1At the dawn of the 21st century, the internet has emerged as a revolutionary innovation with the potential to empower Africa. Similarly, artificial intelligence (AI) is now seen as a game-changer for Africans.
- 1 N. Butcher; M. Wilson-Strydom; M. Baijnath; D. Orlic; M. Smith; B. Neupane; J. Shawe-Taylor, 2022. (...)
2Africa’s population has grown exponentially over the past 50 years, with 60% of its 1.26 billion people under the age of 25. With its vast natural resources, Africa has the potential to become the world’s largest economy if it combines its advantages with investment capital, research and development, effective policies and highly skilled talent1. This would enable the continent to capitalise on the opportunities presented by Industry 4.0.
- 2 Gwagwa, A., Kraemer-Mbula, E., Rizk, N., Rutenberg, I., & De Beer, J. (2020). Artificial intelligen (...)
3Without doubt, AI has the potential to bring transformative change to African economies, leading to a reduction in poverty and inequality, a reorganisation of work, skills and production, reform of agriculture and agribusiness, and improvements in healthcare systems and human capital2.
- 3 Baijnath, M., Wilson-Strydom, M. and Butcher, N., 2022. ARTIFICIAL INTELLIGENCE CAPACITY IN SUB-SAH (...)
4At a regional level, no country in sub-Saharan Africa (SSA) features in the top 10 countries expected to benefit most from AI and the other emerging technologies set to usher in the fourth industrial revolution. Despite this position, the unprecedented advances in AI are having a considerable impact on African societies3.
5While Africa is the least digitalised continent in the world, this could be seen as an obstacle or a handicap to its technological development. But from a regulatory point of view, this obstacle could be transformed into an advantage. Europe took more than 15 years to draw up a regulatory framework on the use of personal data (GDPR) after the digital explosion in the northern hemisphere. By contrast, Africa, which is in its early stages of digital maturity, can develop the necessary regulations in parallel with the digitisation of the continent.
6This type of duality, advantage/disadvantage, opens up a debate on the type of AI-generated innovation that Africa would like to see in terms of the United Nations (UN) Sustainable Development Goals (SDGs), and forces us to question AI innovation in Africa, its opportunities, its risks, and the legal context associated with it.
7The rest of the paper is organised into three sections: Section 1 analyses the opportunities and successes of AI innovation in Africa. Section 2 examines the evolution of the legal framework for AI in Africa. And finally the last section looks at the risks and negative scenarios associated with AI innovation in Africa.
- 4 Kiemde, S. M. A., & Kora, A. D. (2022). Towards an ethics of AI in Africa: rule of education. AI an (...)
- 5 Faria, Julia (2022): Funding value secured by tech startups in Africa 2015-2021. 9 February 2022.
- 6 Oxford Insight, "Government AI Readiness Index 2020 - Oxford Insights," 2020. https://www.oxfordins (...)
8Artificial intelligence is opening Africa and the rest of the world up to numerous new opportunities for innovation. Africa is the continent that should benefit most from AI for its development, despite an ecosystem that is not always favourable to its development (Kiemde et al, 2022)4. Despite the challenges, the number of African technology start-ups has increased significantly in recent years. These startups raised more than $2.1 billion in 2021, the highest amount of venture capital raised since 2015 (Faria, 2022)5. According to a report by Oxford Insights and the International Development Research Centre (IDRC) on how governments can position themselves to take advantage of AI-driven transformation, Kenya, Tunisia, Mauritius, South Africa and Ghana are ranked in the top five countries in terms of government readiness for AI (Oxford Insights, 2020)6. AI impacts all areas including education, health, commerce, financial services and agriculture.
9In education, AI techniques can improve the learning experience through personalisation and real-time adaptation to learners’ profiles and learning styles. Through AI-powered personalised tutoring and learning resource recommendation systems, the efficiency of education is improved and the number of people with access to quality education increases, representing great potential for African countries.
10In healthcare, AI can be used in a variety of ways to improve healthcare services through personalised medical assistance, automated and remote diagnosis, and the discovery of new drugs and vaccines. The greatest success ever achieved in Africa in this field was achieved by the Tunisian-British start-up Instadeep. Founded by two Tunisian engineers, Karim BEGUIR and Zohra SLIM, it was acquired in January 2023 by German biotech specialist BioNTech for €408 million - a record in the field of AI and machine learning! - Instadeep supports BioNTech’s strategy to develop world-leading capabilities in AI-driven drug discovery and the development of next-generation immunotherapies and vaccines to address diseases with significant unmet medical needs. The transaction is expected to add approximately 240 highly skilled professionals to BioNTech’s workforce, including AI, ML, bioengineering, data science and IT development teams.
- 7 Sampene, A. K., Agyeman, F. O., Robert, B., & Wiredu, J. (2022). Artificial Intelligence as a Path (...)
11By analysing financial data and detecting fraudulent activity in real time, AI can help to improve the efficiency and security of financial transactions in Africa. In the financial sector, AI can help reduce the risk of fraud and facilitate access to many people. For example, Kudi is a Nigerian startup founded in 2016 ranked as the leading AI company specialising in Fintech in Africa. Generating over $30 million in monthly revenue, through a chatbot-enabled platform, Kudi collaborates with commercial banks and other financial service providers to deliver a variety of financial products to small informal businesses and individuals, such as money transfers, bill payments, cash withdrawals, loans and insurance products (Samepene et al., 2022)7.
- 8 FU, G. (2022). Toward Ethical Artificial Intelligence in International Development. CDA Insight 202 (...)
12In the agricultural sector, South African startup Aerobotics uses drones and satellite imagery to help farmers optimise crop yields in Malawi, Zimbabwe and Mozambique (Fu, 2022)8. Its AI platform provides advice to identify underperforming areas caused by pests or other deficiencies. It then monitors the impact of agricultural interventions on crops. The startup has generated information on more than one hundred million trees to help growers, investors and insurers improve production and profitability.
- 9 Alupo, C. D., Omeiza, D., & Vernon, D. (2022). Realizing the Potential of AI in Africa: It All Turn (...)
13From an economic point of view, AI is the basis of what is known as its fourth industrial revolution - Industry 4.0. In order not to miss out on this revolution, member countries of trading blocs around the world have developed or are in the process of developing national AI strategies (Alupo et al., 2022)9. AI, combined with other Industry 4.0 technologies (AR/VR, big data, IoT, cybersecurity,...), is at the service of productivity within factories and makes it possible to address various issues such as green production, waste management or energy optimisation. For example, the Internet of Things (IoT) and AI are at the heart of predictive maintenance for industrial equipment. Predictive maintenance is essential for improving the availability of equipment and machinery in industrial production chains. It enables a problem to be detected before it occurs, and the component to be replaced at a time that does not disrupt operations, thereby reducing the risk of unforeseen events and delays in production and delivery times.
- 10 World Bank. (2021). Harnessing Artificial Intelligence for Development on the Post-COVID-19 Era: A (...)
14In order to take advantage of the potential offered by AI technologies in Africa, public policies must be based on existing strengths in African countries and on strategies that take into account current challenges. In general, seven pillars, associated with public policy areas, represent distinct thematic areas that are central to the development of policies to accelerate the development and adoption of AI at the national level (World Bank 2021)10. These seven pillars can be grouped into:
-
Four "core" pillars consisting of promoting and retaining AI talent, developing digital infrastructure and computing power, fostering AI and data governance, and promoting AI research, development and innovation.
-
Two "acceleration" pillars to stimulate entrepreneurship in the private sector and encourage the adoption of AI in the public sector. These include, for example, professional training programmes for public employees and civil servants, or open innovation programmes aimed at co-creating AI solutions to solve concrete problems in the private and public sectors.
-
A "safeguard" pillar aimed at developing and implementing an ethical charter for responsible AI.
- 11 Alupo, C. D., Omeiza, D., & Vernon, D. (2022). Realizing the Potential of AI in Africa: It All Turn (...)
15At all levels, the challenge for Africa’s education system is to adapt to the new needs of Industry 4.0 by focusing, in primary and secondary education, on science, technology, engineering and mathematics (STEM) and, in secondary and tertiary education, on developing the requisite skills that will enable graduates to adapt quickly to the changing world of AI (Alupo et al., 2022)11. Education can be used to facilitate the development of responsible artificial intelligence in Africa. Through the introduction of ethics courses in university education and capacity building for those involved in AI development.
16Artificial intelligence (AI) is bringing significant social and economic benefits to Africa in a wide range of areas, such as agriculture, health and finance. However, the use of AI on the continent also presents considerable risks. A non-exhaustive list of these risks includes AI and data bias, data privacy, lack of transparency, potential harm to vulnerable populations, lack of adequate infrastructure and resources, and AI used for malicious purposes (misinformation, surveillance or perpetuation of corruption and human rights abuses). In addition, if not thoroughly addressed, these risks can hamper efforts to achieve sustainable development in Africa. This section provides an overview of the risks associated with AI innovation in Africa, including a list of high-profile examples of harmful uses of AI on the continent.
- 12 Belenguer, L. AI bias: exploring discriminatory algorithmic decision-making models and the applicat (...)
- 13 Miller, E., Kishi, R., Raleigh, C. et al. An agenda for addressing bias in conflict data. Sci Data (...)
- 14 Amazon scraps secret AI recruiting tool that showed bias against women (Reuters, 2018). https://www(...)
- 15 Ledford, H. Millions of black people affected by racial bias in health-care algorithms. Nature 574, (...)
17AI and data biases are one of the main challenges facing Africa, particularly in terms of digital innovation. AI bias12 (algorithmic bias in general) are systematic and repeatable errors in a computer system that create "unfair" results, such as "favouring" one category over another in a way that is different from the intended function of the algorithm. Bias in the data13 occurs when the data used to train AI systems (or any other automated system) are not representative of the population they are intended to serve. The two types of bias are generally linked and can lead to inaccurate or unfair results for certain groups of people. For example, Amazon’s AI hiring tool scandal in 201514 revealed gender bias against women when selecting new employees. Another example is the millions of African-Americans subject to racial bias in healthcare algorithms in US hospitals15.
- 16 United Nations Conference on Trade and Development, Report on Data Protection and Privacy Legislati (...)
18Data confidentiality should be a major concern when using AI for innovation. Africa is showing encouraging signs in this direction. According to UNCTAD’s analysis of data protection and privacy legislation worldwide16, 33 out of 54 African countries already have data protection and legislation in place. However, as the ways in which AI uses data become increasingly complex, maintaining robust and up-to-date regulatory frameworks remains an ongoing challenge. In other words, regulatory frameworks must keep pace with the evolution of AI and the use of data. The lack of flexible and robust privacy and data security regulations poses a risk to individuals, as it increases the likelihood that personal data will be mishandled or misused. In addition, this lack of regulation can undermine trust in AI systems and increase reluctance to use them, hampering AI innovation.
19Lack of transparency is another risk associated with the development and deployment of AI systems. As highlighted in the UNESCO Recommendation on the Ethics of AI17 "A lack of transparency could undermine the ability to effectively challenge decisions based on results produced by AI systems, and could thus undermine the right to a fair trial and an effective remedy, and limits the areas in which these systems can be legally used". This is of particular concern when using AI solutions in legal systems, where a lack of transparency can lead to human rights violations, as happened in 2020 with a Dutch AI system used to detect welfare fraud18. In addition, as with data privacy, a lack of transparency can lead to the erosion of trust in AI systems, which can hinder their adoption and impact.
20Most African governments are aware of this risk and recognise the power of open government data to increase transparency and innovation through the development of data-driven public services19.
21A fourth risk is that of harming vulnerable populations. For example, the use of AI in the delivery of social services may lead to the exclusion of marginalised groups such as people living in rural areas, people with disabilities, anyone lacking the necessary digital skills or anyone living in areas lacking digital infrastructure. This risk is strongly linked to the digital divide20, which includes, but is not limited to, unequal access to digital technology, creating division and inequality around access to information and resources. The digital divide is a real obstacle for vulnerable populations. It can perpetuate and amplify existing social inequalities, such as discrimination and prejudice. Hence the need to develop and deploy AI systems designed using a human-centred approach. In Africa, this issue is all the more worrying because the digital divide and the knowledge divide concerning the quality and quantity of AI research are widening between and within countries.21
22Another risk associated with AI innovation in Africa is the lack of adequate infrastructure and resources to support the development and deployment of AI systems. This lack includes technical infrastructure, such as a robust digital network providing access to good quality internet, as well as human resources, such as well-trained educators and professionals in AI and related fields. According to the World Economic Forum’s strategic intelligence tool22, the high cost of data and the limited availability of internet access and finance have hampered business development, contributing to sub-Saharan Africa’s position as one of the least competitive economies. Entrepreneurs in the region face a fragmented business landscape between the formal and informal sectors. They also face the risk of low customer adoption due to persistent digital divides, widening wealth inequalities and limited access to banking services.
- 23 CIPESA Report. Disinformation Pathways and Effects on Democracy and Human Rights in Africa: Case st (...)
- 24 Bulelani J. The Rise of Chinese Surveillance Technology in Africa (part 1 of 6). May 31, 2022, EPIC (...)
23AI systems used for malicious purposes are another significant risk associated with AI innovation in Africa. The best-known example is the use of AI to spread disinformation, which has a significant impact on democracy, political stability, national security and public health. A recent CIPESA report23 explores the nature, perpetrators, pathways and effects of disinformation in Cameroon, Ethiopia, Kenya, Nigeria and Uganda, and shows how contextual similarities have underpinned the proliferation of disinformation. Another relevant example is the use of AI to monitor citizens24, involving negative impacts on human rights and privacy.
24These six types of risk associated with the use of AI and data can hamper the ability to innovate through AI in Africa, diminishing its potential to contribute to the continent’s sustainable development. For example, if data bias is present in a healthcare AI system, it can lead to inaccurate diagnoses for certain groups of patients, hindering reproducibility and obstructing the system’s ability to improve healthcare outcomes. Similarly, a lack of transparency in the development and deployment of AI systems can lead to a lack of trust in the technology, preventing its adoption and effective use.
25Beyond the potential damage, the adverse consequences of these risks have already occurred in Africa. In what follows, we present a short list of harmful examples of the use of AI in Africa.
-
- 25 Bulelani J. The Rise of Chinese Surveillance Technology in Africa (part 1 of 6). May 31, 2022, EPIC (...)
The rise of Chinese surveillance technology in Africa.25
-
- 26 Mapping Disinformation in Africa, Africa Center for Strategic Studies, April 26, 2022. https://afri (...)
Russian disinformation campaigns in Africa.26
-
Gender bias in virtual personal assistants in South Africa.27
-
- 28 Mirca Madianou (2021) Nonhuman humanitarianism: when 'AI for good' can be harmful, Information, Com (...)
Biased chatbots used in the humanitarian sector.28
-
Gender and racial bias in bank lending decisions in South Africa29
26The potential of AI innovation to contribute to sustainable development in Africa and improve the lives of its people is undeniable. However, it is essential to be aware of the risks that can arise from its adoption, such as AI and data bias, data privacy, lack of transparency, potential harm to vulnerable populations, lack of adequate infrastructure and resources, and AI being used for malicious purposes. Governments, businesses and individuals must take the necessary steps to ensure that these risks are addressed and that the potential benefits of AI technology are maximised. In addition, it is essential that AI is used responsibly and ethically, taking into account cultural norms, language barriers and other factors that could lead to negative consequences.
27A multi-stakeholder approach is strongly recommended for AI innovation in Africa, in order to capitalise on the continent’s technological potential and demographic explosion, and contribute to its sustainable development.
28AI innovations are multiplying, with a series of initiatives and projects being developed across the continent. It is therefore necessary to put in place a legal and regulatory framework to guide the development of these innovations and ensure that they comply with local laws and regulations. This section aims to explore the legal and regulatory context for AI innovation in Africa, focusing on the challenges faced, potential solutions and implications for the future of the industry.
- 30 The Digital Transformation Strategy for Africa (2020-2030). The Digital Transformation Strategy for (...)
29Legislation on artificial intelligence (AI) innovation in Africa has not yet received much attention from a national and continental perspective, and is still addressed by sectoral legal instruments, such as privacy laws, data protection laws, cybersecurity laws or ICT legislation. The endorsement of the African Union’s 2020 Digital Transformation Strategy30 by the Heads of State is a crucial step towards harnessing emerging technologies on the continent, including AI.
- 31 Ibid
- 32 AUDA-NEPAD. Agenda 2063: Framework document: Auda-NEPAD, NEPAD Home. Available at: https://www.nepa (...)
30The strategy pays particular attention to AI and its links with other emerging technologies (Internet of Things, machine-to-machine communication, 5G, etc.)31, and aims to ensure that policymakers and regulators keep pace with technological advances in order to address new regulatory frontiers and create a solid foundation for digital transformation to reach its full potential. These efforts aim to promote digital skills and capabilities, digital governance, digital infrastructure and digital entrepreneurship. They also aim to foster the innovation needed to create jobs, support startups in line with Agenda 206332, ensure an inclusive digital transformation and accelerate Africa’s socio-economic development.
31The African Union has made great strides in data governance, notably with the adoption of the AU Data Policy Framework in 202233. However, the Malabo Convention on Data Protection and Cybersecurity has still not been ratified and implemented since 2014. As of May 2022, 33 out of 55 African countries have data governance laws, and 5 countries had draft laws34.
32At the national level, other AI governance initiatives have been implemented by African states. These include the drafting of national AI strategies (Egypt and Rwanda), the establishment of regulatory sandboxes (Kenya and Ghana) and the formation of government taskforces to investigate the potential of emerging technologies (Blockchain and AI Taskforce in Kenya and CBDC Taskforce on Digital Currencies in Tanzania). In addition, Botswana, Egypt, Mauritius, Tunisia and Zambia have adopted national AI legislative strategies.
- 35 Kenya Law (2019) Kenya Data Protection Act, Kenya Data Protection Act No. 24 of 2019. Available at: (...)
33Although some of these laws do not apply exclusively to AI, they are no less useful. For example, Kenya’s Data Protection Act35 stipulates limits on how companies can use technology to process, store and transfer personal data. Similarly, South Africa’s Personal Information Protection Act (POPIA) regulates automated decision-making and algorithmic bias.
- 36 Onuoha, R. (2019) AI in Africa: Regional data protection and privacy policy harmonisation, Global I (...)
34While this is a positive step, some of the stipulations in these laws could restrict AI innovation and development in different regions. For example, data localisation laws prevent the commercial use and movement of data needed to create AI on the continent. Therefore, the trend is towards a more unified privacy and data protection framework to better meet the cross-border needs of AI deployment36.
- 37 Hart, R.D. (2018) Saudi Arabia's robot citizen is eroding human rights, Quartz. Available at: https (...)
35Most African countries are finding it difficult to develop laws and regulations relating to artificial intelligence (AI). This complexity is due to the rapidly evolving uses of the technology, such as generative AI in the creative sector, which are subject to intellectual property laws. Questions about the rights of robots, such as the Sophia robot in Saudi Arabia37, further complicate the task of balancing human rights and technological progress.
36Regulators must take into account the power asymmetries between consumers and large technology companies, and support local artificial intelligence start-ups working with data largely controlled by large technology companies. African competition laws still need to evolve to respond effectively to the extractive and monopolistic practices of digital platforms.
37Despite these challenges, the African Union High Level Panel on Emerging Technologies has recommended that AI should be a priority for Africa’s socio-economic development38.
38Ideally, the African Union should develop a continental strategy on artificial intelligence (AI) to enable African countries to improve policy development and implementation, as well as stakeholder engagement on AI challenges and opportunities. The strategy would enable countries to create AI-related technology products and services and model data security and regulatory frameworks. It would also facilitate stakeholder engagement, increase capacity building and strengthen the regulatory environment for AI technology and data management.
39This continental strategy should address existing and systematic implementation challenges faced by African AI innovators, and provide guidance to generate knowledge products that address myths and misconceptions about AI adoption in Africa. The strategy should be developed in consultation with AI experts and stakeholders across the continent, and should integrate existing national AI strategies and consolidate their best practices for the benefit of the continent.
- 39 AUDA-NEPAD (no date) The African Union Artificial Intelligence Continental Strategy for Africa: Aud (...)
- 40 Adams, R. (2022) AI in Africa: Key Concerns and Policy Considerations for the Future of the Contine (...)
40In addition, there is an urgent need for an Africa-focused repository and networking platform to support the AI expert community. This platform should promote local AI capabilities and encourage the use of locally produced African AI products39. This is critical, as continued reliance on foreign AI technologies may result in technologies and solutions on the continent that are not aligned with the region’s development priorities40.
41This platform should provide an enabling environment that facilitates the compliance of market-based AI policies with regulatory processes. To this end, policies and frameworks should be developed to increase data transparency and facilitate data sharing. In addition, collaboration in AI-related economic activities in Africa is necessary to avoid "neocolonialism"41.
42An African AI strategy should lay the foundations for such a platform, increasing the possibilities for the continent to capitalise on the opportunities offered by Industry 4.0.
- 42 Baijnath, M. and Neil Butcher and Associates (2021) The Growth of Artificial Intelligence in Africa (...)
- 43 Ibid
- 44 Ibid
43Overall, African society and businesses can benefit from diversification that includes people from different socio-economic, racial, ethnic and cultural backgrounds42. This is essential to improve the quality of data sets, provide accurate results and mitigate bias43. In addition, it is important to recognise the different perspectives and contributions that can be made in this sector44.
44African countries need to invest in human capital and ensure that AI knowledge is integrated into their educational curricula so that all young people and all economic sectors can benefit. To further facilitate the development of AI, continental, regional and national institutes and innovation ecosystems should be established. However, data protection and management must be improved and awareness of their importance raised.
45With regard to AI regulation, countries that do not have laws in place should consider developing them. These laws should be flexible, principle-based and written in plain language so that individuals and organisations can understand and comply with them. They should also take into account advances in technology and human rights.
46Collectively, efforts should be made to promote AI within the African Union’s regional economic blocs, through the development of digital policies, conversations and infrastructure.
47AI innovation in Africa offers a unique opportunity to develop technological solutions that have a significant impact on the lives of African citizens. However, caution must be exercised in the face of potential risks and challenges. To take full advantage of the potential of AI, public policies must be designed to capitalise on the existing strengths of African countries and address their current weaknesses. Overall, a multi-stakeholder approach to AI innovation in Africa is strongly recommended.
48AI innovation in Africa offers the opportunity to create technological solutions that can have a positive impact on the lives of African citizens. Here are some examples of areas already touched by significant development and positive impact:
-
Education,
-
Health care,
-
Financial Services, and
-
Agriculture.
49For these AI innovations in Africa to be beneficial, however, it is essential to address potential risks, such as:
-
AI and data distortion,
-
Data confidentiality,
-
Lack of transparency,
-
The potential harm to vulnerable populations,
-
Lack of infrastructure and adequate resources, and
-
AI used for malicious purposes.
50Collaboration between governments, private companies, civil society organisations and other stakeholders is needed to promote responsible and ethical innovation in line with existing legal frameworks at national and regional levels. Capacity-building initiatives are also needed to enable African countries to design and implement effective AI regulation that takes into account the changing nature of the technology and the markets in which it is used. These policies should focus on the following:
-
Promoting and retaining AI talent
-
Developing digital infrastructure and computing power
-
Promoting the governance of AI and data
-
Stimulating AI research, development and innovation
-
Encouraging the adoption of AI in the public and private sectors.
51Ultimately, AI innovation in Africa offers the opportunity to create technological solutions that can positively impact the lives of African citizens, but potential risks such as AI and data bias, data privacy and lack of transparency need to be addressed. Collaboration between governments, private companies, civil society organisations and other stakeholders is needed to promote responsible and ethical innovation, and capacity-building initiatives must be implemented to enable African countries to design and implement effective AI regulation.