculture for jobs.
AI for food security
Food insecurity is another serious issue in Africa, affecting around 140 million people who face acute hunger, according to the Global Report on Food Crises 2022. One of the factors that threaten food production in Africa is locust infestations, which can destroy crops and pastures in a matter of hours. In 2022, South Africa experienced the worst locust outbreak in 25 years, affecting five million hectares (19,305 square miles) of grazing land in the Eastern Cape. Some reports believe that the outbreak could have been prevented if there were early warning systems in place.
Google Research Africa is working on two AI solutions to help farmers and authorities deal with locust outbreaks. The first one is a collaboration with InstaDeep, an AI company, and the FAO, the UN Agency for Food and Agriculture, to develop systems that can detect locust outbreaks and suggest preventive or control measures. The second one is an AI model that can use historical data from the FAO and weather variables like temperature and rainfall to predict locust breeding grounds.
Google research scientist, William Ogallo, said that the AI model for locust forecasting is part of a broader project to provide accurate and timely food security predictions that can help farmers plan and allocate their resources efficiently. He also said that Google’s solution is different from existing projects like FEWSNET and Hunger Map Live, which rely on manual surveys of experts to make predictions. Google’s solution aims to use multiple sources of data, including Google’s internal datasets, such as trends data, to create a more effective and scalable model for forecasting food insecurity situations.
By using AI to improve locust and food security forecasting, Google hopes to help millions of people in Africa and the world who are affected by hunger and malnutrition. This is how Google is using AI to address one of the most critical challenges in Africa and beyond.
Mapping Africa through Open Buildings Dataset
Many African capitals have well-planned cities, but outside of them, the reality is different. This is why Aderonke “Ade” Ajose-Adeyemi, entrepreneur and CEO of Losode, an e-commerce platform believes the e-commerce sector still has poor logistics infrastructure. Poor economic planning leads to problems like inaccurate maps and population data. For example, in many rural areas in Nigeria, it is hard to provide electricity because the government does not have precise information about where the communities are, what kind of buildings they have, and what their electricity needs are. Surveying these areas is not feasible, as the roads are often in bad condition and inaccessible during the rainy season. Without proper mapping, it is difficult to plan infrastructure for these areas.
Google Research Africa has been working on a solution to map Africa’s built environment using satellite imagery and machine learning. They trained a deep-learning model to detect the outlines of buildings from high-resolution satellite images. With this data, they created a large-scale open data set that shows the location and shape of buildings in Africa. This data set can help with urban planning, population estimation, humanitarian response, and service delivery to underserved communities. Abigail Anka, a research software engineer on this project, said that the project has evolved since it started in 2019. The third version of the data set is already available, and she expects more improvements every two to three years. This project has increased the number of buildings in Africa on Google Maps by four times, reaching 250 million. The project has also expanded to other continents, providing information on 1.8 billion buildings worldwide.
Improving maternal health outcomes
As of 2020, the maternal mortality rate in the African Region was 531 deaths per 100,000 live births, accounting for 69% of global maternal deaths. One of the projects he mentioned is about using AI to improve ultrasound screening for pregnant women. Ultrasound is an important tool for detecting pregnancy complications early, but it is not widely available in Africa, due to the lack of trained sonographers and the complexity of ultrasound devices. Google Research Africa is collaborating with Jacaranda Health, a maternal health organization, to develop AI models that can simplify the use of handheld ultrasound devices. The goal is to enable non-experts to capture and interpret ultrasound images easily and increase the access of pregnant women in underserved areas to ultrasound screening. This can help reduce maternal mortality.
Another project that Google supports for maternal health in Africa is the Mama’s Hub Android app, which collects timely information about expectant mothers. The app is built using the Open Health Stack (OHS), a set of open-source resources that developers can use to create digital health solutions. OHS is also a project supported by Google.