Microsoft is making some good noises about AI in our motherlands
When it comes to game-changing developments like artificial intelligence (AI), and messenger RNA vaccines, Africa is often like the grumpy lad in the township along the highway. ILLUSTRATION | JOSEPH NYAGAH | NMG
Journalist, writer and curator of the Wall of Great Africans
Tech giant Microsoft recently posted a blog — or better a visionary white paper — titled AI in Africa: Meeting the Opportunity. It is not for reading on an easy Sunday breakfast, but if you are sitting somewhere in Nairobi stranded by floods, it is worth the slog if one glosses over a lot of the self-back-patting.
Perhaps except for mobile phones, when it comes to game-changing developments like artificial intelligence (AI), and messenger RNA vaccines, Africa is often like the grumpy lad in the township along the highway.
He sits by the highway, counting the vehicles as they drive by, bemoaning his misfortunes, cursing those who whizz past in fancy cars, and railing against the local MP who promised to make them rich, but has never returned since they elected him.
Life goes by, the world moves forward and, as someone put it, Africa then waits “to talk about leapfrogging”. Microsoft does a great service because it does a whole lot of good showing that AI in Africa is not all hot air. Innovative businesses, organisations, and governments have been doing important work with it. I must confess I was a little awed to see the extent of their exertions.
There are many brilliant examples all over Africa, but we shall be tribal and focus on the standout cases from the East African Community.
Zipline: If you are feeling lazy, best to jump to the “AI in Action in Africa” section. Here is a sample. It starts with Zipline, which was birthed in Rwanda in 2016. It started using drones to deliver blood and medical products to hospitals and remote health centres and has since expanded to food, retail, agriculture products, and animal health products. It has taken the use of AI to manage and optimise flight paths and delivery schedules to a high level.
The report says this system enables rural hospitals and clinics to place orders with Zipline and receive essential medical items like blood, platelets, and frozen plasma in as little as 30 minutes. With the success in Rwanda, Zipline expanded to Ghana, Nigeria, Tanzania, Kenya, and Cote d’Ivoire.
Amref Health Africa: With the Ministry of Health in Kenya, Amref is developing what Microsoft calls an “innovative spatio-temporal machine learning model”. Spatial refers to space, and temporal refers to time. What these clever people are saying is that the AI model they are developing will look at past data (in this case 50 years), and real-time data from community and other health workers. They will use that to detect malnutrition hotspots in Kenya and intervene in a timely fashion to save lives.
M-Kopa: M-Kopa, established in Kenya in 2012, is a quite impressive fintech company headquartered in Nairobi. It lends money to the “small people” so they can buy essential things like solar lighting, televisions and refrigerators. M-Kopa, which has since expanded to Uganda, Nigeria, Ghana and South Africa, uses AI to do many things, including generating daily forecasts for loan repayments. It has considerably improved customer repayment performance.
AKI, Association of Kenya Insurers: The insurance industry is notoriously plagued by fraud and can be insufferably bureaucratic and inefficient. AKI called in Sheriff AI. Among many things, the industry has slashed fraud (which previously comprised 25 percent of the claims), cut the time required to file a claim from 10-12 days to just six minutes, and brought down the cost of processing insurance certificates by 50 percent.
Pulse Lab Kampala: “Uganda, known for its progressive refugee protection policies, has witnessed a rapid increase in its refugee population…This surge has strained resources and heightened tensions between local communities and the refugee population.
To address these challenges, Pulse Lab Kampala, in collaboration with the UNDP Operations Coordination Office, has developed a Radio Content Analysis Tool that employs AI for real-time analysis of public radio discussions about refugees, functioning similarly to social media analysis tools.
This AI-powered tool enables them to gauge the level of concern and sentiment among local communities towards refugees. The insights gained serve as an early warning system and assist in making more informed decisions.
Forest Guard: An AI-powered solution to tackle illegal logging in Kenya which was conceived by four interns during a Microsoft hackathon, the device is equipped with sound and geosensors, and machine learning capabilities. It can identify the sound of chainsaws, a common indicator of illegal logging activities. When illegal logging is detected, the system automatically sends SMS and email alerts to forest officials.
There is a lot of cool stuff happening outside the EAC, including the Elephant Listening Project in the Nouabalé-Ndoki National Park in the Republic of Congo. The African forest elephant has experienced a massive population decline, with numbers falling by over 86 percent between 1989 and 2020. “Recognised as a keystone species, their survival is crucial for the health of their ecosystems,” the report said.
The Elephant Listening Project uses acoustic sensors that track their movements in the wild. The start-up Conservation Metrics developed an AI tool capable of distinguishing elephant calls from other rainforest sounds. These sensors capture audio files that are then processed using machine learning and deep neural networks.
The remoteness of their habitats and the large size of the sound files often resulted in lengthy data collection and analysis periods, delaying timely interventions against immediate threats like poaching.
The application of AI speeds the analysis of these files, allowing scientists to promptly alert park rangers about potential threats.
Brilliant!
Charles Onyango-Obbo is a journalist, writer, and curator of the ‘Wall of Great Africans’. X@cobbo3
