Abstract
This study explores technical considerations for designing, developing, adopting, and using AI-based systems in Africa. Africa did not benefit as intended from the first three industrial revolutions. Cognizant of this fact, the continent is now expected to be aware of and ready to tap into the opportunities created by AI and the Fourth Industrial Revolution to fix chronic problems connected to efficiency while minimizing the unintended consequences AI might pose. Data for the study was gathered through focus group discussion (FGD), key informant interview (KII), and document review. The outcome of the study reveals that AI model adoption issues, AI biases, AI data availability, security, and privacy, AI model accuracy and quality, and AI resources have emerged as major technical considerations for adopting and using AI in the African context. The chapter provides valuable insights that would serve as input for policy formulation and AI capacity development endeavors.
TopIntroduction
Nations around the world are showing a vested interest in tracking developments in AI and other emerging technologies. This is because AI and the associated technologies have the potential to disrupt and transform prevailing socio-economic activities. Scholars from around the world argue that AI would be a key driver of the fourth industrial revolution which is believed to be the next stage of socio-economic development. For various reasons, Africa could not benefit as intended from the first three industrial revolutions. A more advanced AI technology emerged before the continent could optimally use the conventional ICTs improve the backbone telecom infrastructure and develop the necessary human resources. In connection to this, Ade-Ibijola and Okonkwo (2023) point out that a lack of technical skills, uncertainty, lack of structured data, lack of government policies, ethics, and user attitude are likely to impede the adoption of AI-based technologies in Africa.
Artificial intelligence (AI) is a relatively new phenomenon in the socio-economic arena of Africa. As the continent is entangled with various challenges, rapidly emerging technologies like AI have grabbed less attention from policymakers in Africa. However, currently, a consensus seems to have been reached among the scientific community, policymakers, and African leaders recognizing that AI has enormous potential to fix some of Africa's chronic problems connected with efficiency and optimal use of resources. Among others, Africa's agricultural and healthcare sectors generate a huge amount of data that can be used to solve critical problems in these sectors.
Applications of AI that center on solving real-world problems affecting the lives of millions of ordinary people are supposed to bring about meaningful transformation in the continent (Nayebare, 2019). So far, most AI applications adopted in the cited sectors are developed in Western countries (Berhane, 2020). However, some initiatives are underway in different parts of the continent to foster local AI capacity development. For instance, Arakpogun et al. (2021) disclosed that inspired by AI Labs developed with the support of leading tech companies such as IBM in Kenya, Google Lab in Ghana and other labs established by initiatives of universities like the University of Cape Town and Makerere University, around 100 AI start-ups have emerged across Africa.
Although AI-related matters appear daunting to Africa given the current state of technological development and readiness to adopt such cutting-edge technologies, it is imperative to understand the potential impact of the technology and reduce potential marginalization that would occur because of disregarding this technology. To this end, studies have been undertaken to understand the current state of the continent in terms of adopting the technology. For instance, Baguma et al. (2022) propose an AI readiness index for Africa. The authors argue that the readiness index signals the continent’s capacity to harness AI for socio-economic development. tied to the proposal of Baguma et al. (2022), other scholars such as (Arakpogun et al. 2021; Ade-Ibijola & Okonkwo 2023; Butcher et al. 2021) have expressed the challenges AI might pose in Africa and the opportunities the technology might bring to rectify the continent’s persistent problems. These scholars contend that AI tends to amplify the historic structural inequalities among African citizens causing a disproportionate level of access to resources such as education, employment, healthcare services, etc., and eventually tends to transfer existing inequalities into a digital space and create a further digital divide. Particularly, Sub-Saharan African (SSA) countries are likely to encounter another round of digital divide due to inadequate telecom infrastructure, unreliable electric power, unaffordable smartphones, and a lack of digital skills (Arakpogun et al., 2021). AI involves the fusion of a range of technologies. This will entail another challenge of regulations and governance as most SSA governments lack the required institutional capacity, skills, and financial resources. The inability to regulate dynamic and complex AI technologies can lead to unintended consequences surrounding citizens’ privacy and data security and eventually threaten national security (Arakpogun et al., 2021).
Key Terms in this Chapter
Trained Models: A set of computer programs that involve statistical models such as linear or logistic regressions, Neural Networks, or deep learning trained and validated using massive datasets and ready to be reused in other settings with or without fine tuning.
Digital Ecosystem: An integrated set of entities and components linked to interact, share
Resources: and reinforce each other in order to establish a robust self-regulating AI environment.
AI bias: A scenario where a computer program carries an inherent human bias or bias created in the process of developing the computer algorithm, eventually leading the AI system to generate discriminatory and unbalanced results.
AI Resources: An infrastructure that includes database systems, hardware, software, communication systems, and essential procedures to develop, deploy, and set up AI-based technologies in a particular setting.
AI Algorithm: A computer program typically designed to perform high-level tasks such as prediction, classification, and revealing other interesting patterns from huge datasets.
AI Model: A packaged computer program designed to learn from massive data sets and generate a pattern.
Machine Learning: A phenomenon where a computer program is designed to learn from huge datasets, sensors, and other data sources and generate interesting patterns, make a suggestion, or make relevant decisions.
Artificial Intelligence (AI): A computer system that mimics a human being and performs some tasks which mainly used to be carried out typically by a human being such as decision-making and logical reasoning.