Artificial intelligence apps have the potential to boost the African continent’s economic growth by close to US$3 trillion over the next six years – and mobile operators will be key to unlocking that, says the GSMA, the world’s biggest mobile operator association.
The GSMA released a new report on Monday that identifies over 90 AI use case applications in Kenya, Nigeria, and South Africa that it says can “drive socio-economic and climate impact.”
According to the report – “AI for Africa: Use Cases Delivering Impact”, which combines existing research with interviews with experts across civil society, NGOs, academia and the private sector – the vast majority of potential AI use cases in Africa are related to agriculture (49%), climate action (26%) and energy (24%).
The GSMA says these use cases could be part of a wave of emerging AI apps that, according to a March 2024 study from AI4D Africa, could boost the African continent’s economic growth by $2.9 trillion by 2030, the equivalent of increasing annual GDP growth by 3%.
For example, in sub-Saharan Africa – where up to 80% of food is produced by traditional smallholder farmers who lack access to information that would help improve yields –“the majority of AI use cases in agriculture involve machine learning (ML) enabled digital advisory services, which equip farmers with data-driven advice to adopt climate-smart farming practices and optimise productivity.”
Meanwhile, AI can also enable more affordable and reliable energy services with use cases such as predictive maintenance, smart energy management, energy access assessment and productive use financing to monitor and extend services in energy-scarce areas, the report says.
With African countries being disproportionately impacted by climate change, AI could also support use cases to mitigate that impact, such as biodiversity monitoring, wildlife protection and early warning systems for climate emergencies and other natural disasters.
According to the GSMA report, most (98%) of these kinds of use cases are essentially predictive AI applications that leverage ML approaches, due to the availability of historical datasets, ease of application and lower computation requirements compared with generative AI models.
This is key because major barriers to AI adoption for African countries include limited availability of data centres and the rising cost of cost of hardware and software. AI requires GPU-level compute power, and in countries like South Africa and Kenya, for example, the cost of a GPU representing 22% and 75% of GDP per capita, respectively, the report says. Even when data centres get built, the energy demand is an extra expense that also taxes the power grid.
Which is why the GSMA report recommends taking advantage of the fact that mobile networks are the primary modes of internet access to ease the availability of smartphones and use mobile-based AI solutions as “a practical way to circumvent current limitations and tap into AI’s full potential across the continent.”
Essentially, the idea is to reducing reliance on high-powered data centres by using mobile networks to develop distributed or hyperlocal edge computing, where tasks occur on devices including phones and laptops. After foundational models are trained on large datasets, AI models can be transferred to smartphones for fine tuning.
“With smartphone penetration at 51% and expected to reach 88% by 2030, mobile-based edge computing will be central to expanding the proliferation and capabilities of AI in Africa,” the GSMA report says.
That said, it’s not all about mobile. The GSMA report also lists other barriers and challenges that need to be addressed. Key among these is the lack of availability of local-language data, which is needed to train AI models so that they “reflect the complexities and nuances of African markets rather than mimic data from the Global North.”
Max Cuvellier Giacomelli, GSMA’s head of Mobile for Development, said that there also “needs to be a strong focus on increasing skills for both AI builders and users, especially among underserved populations. Better training programmes are essential, particularly in the face of a global brain drain on AI talent.”
Giacomelli added that AI development in Africa will further require strong partnerships across a broad ecosystem of partners including ‘big tech’, NGOs, governments, and mobile operators.
Also, he continued, “Policies must also evolve to address inequality, ethics, and human rights concerns in AI deployment.”
