How AI is impacting policy processes and outcomes in Africa

While it has been in existence for almost 70 years, artificial intelligence (AI) has re-emerged over the past decade, powered by advances in deep learning and computing architecture to complete tasks such as object recognition, predictive modeling, text, image, and video generation, and decision-making. The emergence of popular commercial tools, such as ChatGPT, developed by OpenAI in late 2022, has led to the rising interest of companies, startups, and institutions in adopting and developing AI. However, given the negative impacts of AI in fields such as healthcare, education, and policing, countries have also been rapidly increasing their respective capabilities to effectively govern AI. In 2017, Canada released the world’s first national AI strategy, and since then, tens of countries have followed, including the United Kingdom, Egypt, Mauritius, Singapore, and Brazil.

With the European Union being the only regional body having enacted formal AI regulation, there is much room for countries around the world to begin implementing formal governance mechanisms to ensure that AI is developed and implemented in responsible ways. However, given the concentration of AI development in countries such as the United States, the United Kingdom, and China, concerns that Africa may fall behind in regulating AI abound, given existing challenges in data protections across the continent. Data governance has formed the foundation for many national AI policies, and the EU’s General Data Protection Regulation (GDPR) can arguably be seen as a mechanism that aided the successful development and enactment of the EU AI Act. Given that 36 of 54 African countries have enacted formal data protection regulations and the African Union (AU) recently ratified the Malabo Convention, a legal framework for data protection and cybersecurity, in June 2023, such regulation could help bolster effective AI regulation throughout the continent, as framed in a piece published by the Brookings Institution.

However, African governments have notable challenges, primarily surrounding addressing formal development concerns like effective education, healthcare, infrastructure, and skilled technical workforces. However, these issues will have to be addressed in tandem with enacting formal AI regulation to preempt harms to citizens and hold companies accountable for deleterious AI systems. Additionally, the concerning adoption of digital surveillance technologies by African governments and the increasing outsourcing of data annotation and content moderation labor to countries in East Africa have shifted priorities for AI and data governance within Africa. In light of these considerations, establishing robust AI governance measures across the African continent is necessary to implement effective safeguards.

African consumers, educational institutions, governments, and companies are rapidly adopting AI to aid in content creation, improve the delivery of public services, and streamline business processes. While there is little information available about consumer adoption of AI tools across the continent, it has rapidly increased due to the introduction of ChatGPT, DALL-E, Midjourney, and other commercial AI tools. Research from the 2024 Stanford AI Index shows that 27% of Kenyans use ChatGPT daily, coming third behind India and Pakistan. Google search trends also reveal rising consumer interest in AI across Africa, with searches related to AI rising 270% over the last year and 400% over the last five years in countries such as Kenya. African educational institutions, such as the University of Pretoria in South Africa, Makerere University in Uganda, and Kwame Nkrumah University of Science and Technology in Ghana, have developed prominent AI labs and published research focused on leveraging AI for social impact. The African Institute for Mathematical Sciences (AIMS), launched in 2003 and with educational centers in Cameroon, Ghana, Senegal, South Africa, and Rwanda, has established a number of Master’s and PhD programs in AI, machine learning, math, and data science, which have graduated over 3000 students. Additionally, large tech companies have developed AI research labs in Africa, with Microsoft establishing in 2020 the Microsoft Africa Research Institute in Nairobi, Kenya, and in 2018, Google launching an AI research lab in Accra, Ghana, and in 2013, IBM Research launching labs in Nairobi, Kenya and in Johannesburg in 2016.

Over the past few years, the AI ecosystem within Africa has rapidly expanded with a growing number of startups focusing on developing AI applications for a variety of use cases. These startups are spread across the continent, with Intron Health in Nigeria developing natural language processing tools to understand African accents in clinical settings, minoHealth AI Labs in Ghana developing AI systems to diagnose 14 chest conditions, iCog Labs in Ethiopia developing general-purpose AI solutions, including an Amharic-speaking robot, and Lelapa AI in South Africa working on building natural language model solutions for African languages. Many efforts in bolstering the AI ecosystem within Africa started with the establishment of local, grassroots communities such as Deep Learning Indaba, AI Saturdays Lagos, Data Science Africa, Ghana NLP, and Masakhane, whose founders spun off Lelapa AI from the organization’s significant efforts. These groups have been essential in training the next generation of AI researchers in Africa, providing community, hosting conferences and workshops, and publishing research in premier machine learning and AI venues.

Policy conversations on AI are gaining traction across Africa. These conversations can be understood in two parts: the basis for national AI strategy or regulations and the policy or governance changes linked to AI. Regarding AI regulations, some arguments favor implementing appropriate regulations continent-wide, while others advocate for enhancing AI literacy to ensure its effective utilization. Yet, African countries can simultaneously adopt these perspectives in AI-oriented policy initiatives.

Beyond the Malabo Convention, which serves as the benchmark for AI policy in Africa, the African Union Development Agency produced a draft policy that addresses AI legislation for African nations. African leaders are expected to endorse the continental AI strategy by February 2025. Countries lacking AI policies would utilize this framework to formulate their strategy, while those with existing policies may be required to align them with the AU’s guidelines. So far, only seven African countries (Benin, Egypt, Ghana, Mauritius, Rwanda, Senegal, and Tunisia) have developed national AI programs with varying levels of implementation. Rwanda, for example, has AI-specific investment programs such as Rwanda’s Seed Investment Fund, which was established to foster an environment where the government can co-invest with angel and venture capital investors in AI startups. Egypt’s AI Strategy aims to enhance funding for AI start-ups, establish AI startup incubators, and incentivize enterprises to buy AI products locally rather than import them.

Amid this backdrop, some African countries already use AI to address development concerns without a clear strategy, potentially determining their stance on AI policy development. AI influences policy and governance outcomes in some African countries through various adopted initiatives to bolster social and economic growth. For example, Morocco is harnessing the potential of AI, having recognized its benefits in building small businesses, startups, and service outsourcing. To this end, the government has established specific research and development centres in various towns, including Oujda, Rabat, Fes, and Benguerir. AI has been used in Togo to target poor cantons for social funds and in Zambia to combat misinformation during elections. Notably, some of these activities are stimulated by the assistance of international partners.

AI has great potential for good within Africa, but it also carries tremendous risks. Sadly, AI risks may disproportionately impact those least equipped to mitigate them, particularly in Africa. Studies have shown that AI can perpetuate biases, exacerbate injustices, and violate human rights. AI has negative implications for transnational organized crime, as criminals leverage this technology to attack personal datasets and applications and impersonate to extort or defraud. In South Africa, impersonation fraud increased by 356% between April 2022 and April 2023. AI has the potential to influence public opinion by spreading misinformation or disinformation, which is a soft power. For example, a leaked audio that turned out to be a deepfake was widely distributed over WhatsApp, Facebook, and other social media sites during Nigeria’s 2023 general elections. The fake audio purported to depict a secret conversation between leaders of the biggest opposition party, the People’s Democratic Party (PDP), planning to compromise elections. The rise of sophisticated deep fake technologies and massive language models raises socio-technical problems about democratic elections.

As African countries work to build adequate AI policies, proactive steps to mitigate AI concerns remain in their early phases. Nonetheless, many African countries would likely prioritize keeping up with AI breakthroughs and maximizing its potential use. This may mask possible threats and delay solutions to future difficulties.

There are several ways in which African countries can design and adopt their own governance approaches when addressing the current and future roles of AI, including:

• Although African countries do not need to wait to implement AI strategies to capitalize on their potential, they should prioritize the development of comprehensive national AI strategies that include regulations and investment opportunities. These countries should acknowledge country-specific peculiarities while aligning with existing continental and global policies.
• AI-related policy activities and interventions should be integrated into more significant initiatives focused on privacy, security, data access, intellectual property protection, human rights, and cross-border data interchange. A concerted effort by diverse stakeholders is necessary to connect these policy nodes and weave them into a coherent action plan for policymakers.
• African governments should prioritise partnership and knowledge-sharing opportunities to boost the continent’s AI initiatives. This involves working with other governments, regional organisations, academic institutions, and industry players to share AI governance best practices, resources, and expertise.
• African governments and partners should begin to identify, amplify, and mitigate the potential risks of AI. This can occur through prioritizing ethical AI education and increasing awareness of AI harms. This includes integrating AI ethics and responsible AI principles into educational curricula and raising public awareness about the implications of AI technologies. Given the evolving nature of AI, governments and partners should commit to continuous research to track emerging threats and provide early warning support.

These recommendations are initial starts into what promises to be a robust and evolving debate.

The evolution of AI poses opportunities and difficulties for Africa. The youthful and enthusiastic African population is a valuable opportunity to harness the full potential of AI in driving socioeconomic development. As African countries manage the intricacies of AI policies, regional and international collaboration will be critical to establishing a long-term and equitable AI ecosystem. Africa can capitalize on AI’s transformative potential while mitigating adverse outcomes by addressing concerns such as data governance and ethical considerations. In this shifting climate, proactive engagement with stakeholders such as governments, industry, academia, and civil society will be important in crafting a future in which artificial intelligence contributes positively to Africa’s development aspirations.