According to a recent survey, the top three African nations using AI are Kenya, Nigeria, and South Africa.
According to the paper AI for Africa: Use Cases Providing Effect, of the 90 applications discovered in the three nations, 40–49 are from Kenya, 30-39 from Nigeria, and 20–29 from South Africa.
The paper, which focuses on Kenya, Nigeria, and South Africa, was created using previously conducted research as well as interviews with leaders in civil society, non-governmental organizations (NGOs), academia, and the commercial sector. It also offers trends and insights specific to an area.
Agriculture accounts for over half (49%) of AI applications, with climate action and energy coming in second and third, with 26 and 24 percent, respectively. This is consistent with the report’s findings that agriculture still plays a major role in Kenya and Nigeria.
Currently, the continent contributes only $0.4 trillion, or 2.5 percent, to the $16.5 trillion global AI industry. Africa’s GDP is expected to grow by $2.9 trillion to over $5 trillion thanks to AI.
However, if significant issues are not resolved, this will not be simple. The lack of readily available, high-quality, locally relevant data is one of the primary obstacles in Africa.
Major advances in AI have been made, mostly in the Global North, where there are abundant datasets, strong processing power, and models that can be trained quickly and effectively with vast amounts of data.
According to the paper, they “may not be appropriate or representative for African contexts, and carry inherent risks of exacerbating biases present in the data they are trained on.” It demanded locally relevant data and gave local talent the authority to handle and analyze it.
Prejudice toward women is one danger. Women are underrepresented in datasets worldwide. The lack of gender-disaggregated data gathering in African nations makes this worse.
The small number of languages in Africa’s dataset is another obstacle. The majority of the data used to train huge language models, such as Open AI’s ChatGPT, which understands and produces human language text, comes from Western and English-speaking nations. Furthermore, they may result in biases and errors when applied to Africa. The research said that achieving diversity and accessibility requires a strong emphasis on local languages.
Infrastructure is another difficulty. The continent requires cloud computing systems, graphics processing units (GPU), and high-performance computing to train AI.
GPU costs in South Africa and Kenya, respectively, are nine and thirty-one times more than those in affluent nations.
African nations also need dependable energy, high-speed broadband, mobile internet, and facilities with storage capacity, such as data centers.
Africa lacks sufficient data centers, thus its data is often transported via undersea fiber-optic cables and housed on remote servers. Still, there are very few of these lines that reach Africa.
According to the research, “this means that the continent is particularly vulnerable to disruption in these networks, unlike other parts of the world where traffic can be easily rerouted and network redundancy is robust.” It also said that by 2030, African nations would have to more than treble their ability to host data centers.
Next, there are requests to increase the energy efficiency of data centers. On this front, there has been some progress. The report’s authors also noted that data centers including Africa Data Centers in Nigeria and South Africa’s Amazon Web Services are investigating sustainable energy sources. The Ecocloud Data Center in Kenya is the first data center in Africa to run entirely on geothermal energy.