Ethical Horizons - Mapping AI Policy in Africa

Ethical Horizons

Mapping AI Policy in Africa May, 2024

1

Ethical Horizons - Mapping AI Policy in Africa May, 2024

Funded by: Bill and Melinda Gates Foundation (BMGF). Lead and Coordinated by: Nanko Madu, Lydia Ezenwa, Philip Adebayo, and Timothy Mugume AI Policy Experts: Ms. Lavina Ramkissoon, Mr. Serge Ntamack Anglophone Lead AI Researcher and Report Writer: Dr. Ing. Judith Leo

AI Data Analysts: Eng. Samiiha Nalwooga

2

Ethical Horizons - Mapping AI Policy in Africa May, 2024

Foreword As the world stands at the precipice of a technological revolution driven by artificial intelligence (AI), the ethical implications of this transformative force have never been more pressing. In the heart of this global conversation, Africa emerges not only as a participant but as a pivotal protagonist in shaping the ethical contours of AI adoption and governance. It is with great anticipation and profound optimism that the Bill and Melinda Gates Foundation, in collaboration with AfriLabs, presents this ground breaking initiative. Therefore, the mandate is to impact the continent over the next decade through four Strategic Pillars one of which is Research, Evidence, and Learning. This pillar builds on innovative ways to move the continent forward through collaboration and research on what works, where it works, and how it works, and dissemination of learning across Africa and the wider innovation ecosystem, network, and experience gained so far with over 470 innovation centres across 227 cities and 52 African countries is conducting a project to understand the status of AI Ethical Policies in terms of development and application stages and their implications to innovators and entrepreneurs in African countries. This forward-thinking collaborative research seeks to do more than simply assess the current state of affairs; it aspires to catalyse meaningful change. By mapping the status of AI ethical policies and examining their impact on African innovators and entrepreneurs. The study aims to provide a roadmap for excellence in AI adoption, development, and use —one that prioritises national values, social-economic growth and sustainable development.

3

Ethical Horizons - Mapping AI Policy in Africa May, 2024

Executive Summary

Artificial intelligence (AI) is a rapidly growing sector within the African innovation ecosystem. Amidst rapid technological advancements and growth in the development and application of AI solutions in African countries, it has been noted that AI technologies pose ethical quandaries like privacy violations, algorithmic bias, and safety concerns. As a result, governments, organizations, and institutions are drafting or adopting AI ethical policies to guarantee responsible and moral development and application of AI technologies. African nations such as Tunisia, Egypt, and Nigeria have made the initial moves to establish governance and policies to direct the development and application of AI in their countries, while other nations are still lagging. Thus, there is a need to explore the sentiments on AI ethical policies of innovators and entrepreneurs in Africa, as well as the opportunities and difficulties they encounter when implementing ethical AI systems.

The goal is to foster excellence in AI ethical policy adoption, development, and use in Africa. The study utilized both qualitative and quantitative research methods with the use of proprietary vocabulary to discover and extract quality information from the literature, and from AfriLabs’ enormous datasets of African policymakers, Governments, and innovation ecosystems. A snowballing approach was utilized to increase the coverage and diversity of the collected data and its stakeholders. The following data collection methods were deployed; desk research of stakeholders by country, review of AI stakeholders’ engagement through key information interviews and focus group discussions, case studies, and online surveys. A webinar session was used to discuss findings and get feedback and inputs from relevant stakeholders and contributors in the ecosystem. A total of 300 publications/ resources and 165 participants contributed in the secondary and primary data collection respectively. The result shows there is a small number of about 10% of the collected responses indicating the adoption, development, and use of AI ethical policies in African countries are in the initial stage. The major challenges that hinder the adoption and development of AI ethical policies include limited understanding of AI, funding, lack of access to data, inadequate infrastructure such as internet connectivity, and skills shortage. The study recommends AI startups enhance collaborative networks for resource sharing within the AI stakeholder’s ecosystem, and development of adequate infrastructure that will enhance the adoption of AI ethical policies.

This project aims to:

1.Map and identify AI ethical policies in Africa

2.Examine sentiments about AI eth- ics among African AI practitioners

3.Analyse implications of these policies on African innovators and entrepreneurs. 4.Provide recommendations for promoting, complying with, and leveraging these AI ethical policies

4

Ethical Horizons - Mapping AI Policy in Africa May, 2024

Table of Contents

Executive Summary

4

List of Figures

6

1.0 Introduction

7

2.0 Rationale of the Study

8

3.0 Key Definition and Methodology 3.1 Key Definition 3.1.1 AI Stakeholders 3.1.2 AI Strategy and Policy 3.2 Methodology

10 10 11 13

4.0 Results and Discussion 4.1 Results 4.1.1 Mapping the Landscape of AI Ethical Policies in Africa

14 14 14 15 16 18 24 26 26 26

4.1.2 Case Study Analysis on AI Ethical Policies Landscape in Africa 4.1.3 Perspective from AI Stakeholders on AI Ethical Policies in Africa 4.1.4 Mapping Impact of AI Ethical Policies in Africa 4.2 Discussion

5.0 Conclusion and Recommendations 5.1 Conclusion 5.2 Recommendations

6.0 Limitation of the Study

29

7.0 Data Availability Statement

29

8.0 Funding

29

9.0 Conflict of Interest

29

Reference

30

5

Ethical Horizons - Mapping AI Policy in Africa May, 2024

List of Figures

The following Table 1 is the definition of terminologies that will be used in the study;

6

Ethical Horizons - Mapping AI Policy in Africa May, 2024

1.0 Introduction

Artificial Intelligence (AI) has been described as a general-purpose technology with potentially far-reaching societal impacts (Rogerson et al. 2022)(Eke 2023). It is a rapidly growing sector within the African innovation ecosystem and many African countries are pinning hopes on AI to help accelerate growth, productivity, and competitiveness, leading to achieve the Sustainable Development Goals (s) (Sampene et al.2022; Sedola, Pescino, and Greene 2021)(ITU and UNDP 2023). As AI continues to advance at a rapid pace globally (Law 2023), it becomes imperative to ensure that its development and deployment adhere to robust ethical policies that set standards to safeguard the rights and dignity of individuals and communities (UNESCO 2022; Jaldi 2023; Brief 2023). By definition, AI ethical policies and regulations serve as essential guardrails in guiding the responsible development and use of AI innovations and solutions. These policies encompass principles such as transparency, fairness, accountability, privacy protection, and mitigation of bias (UNESCO 2022; Gwagwa, Kachidza, et al. 2021). In the African context, where diverse cultural, social, and economic factors shape technological adoption and impact, the formulation and implementation of effective ethical policies and regulations are paramount.

It has been noted that AI technologies pose ethical quandaries like privacy violations, algorithmic bias, and safety concerns (Okolo, Aruleba, and Obaido 2023). Hence, in recent years, the concept of AI ethical policy has gained significant traction worldwide, as the ethical implications of AI technologies have become increasingly apparent (Jaldi 2023). As a result, governments, organizations, and institutions are drafting or adopting AI ethical policies to guarantee responsible development and application of AI technologies African nations such as Tunisia, Egypt, and Nigeria have made the initial moves to establish governance and policies frameworks to direct the development and application of AI while other nations are still lagging (Chatbri, Jemmali, and Hannachi 2019; Demaidi 2023).The main objective of this project is to map and identify AI ethical policies being developed and applied across Africa by governments, examining the sentiments among practitioners, the beneficial and detrimental implications that these policies may have on innovators and entrepreneurs, and provide actionable recommendations to promote, comply with and leverage these policies . The report consists of five sections : introduction, literature review, methodology, results and discussion, and lastly, recommendations and conclusion.

7

Ethical Horizons - Mapping AI Policy in Africa May, 2024

2.0 Rationale of the Study

At its core, AI ethics pertains to the moral principles and guidelines governing the development, deployment, and use of AI systems (Dwi Hadya Jayani 2019; Eke 2023). These principles encompass a wide range of considerations, including transparency, fairness, accountability, privacy protection, and the mitigation of bias (Gwagwa, Kachidza, et al. 2021). The importance of AI ethics cannot be overstated, particularly in the African context, where the rapid proliferation of AI technologies intersects with unique socio-cultural, economic, and political dynamics (Gwagwa, Kachidza, et al. 2021). Ethical considerations are essential to ensure that AI technologies are developed and utilised in a manner that promotes the well- being and empowerment of all individuals and communities across the continent (Brief 2023). By adhering to ethical principles, stakeholders can mitigate potential harms associated with AI, safeguard fundamental rights, and foster trust in these transformative technologies (Demaidi 2023). This study aims to examine existing ethical frameworks and regulatory approaches related to AI in Africa, identifying gaps, challenges, and opportunities for improvement. Key objectives include assessing the adequacy of current regulatory mechanisms in addressing emerging ethical challenges posed by AI.

Additionally, the study aims to explore avenues for enhancing multi-stakeholder collaboration and capacity-building efforts to promote ethical AI practices across the continent. It also analyses the intersection of AI with relevant guidelines and recommendations for ethical AI governance with the goal to shape AI’s future in Africa, ensuring it is ethical, innovative, rights- protective, and inclusive. In the global context of AI ethics, diverse ethical frameworks and principles have emerged to guide the responsible development and deployment of AI technologies (Ndzendze and Marwala 2023). For example, the UK’s Bletchley Declaration (UK Government 2023), inspired by the historic Bletchley Park, emphasises the importance of transparency, fairness, accountability, and inclusivity in AI governance and the United Nations through the United Nations Centre for Artificial Intelligence and Robotics (UNICRI) initiative, have aimed to promote human-centric AI systems that respect fundamental rights and dignity (International Telecommunication Union 2021). Other countries such as India and Canada, have pioneered approaches to AI ethics, reflecting regional nuances and priorities (Eke 2023). India’s National Strategy for AI underscores the need for ethical, transparent, and accountable systems to foster socio-economic growth

8

Ethical Horizons - Mapping AI Policy in Africa May, 2024

and inclusive development, while Canada’s AI Strategy emphasises human-centric principles and the importance of diversity and inclusion in AI innovation (Pizzi, Romanoff, and Engelhardt 2020; Saheb 2024; World Health Organization 2024; Bhalla, Brooks, and Leach 2023). The principles of EU AI Act weight the advantages of AI innovation against possible concerns, assuring ethical and responsible deployment across the EU. It aims to advance the EU’s values and AI governance principles while also encouraging innovation and competitiveness in the global AI landscape. It balances risk and innovation, supporting AI technology development while minimising any risks associated with AI systems. The EU AI Act addresses both elements in the following ways (Parlamento Europeu 2024; Meltzer and Tielemans 2022; Friday and Act 2023) (i) Risk-Focused Approach, which prioritises high-risk AI systems that may jeopardise safety, basic rights, or social values. These high-risk systems must adhere to stricter criteria, which include detailed risk assessments, transparency obligations, human monitoring, and documentation/reporting requirements; the legislation prohibits some AI approaches that damage individuals or society. For example, it prohibits AI systems from manipulating human behaviour, exploiting vulnerabilities, participating in social scoring, or evading legal obligations; and stresses on the point that the high-risk AI systems require rigorous conformity evaluation to fulfil regulatory criteria. This includes risk assessments, mitigation techniques, certification, and ongoing compliance. Aspects which encourage trust and openness in AI systems by providing clear user information, explainable AI judgments, and accountability procedures for accidents or harm. The Act fosters responsible AI development that respects ethical principles, societal values, and fundamental rights; and promotes international cooperation and aligns AI regulations with partners globally. (ii) Innovation-Focused

The US Algorithmic Accountability Act (US AAA) of 2022 takes a risk-based strategy to address algorithm-related concerns and their societal consequences. It aims to balance the benefits and risks of automated decision systems (ADS) by requiring organizations to conduct impact evaluations before deployment and enhance decision-making procedures post-implementation (Mökander et al. 2022). Hence, the Act has created a governance architecture that holds bad actors responsible while empowering good players to show ethical, legal, and safe ADS. Collectively, these global efforts underscore the imperative of developing ethical frameworks that prioritise human values and rights in the rapidly evolving landscape of artificial intelligence. Additionally, the Governmentt of China make AI work for economic development, for welfare of humanity in line with ethics while ensuring safety, fairness, promoting robust standards and regulations to contain perceived social, political, and security and data protection or personal information protection risks (Lin et al. 2024). AcrossAfrica,variouscountrieshavemadestrides in developing ethical frameworks and principles to guide AI governance (Law 2023). While the approaches may vary, common themes include transparency, fairness, accountability, and human rights protection. Some countries have conducted extensive research and gathered data on ethical considerations related to AI, while others have formulated comprehensive strategies or policies to address the need and adoption of AI ethical policy (Dwi Hadya Jayani 2019)(Broadridge 2019). It is important to realize that recognized frameworks such as the EU AI Act might provide useful inspiration for African countries to benefit from getting insights and adopt some of the AI policies to fit African’s settings, values and needs. Therefore, the ultimate motive should be to adopt the existing legislation that supports responsible AI innovation, safeguard African societal and cultural interests, and contribute to long-term growth. However, African countries need to continue refining their ethical frameworks and policies in response to evolving technological landscapes and their peculiar societal needs. Collaboration between governments, industry stakeholders, civil society organisations, and academia is crucial to ensure that AI technologies are developed and deployed in a manner that aligns with ethical principles and fosters inclusive and sustainable development across the continent.

9

Ethical Horizons - Mapping AI Policy in Africa May, 2024

3.0 Key Definition and Methodology

3.1 Key Definition 3.1.1 AI Stakeholders

AI Stakeholder refers to an individual, group of people, or company that has invested in AI, either in the form of direct support for research and development or a vested interest in the success of the AI, or who is affected by or has an interest in AI and its social impact initiative (Dwi Hadya Jayani 2019; Gwagwa, Kachidza, et al. 2021). They may include users, community members, partners, funders, investors, civil society organization, policymakers, regulators, researchers, and others (Sinde et al. 2023). The stakeholders will be categorized into four categories which include firms, academia, civil society, and government. Additionally, these stakeholders are in different stages for AI ethical policy adoption and implementation, such as the initial stage, intermediate stage, and final stage, as briefly described in Figure 1. This study has considered stakeholders in all three stages to understand the status of AI ethical policy at a country level and at the African continent level.

10

Ethical Horizons - Mapping AI Policy in Africa May, 2024

Figure 1: Description of AI Stakeholders and their Stages.

FIRMS Encompasses companies in the private sector engaged in the AI domain, varying from emerging Al companies to

CIVIL SOCIETY Encompasses civil society organization in the Al domain

GOVERNMENT Encompasses entities such as policy-making bodies, regulatory agencies, and organizaions withing the public sectors.

multinational corporations.

Initial Stage Is a stage at which African country in the initial stage of collecting information that will lead to start writing the Al strategies.

Intermediate Stage Is a stage at which African country have already started writing or validation the Al strategies.

Final Stage Is a stage at which African country has already implemented and adopted the Al policy or Al strategy or data protection law and Act which is linked to Al.

3.1.2 AI Strategy and Policy AI Strategy outlines the overarching goals, objectives, and approaches that an organization or entity intends to pursue to achieve its mission and vision through the application and use of AI (Eke 2023; Demaidi 2023). It sets out the direction and priorities for the organization, detailing the strategies and tactics that will be employed to accomplish its desired outcomes (Dwi Hadya Jayani 2019). The purpose of a AI strategy document is to provide a roadmap for organizational growth, competitiveness, and sustainability over a defined period (Adams 2022; Demaidi 2023; van Buren, Chew, and Eggers 2020). AI Policy outlines the principles, rules, and guidelines that govern decision-making and actions within an organization or institution through the application and use of AI (Adams 2022). It provides a framework for addressing specific issues or achieving particular objectives. Policy documents typically define the organization’s stance on various matters, establish procedures for implementation, and outline responsibilities for compliance (Bhalla, Brooks, and Leach 2023; Saheb 2024). The segmentation in

policymaking is critical in understanding the flow of actions to come, which generally start from strategy to policy and its enforcement. As indicated, the purpose of the AI Ethics Policy is to guide AI stakeholders in the development, deployment, and use of AI technologies in an ethical and responsible manner. The policy outlines the principles that the stakeholders must follow to ensure that AI systems are designed and used in a way that respects human rights, promotes fairness, and enhances the well-being of all stakeholders. These policies generally include (Sedola, Pescino, and Greene 2021)(Arakpogun et al. 2021)(Kaack et al. 2022) (Gwagwa, Kazim, et al. 2021);

11

Ethical Horizons - Mapping AI Policy in Africa May, 2024

(i) Respect for Human Rights: AI stakeholders are committed to respecting and upholding human rights in all AI-related activities. AI systems must be designed and used in a way that respects the rights to privacy, non-discrimination, and freedom of expression, among others

(ii) Fairness and Non- Discrimination: AI systems must be developed and used in a manner that is fair, unbiased, and inclusive. AI stakeholders should actively work to identify and mitigate biases in AI systems, ensuring that they do not perpetuate or exacerbate existing inequalities.

(iii) Privacy and Data Protection: AI stakeholders should prioritize the privacy and data protection of individuals when designing and using AI systems. The AI stakeholders should comply with all applicable data protection and privacy laws and adhere to industry best practices for data handling and storage.

(iv) and Explainability: AI systems used by AI stakeholders must be transparent in their operation, and the organisation must provide clear explanations of the logic and reasoning behind AI-generated outputs. For example, companies should maintain a centralized system for AI governance and compliance efforts that provides comprehensive Transparency transparency across the organization of proposed and active AI efforts. Also, they should ensure that the organization is trained and has necessary access to the system for their ongoing work and reporting. Stakeholders should be able to understand how AI systems work and the basis for the decisions they produce.

(v) Accountability and Re- sponsibility: AI stakeholders should hold themselves ac- countable for the ethical use of AI and the outcomes gen- erated by AI systems. The or- ganisation is responsible for ensuring that AI technologies align with this policy and for addressing any adverse im- pacts resulting from their use.

(vi) Safety and Security: AI systems must be designed, developed, and used with a focus on safety and security. AI stakeholders should prior- itize the protection of indi- viduals and society from po- tential harm and will work to minimize risks associated with AI technologies.

(vii) Human-Centered AI : AI systems should be designed to augment and empower human decision-making, not replace it. AI stakeholders should promote human-AI collaboration and ensure that humans retain control over AI systems and their outputs.

(viii) Environmental and Social Sustainability: AI stakeholders should strive to develop and use AI technologies in a manner that promotes environmental and social sustainability, minimizing negative impacts on the environment and communities.

12

Ethical Horizons - Mapping AI Policy in Africa May, 2024 Ethical Horizons - Mapping AI Policy in Africa May, 2024

3.2 Methodology The methodology of this study was guided by the vision to map AI ethical policies and its implication on African innovators and entrepreneurs with a focus on wide and diverse coverage in both rural and urban areas, and inclusivity in the AI innovation ecosystem. The methodology deployed both qualitative and quantitative research methods with the utilization of proprietary vocabulary to discover and extract quality information from the literature and AfriLabs’ enormous datasets of African policymakers, Governments, community service, and public sectors, and also, related stakeholders with a snowballing approach to

have wide and diverse coverage of the data. Throughout the process, iterative methods were used in each stage to review and incorporate missing and new required information and also, examine thoroughly their valuable insights on the challenges, opportunities, and impacts on African innovators and entrepreneurs leading to serve as a resource to inform and provide actionable recommendations that promote excellence in the adoption and use of AI ethical policy leading to the social-economic growth and sustainable development in Africa. The following Figure 2 briefly describes the methods that will be applied towards fulfilling the vision of this project;

Figure 2: Description of Methodology for Mapping AI Ethical Policy in Africa.

01

Initial Preparation

Data Reporting and Dissemination

02

06

Secondary Data Collection

Report data using tools such as PowerBI and Tableau & Data dissemination to AI-based AfriLabs Network, & African and Global AI Conferences and Journals.

Iterative

Primary Data Collection

Data Analysis and Synthesis

Engage African Al Stakeholders through the use of Key Informant Interviews (KIls), Focus Group Discussion (FDGs), Case Studies, Startups and Hubs Referral, Surveys and WebinarSessions.

Conduct qualitative and quantitative analysis using tools such as PowerBI, Python, and DAX to analyse and synthesize the clean collected data.

03

05

Data Cleaning and Annotation

Data cleaning and annotation through the use of tools such as Python and Excel.

04

13

Ethical Horizons - Mapping AI Policy in Africa May, 2024

4.0 Results and Discussion

4.1 Results 4.1.1 Mapping the Landscape of AI Ethical Policies in Africa Mapping the current status of AI ethical policies is an important step towards identifying and examining the sentiments about AI ethics and policies being developed and applied among African AI practitioners, as well as examining the beneficial and detrimental implications that these policies may have on innovators and entrepreneurs. Then, provide actionable recommendations that promote excellence in the adoption, development, and use of AI ethical policy for African innovators and identify the strategies that policymakers, innovators and entrepreneurs can use to promote, comply with and leverage these AI ethical policies in Africa. There are several efforts that have been made so far in the continent to develop regulatory frameworks, strategic initiatives, and policies. However, the engagement with AI governance is multifaceted, reflecting the varying degrees of technological advancement, regulatory capacity, and socio-economic priorities across

different countries and at pan African level. These efforts include; i. In 2018, the African Development Bank reviewed the implication of emerging technologies and then, recommended the need of a coordinated regional vision to seize the opportunities provided by emerging technologies by strengthening human resources capacity, develop adaptive and collaborative regulations and nurture the adoption of emerging technologies in the key developmental sectors (African Development Bank 2018). ii. (In 2021, the African Commission on Human and Peoples’ Rights (ACHPR) directed member states to ensure that the development and use of AI, robotics, and other new and emerging technologies are compatible with the rights enshrined in the African Charter, particularly the human rights principle (Brief 2023; Law 2023). iii. The Smart Africa Alliance, together with several partners, has also developed an AI for Africa Blueprint which emphasizes human capital development (UNESCO 2022).

14

Ethical Horizons - Mapping AI Policy in Africa May, 2024

iv. The African Union Development Agency (AUDA-NEPAD) and the African Union High- Level Panel on Emerging Technologies (APET) which emphasize on the need to address concerns about job losses and the need to enhance job creation opportunities through the integration of AI in various developmental sectors (Tikoudi 2023; Engines and Growth 2023).

Based on these efforts, other African countries have gone ahead towards developing AI strategies, policy drafts, published policies and some have already started implementing them. For example, Mauritius, Egypt, Benin, and Senegal have developed AI strategies in 2018, 2021, and 2023 respectively. Whereas Rwanda (2023) have developed and published AI policy to serve as a roadmap towards harnessing the benefits of AI and mitigate its risks as briefly described in Figure 3.

Figure 3: Landscape of AI Policies, Strategies, and Data Protection Laws.

KEYS

Have Al Strategies

Have Al Policies

Drafting Al Regulations

Have Data Protection Laws

No Data Protection Laws

4.1.2 Case Study Analysis on AI Ethical Policies Landscape in Africa Most of the existing AI Strategies and Policies in Africa are in line with the African Union’s Agenda 2063 and strategies for digital transformation, science, technology and innovation (Moono 2021). For example, Benin have developed AI and big data strategy which focused on strengthening the national data infrastructure, promoting AI solutions, developing human capital, research and innovation; Egypt have developed National AI strategy which focuses on building human capacity by creating the necessary enabling environment; Mauritius’ AI strategy is guided by accountability, ethics and inclusiveness to build public trust and create a robust AI ecosystem(Brief 2023); Rwanda’s AI policy aim to harness the benefits of AI and mitigate its risks by building the skills, creating an open and secure data ecosystem, driving public sector transformation and accelerating

responsible AI adoption; Senegal’s AI strategy which aim to develop human capacity in AI, support solutions that solve development problems, foster partnerships between the public and private sectors and create an inclusive and trusted AI ecosystem (Engines and Growth 2023; StandfordUniversity 2023) Other countries such as, Ethiopia, Tanzania, Ghana, Kenya, Morocco, Nigeria, South Africa, Tunisia, and Uganda have also taken steps to define AI policies and establish institutions to drive AI development (Chatbri, Jemmali, and Hannachi 2019; Akello 2022; StandfordUniversity 2023). There has also been progress in addressing data issues across Africa and therefore, so far, the number of countries with data protection laws has doubled compared to the statistics of the last decade. The following are some of the case study analyses on the key AI principles in the AI strategies and policies in Africa;

15

Ethical Horizons - Mapping AI Policy in Africa May, 2024

Figure 4: Key AI principles in the AI Strategies and Policies in Africa.

Figure 4 briefly describes the key AI principles that have been promoted in the AI strategy and policy of some of the African countries. For example, Mauritius AI strategy has capture the issues of inclusive growth, sustainable development and well-being, human-centered values, fairness, robustness, security and safety, whereas Tunisia National AI strategy has captured inclusive growth, sustainable development and well-being, human-centered values, fairness, robustness, security and safety (Chatbri, Jemmali, and Hannachi 2019); Morocco National AI strategy has captured inclusive growth, sustainable development and well-being, human-centered values, fairness, robustness, security and safety, transparency and explain-ability, and accountability; Egypt National AI strategy has capture the issues of inclusive growth, sustainable development and well-being, human-centered values, and fairness; Sierra Leone National Digital and Innovation strategy has capture the issues of inclusive growth, sustainable development and well-being, human-centered values, fairness, robustness, security and safety; Senegal National Strategy for development of AI has capture the issues of fairness, , security, safety, transparency and explain-ability; Rwanda National AI policy has capture the issues of robustness, security and safety; and lastly, Benin National AI and Big Data strategy has capture the issues of inclusive growth, sustainable development

and well-being, robustness, security and safety (StandfordUniversity 2023).

4.1.3 Perspective from AI Stakeholders on AI Ethical Policies in Africa The study findings are in line with the white paper draft of African Union’s which provides a comprehensive plan for harnessing AI technology to propel socio-economic growth across Africa. The document provides a comprehensive plan for harnessing AI that can transform industries like healthcare, education, agriculture, and transportation, benefiting African governments through technological advancement, service delivery improvement, efficiency, and economic growth (Sampene et al. 2022; StandfordUniversity 2023). The study findings also have identified a diverse landscape of sectors requiring attention in AI policy formulation in Africa. The focus remains on sectors handling sensitive data, driving economic growth, shaping human capital, and facing unique AI-related challenges, with an overarching goal of fostering ethical AI use and regulatory coherence across sectors. Additionally, the responses emphasized on the critical importance of Healthcare, Finance

16

Ethical Horizons - Mapping AI Policy in Africa May, 2024

& Banking, Agriculture, and Education. Healthcare stands out due to its handling of sensitive personal data, requiring robust policies for data protection and ethical AI use. Finance & Banking follow closely, noting the sector’s rapid adoption of AI and the need for tailored policies for financial transactions and data security. Agriculture is highlighted for its economic significance and potential AI-driven transformations to enhance food security and economic growth. Education is seen as pivotal

for shaping human intelligence, digital literacy, and ethical AI use in educational technologies. While other sectors such as Technology, Media, E-commerce, Telecommunications, and Security are also recognized with lesser frequency compared to Healthcare, Finance & Banking, Agriculture, and Education sectors. These sectors present unique challenges and opportunities related to AI, ranging from data management to regulatory oversight and cybersecurity as shown in Figure 5.

Figure 5: Perspective from AI stakeholders on AI Ethical Policy in Africa.

In connection with the identified sectors, the study outlines the top five safety considerations that should be considered when developing AI ethical policies for African countries which include; data privacy and security, reliability and robustness, accountability and responsibility, bias and fairness, and African culture considerations as described in Figure 6.

17

Ethical Horizons - Mapping AI Policy in Africa May, 2024

Figure 6: Top Five AI Ethical Consideration in Africa.

4.1.4 Mapping Impact of AI Ethical Policies in Africa

Africa has been playing a central role in the global AI supply chain, particularly in the early phases whereby countries like Egypt, Rwanda, and Mauritius have published comprehensive strategies (World Bank 2020). Yet, foreign AI technologies dominate the continent, offering AI solutions which are not compatible with the African cultural norms, values, and conduct. Therefore, there is a need for African countries to develop, adopt, and implement AI strategies and policies that favor the local economies and ecosystems. The following in Figure 7 briefly detail the study findings on the adoption and implementation rate of AI ethical policy in Africa, whereby 48% of African countries have implemented and adopted the AI policy or AI strategy or data protection law and Act which is linked to AI, 40% have already started writing or validation the AI strategies, and 12% are in the initial stage of collecting information that will lead to start writing the AI strategies.

(i)Adoption Rate of AI Ethical Policy in Africa

Artificial Intelligence is a principal policy concern globally, in Africa there is an increasing number of countries developing and implementing AI policies, AI strategies and data protection laws and Acts which are in line with the AI ethical conduct (Brief 2023). The adoption and implementation of AI policy facilitates the delivery of governmental services to citizens, unlocking the value of AI to accelerate responsible predictive and prescriptive processes, allowing governments and agencies in the region to drive better services and decision making (Hofmann 2021). However, the implementation and adoption of AI ethical policy and its technologies in Africa faces challenges, including lack of government support, lack of knowledge and technical skills, uncertainty, lack of structured data, and user attitudes. As the capabilities for autonomous decision-making grow, there is an urgent need to rethink responsibility.

18

Ethical Horizons - Mapping AI Policy in Africa May, 2024

Figure 7: Adoption and Implementation Rate of AI Ethical Policy in Africa.

Initial stage 12%

Intermediate stage 40%

Final stage 48%

AI Adoption Rate in Africa

(ii) AI Ethical Concerns in Africa

African context, where the rapid proliferation of AI technologies intersects with unique socio- cultural, economic, and political dynamics. Based on the study findings, the following are several unique ethical concerns that have shaped the discourse surrounding AI adoption and governance in Africa. These considerations are influenced by diverse key factors which include; cultural norms, historical legacies, and socio-economic disparities prevalent across the continent which include; cultural sensitivity, inclusiveness and equity, data sovereignty and privacy, and computational growth as detailed in Figure 8.

In recent years, the concept of AI ethics has gained significant traction worldwide, as the ethical implications of AI technologies have become increasingly apparent. At its core, AI ethics pertains to the moral principles and guidelines governing the development, deployment, and use of AI systems, which encompass a wide range of considerations, including transparency, fairness, accountability, privacy protection, and the mitigation of bias. Therefore, the importance and impact of AI ethics cannot be overstated, particularly in the

Figure 8: AI Ethical Concerns in Africa.

19

Ethical Horizons - Mapping AI Policy in Africa May, 2024

(iii) AI Ethical Challenges and Opportunities in Africa As noted in the literature, most of the developed nations such as Europe, USA, and China are leading the use and development of AI (Fuster 2020). In addition, these countries are likely to reap most of the social-economic benefits and dominate policy debates. Therefore, in order to mitigate the counter risk, it is important to identify the existing

challenges and opportunities so that they can be well aligned in AI strategies and policies for sustainable transformation and social economic development of African countries. The following in Table 2 briefly describe the AI Ethical Challenges and Opportunities in Africa as indicated by (Okolo, Aruleba, and Obaido 2023)(Amankwah-Amoah and Lu 2022)(Jaldi 2023)(UNESCO 2022)(World Bank 2020)(Law 2023)(ITU and UNDP 2023).

Table 1: AI Ethical Challenges in Africa

Common Leading Voices Overwhelming unified voice leading the AI conversation in Africa is a must for the continent’s survival. Granted the African Union plays a hand, but a firmer one may be required to pull and streamline all the efforts, investments, and applications. More so is the survival of the ecosystem being built.

Privacy and Protection The proliferation of digital technologies and data-driven business models has raised concerns about privacy and data protection. With vast amounts of personal data being collected, stored, and processed, individuals are increasingly vulnerable to privacy breaches and data exploitation. However, advances in privacy-enhancing technologies (PETs) and encryption techniques offer opportunities to strengthen privacy protections and empower individuals to control their personal data. Accountability Ensuring accountability in digital ecosystems poses challenges due to the complexity of digital supply chains and the involvement of multiple stakeholders. However, blockchain technology and distributed ledger systems offer opportunities to enhance accountability and transparency by providing immutable records of transactions and interactions. By leveraging blockchain for supply chain management and auditing, stakeholders can enhance trust and accountability in digital transactions. Visibility Enhancing visibility and transparency in digital ecosystems is crucial for building trust and accountability. However, the opacity of algorithms and decision-making processes presents challenges to transparency and explainability. Yet, emerging technologies such as explainable AI (XAI) and algorithmic auditing tools offer opportunities to enhance visibility and accountability by providing insights into algorithmic

Sustainability Digital economies have the potential to drive sustainable development and mitigate environmental impacts through innovations such as smart grids, energy-efficient technologies, and circular economy models. However, the proliferation of digital devices and data centers also poses environmental challenges, including energy consumption and electronic waste. By adopting green computing practices, renewable energy sources, and eco-friendly design principles, stakeholders can harness the transformative potential of digital technologies while minimising their environmental footprint. Safety and Security Ensuring safety and security in digital ecosystems is paramount to building trust and confidence among users. However, the increasing frequency and sophistication of cyber threats pose challenges to cybersecurity and data protection. Nevertheless, advancements in cybersecurity technologies, such as artificial intelligence (AI) and machine learning (ML), offer opportunities to enhance threat detection and response capabilities.

20

Ethical Horizons - Mapping AI Policy in Africa May, 2024

By leveraging AI-powered security solutions and adopting best practices for cybersecurity hygiene, stakeholders can mitigate risks and safeguard digital infrastructure and assets.

decision-making and bias detection. By promoting algorithmic transparency and accountability, stakeholders can foster trust and confidence in digital systems.

Human Largely in the Loop The human factor remains indispensable in digital economies, requiring attention to ensure human-centric design and decision- making. Challenges arise from the potential for automation bias, job displacement, and erosion of human agency in digital environments. However, the concept of “human in the loop” emphasises the importance of human oversight and intervention in automated systems. By integrating human judgement and values into algorithmic decision-making processes, stakeholders can mitigate risks and ensure that digital technologies serve human interests and well-being. Infrastructure Building resilient and inclusive digital infrastructure is essential for promoting equitable access and participation in digital economies. Challenges include disparities in broadband access, digital literacy, and infrastructure investment between urban and rural areas and developed and developing regions. However, initiatives such as the UN’s Sustainable Development Goals (SDGs) and the World Bank’s Digital Development Principles offer frameworks for promoting inclusive and sustainable infrastructure development. By investing in broadband infrastructure, digital skills training, and digital inclusion initiatives, stakeholders can bridge the digital divide and promote equitable access to digital opportunities.

Rights (Children, Human, AI) Protecting and upholding rights in digital economies is essential for promoting human dignity, equality, and justice. Challenges arise from the potential infringement of rights, including children’s rights to privacy and protection, human rights to freedom of expression and access to information, and AI rights to accountability and ethical treatment. Nevertheless, legal frameworks such as the UN Convention on the Rights of the Child and the Universal Declaration of Human Rights provide a foundation for protecting rights in digital environments. By respecting and upholding rights-based principles, stakeholders can ensure that digital technologies contribute to the advancement of human rights and well- being. Literacy Promoting digital literacy and education is essential for empowering individuals and communities to navigate digital environments safely and responsibly. Challenges arise from disparities in digital literacy levels and access to educational resources, particularly among marginalized and underserved populations. Nevertheless, initiatives such as UNESCO’s Media and Information Literacy (MIL) program and the EU’s Digital Education Action Plan offer strategies for promoting digital literacy and lifelong learning. By investing in digital literacy programs, educational resources, and community-based initiatives, stakeholders can empower individuals to become critical thinkers, responsible digital citizens, and active participants in digital economies.

21

Ethical Horizons - Mapping AI Policy in Africa May, 2024

Table 2: AI Ethical Policy Opportunities/ Benefits in Africa

AI regulations can provide industry ethical guidelines to follow when developing AI technologies.

Clarity and Guidance

African AI developers can profit globally by exporting products, expanding financial opportunities beyond domestic markets. Increased consumer trust as the developed algorithms can explain how they came to the predictions, detection or diagnosis ethically.

Globalization

Building Consumer Trust

Eliminating ethical errors in AI will lead to the development of precise and accurate models, ultimately fostering business growth.

Fostering Business Growth

(iv) Case Studies on the Impact of AI Ethical Policy in Africa Figure 9 briefly describes eleven (11) AI and ICT National strategies for Mauritius, Algeria, Tunisia, Morocco, Egypt, Sierra Leone, Senegal, Benin, Tanzania, Nigeria, and Kenya; and one AI policy in Rwanda. Based on the study findings, it is vividly that African countries recognize the role of AI to sustain and enhance productivity gains through automation of processes in different sectors hence empowering the labor workforce, leading to social economic development. For example, AI is applied and used in agriculture to increase yields and poverty reduction; in

education system to become more accessible, personalized and efficient, in healthcare to standardize quality and create more accessible services, and in securing and expand accessibility of financial services to mention a few (Bradley, Wingfield, and Metzger 2020). In addition, governmental initiatives and intra- governmental organizations are playing a role to support responsible AI policies such as the AU Development Agency which supports the revolution of emerging technologies in Africa for social economic development, and the UNESCO which has conducted AI need assessments to identify priority areas in Africa (UNESCO 2022).

Figure 9: Case Studies on the Impact of AI Ethical Policy in Africa.

Benin National Al and Big Data Strategy 2018 It aims to leverage extensive data and Al, building skills to become a West African Al innovation hub, boosting investment appeal, and leading in regional Al development.

Mauritius Al Strategy 2018 Leverage AI and other emerging technologies to address social and financila issues, revitalize traditional economic sectors, and create a new development pillar.

Tunisia National Al Strategy 2021 Establish Tunisia as an Al hub in Africa and the MENA region. This ambition is supported by creating a conducive environment for Al development and application across various sectors.

Morocco National Al Strategy 2021 Establish a strategy for development of Al ecosystem and choosing priority sectors with dedicated roadmaps.

22

Ethical Horizons - Mapping AI Policy in Africa May, 2024

Algeria National Strategy on Research & Innovation in AI, 2021 Improve performance in a number of so-called priority segments, such as higher education, health, transport, energy and technologies.

Egypt National Al Strategy 2021 Utilize Al technologies to support the achievement of SDGs, establish Egypt as a regional cooperation facilitator and active international player in Al.

Sierra Leon National Digital and Innovation Strategy 2022 Prepare for the Fourth Industrial Revolution, enhancing universities, backing startups with tech and leverage Al to meet development annd sustainability objectives.

Senegal

National

Strategy for Development of AI 2023 Integrate data at its development core. It fosters a digital ecosystem promoting sustainable growth, innovation, and value, in pursuit of a visionary digital future.

Tanzania National Al Strategy 2024 Leverage Al and other emerging technologies to address social and financial Issues, revitalize traditional economic sectors, and create a new development pillar.

Nigeria AI Strategy 2024 Enhance Al and other emerging technologies to address social and financila issues, revitalize traditional economic sectors, and create a new development pillar.

Kenya Al Strategy 2024 Leveraged Al to solve Kenya’s most prevalent problems that cut across various sectors including health, education, agriculture, finance and business.

Rwanda National Al Policy 2023 Rwanda seeks to use Al to drive sustainable, inclusive growth, positioning itself as a leader in ICT and innovation locally and globally.

(v) Promotion and Compliance Strategies for AI Ethical Policy in Africa The following are the identified strategies for promoting and complying with AI ethical policies.

23

Ethical Horizons - Mapping AI Policy in Africa May, 2024

Figure 10: Promotion and Compliance Strategies for AI Ethical Policy in Africa.

Monitoring: • Good monitoring and best practices should be put in place. • Monitor the use of Al policies from time to time.

M

Education and Awareness: • Create awareness and education campaigns about Al and its ethics. • Talk about the bennefits of Al and how it cann increase productivity.

E

P

Penalties: • Enforce penalties for non-compliance.

I

Incentives for complicance: • Provide incentives to promote compliance.

C

Collaboration: • Policymalers should engage the stakehodlers when developing the Al policies from the inception to implemenation stage.

4.2 Discussion Artificial Intelligence is the foundation of digital transformation and business innovation of Africa, hence, leading the new stage of advancing industries and technologies and making decisions that can affect the future in Africa (Stern et al. 2023). With the wide use and application of AI in everything, from automated predictive models in agriculture and financial sectors to diagnostic systems in the healthcare domain, concerns have emerged on the liability and responsibility in the use and application of AI. Liability is the obligation of an individual to compensate the injured party for the damage suffered. “African legal systems, like other international legal systems, do not provide a clear answer to this question (Jaldi 2023). AI is an immaterial entity with no legal personality or assets, so it cannot be held responsible for its actions. As a result, and because artificial intelligence is immaterial (software), but embedded in machines (hardware), it is difficult

to extract its specific behavior beyond the action it co-commands with man. However, its growing autonomy (co-learning) will make it increasingly independent of human action”. Hence, there is a need for a policy for AI which guides effective protection to potential victims. Whereas, Responsible AI principles must ensure that transparency, fairness, and accountability are guaranteed in the underlying principles of adoption, development, implementation and use with a high human-centred consideration. The study findings indicate that some of the data-driven decision-making solutions are susceptible to inaccuracies, discriminatory outcomes, embedded and exacerbated bias, and even unintended consequences due to various limitations that occur through the process. Nevertheless, there is a scarcity of pertinent policies that prioritise AI development and its application in Africa while addressing potential social repercussions.

24

Ethical Horizons - Mapping AI Policy in Africa May, 2024

Page 1 Page 2 Page 3 Page 4 Page 5 Page 6 Page 7 Page 8 Page 9 Page 10 Page 11 Page 12 Page 13 Page 14 Page 15 Page 16 Page 17 Page 18 Page 19 Page 20 Page 21 Page 22 Page 23 Page 24 Page 25 Page 26 Page 27 Page 28 Page 29 Page 30 Page 31 Page 32 Page 33

www.afrilabs.com

Made with FlippingBook - Online Brochure Maker