In response to the call to facilitate multistakeholder engagement and contribution into the development of a continental AI Strategy, TeachSombody Co-founder Kadian Davis-Owusu PhD, provided recommendations on the following aspects:

• Addressing and Mitigating AI Risks to Africa
• Creating an AI Ecosystem and Capabilities for Harnessing the Potential Benefits of AI

The responses are outlined below. 

Addressing and Mitigating AI Risks to Africa (Form 1)

What ethical risks does AI pose to African people, communities and value systems?

AI systems can perpetuate and even amplify existing biases if they are trained on biased data sets, which may not or misrepresent diverse groups in multi-cultural and multilingual African societies. Moreover, there is growing concern about the relevance of training data originating from the Global North as these datasets are often misrepresentative of Africa’s cultural, linguistic, or systemic realities. Ultimately, this can lead to AI systems that are irrelevant and ineffectual in African contexts. This mismatch can affect the AI system’s understanding and interpretation of the local nuances in language, culture, belief systems, agriculture, education, healthcare, employment, and justice systems. Moreover, AI systems that do not consider local contexts may risk infringing on fundamental rights of privacy and equality. This can erode trust and alienate specific indigenous communities, undermining the effectiveness of AI applications. In addition, the deployment of unfair AI systems in Africa could disproportionately affect vulnerable groups including women and children. This is especially important in scenarios like access to education and healthcare, job recruitment, and loan approvals, where biased AI could reinforce existing inequalities. Lastly, the deployment of AI systems could automate many jobs that could result in significant shifts in the labour market. This might escalate unemployment and deepen economic inequalities, which is already a significant challenge in Africa. 

What risks does AI pose to democracy, social cohesion and the rule of law in Africa?

Democracy - Deepfakes AI technology can erode democratic processes through the dissemination of targeted disinformation, misinformation, and manipulation of social media platforms. This can undermine fair and transparent democratic processes, spread false narratives, sway elections, and ultimately erode the public trust in knowledge and informational sources. In addition, the use of AI in surveillance, emotion recognition, risk assessments, predictive analysis, and biometric monitoring can lead to excessive monitoring and violation of human dignity and people’s fundamental rights to privacy and freedom of movement. Moreover, this could stifle freedom of expression and the right to freedom of association and assembly, which are critical elements of African democratic societies.

Social Cohesion - AI systems if not designed inclusively can exacerbate social inequalities by favouring certain groups over others based on unfair and biased algorithms. This can worsen existing disparities in gender, education, digital access, and employment, which are notably significant in Africa.

Rule of Law - AI applications in judicial and law enforcement sectors should be transparent and accountable. Otherwise they could lead to biased decision-making  based on certain demographics such as country or region of origin, address, tribe, educational level, or employment status  and could perpetuate injustices that undermine the rule of law. 

What are the key safety and security issues for AI in Africa?

Key safety and security issues for AI in Africa include but are not limited to privacy and security concerns, cybersecurity threats e.g., malicious attacks on financial, transportation, energy, and telecommunication systems could lead to widespread destruction. Threats to reliability and safety of AI systems especially in sectors such as health care and public safety could lead to loss of life. 

What governance frameworks are best suited to address and mitigate identified risks, and where might regulation be needed? What will be the role of governments and regional institutions?

To address and mitigate the challenges posed by the adoption of AI technologies in the African context, several governance and regulatory frameworks are needed. The role of governments, regional institutions, civil society, technological organisations, and academia is crucial in establishing, implementing, and monitoring these frameworks.

To begin with, adoption of ethical guidelines to ensure AI technologies are developed and exploited in a way that is fair, transparent, interpretable, explainable, just, environmentally friendly and sustainable, respects privacy and security, and includes human oversight. These guidelines should address issues like bias, discrimination, and privacy. Reference to existing frameworks e.g., Microsoft's Responsible AI, Google's AI Principles, IBM's AI Ethics, OECD AI Principles, UNESCO's recommendation on the ethics of artificial intelligence, Mozilla's Trustworthy AI, Mozilla AI Theory of Change, The Aletheia Framework,  WEF’s AI Oversight Toolkit for Boards of Directors, AI Ethics Maturity Model, Salesforce, and Artificial Intelligence Ethics Framework for the Intelligence Community, in consultation with public participation and stakeholder engagement could serve as a template to design the Recommendations for AI Ethics specific to the region of Africa and ensure AI governance is comprehensive and inclusive. 

Secondly, African regulatory bodies should ensure that AI products meet specific standards prior to deployment including requirements for transparency in algorithms, representative training datasets, compliance with privacy laws and data standards, and impact assessments to understand how the AI system will affect different demographics in the African context.

Thirdly, as it relates to the regulatory needs, strict data protection laws are needed across all member states of the African Union to govern the collection, storage, and use of data by AI applications, ensuring that privacy rights are maintained. Sector specific regulation might be an added benefit in areas such as healthcare, migration, and criminal justice to address specific risks in those fields.

Finally, in terms of the roles of governments and regional institutions, I expect that African Union together with African Governments will lead the formulation of AI policies and frameworks  that promote innovation and avoid harm for individuals and society. In addition, they will train regulators and law enforcement on AI and its implications. Furthermore, more capacity building initiatives should be put in place to create AI awareness among the general population on a national and regional level. Lastly, more corporation is needed with funding agencies to support research in responsible AI, especially studies focussing on developing AI that considers the local context and places the citizens’ needs at the forefront of their research agenda. 

Creating an AI Ecosystem and Capabilities for Harnessing the Potential Benefits of AI Form 2

What steps should Member States take to build the capacities of higher education to teach, research and innovate on AI and address the skills gap?

To build the capacities of higher education to teach, research and innovate on AI and address the skills gap, member states can take several strategic steps such as:

Curriculum Development - Updating and expanding higher education curricula to AI related courses not only for engineering and computer science related disciplines but also humanities, social sciences and medical related fields.  

Faculty Training -  There is a pressing need for comprehensive training programs to empower educators in responsibly leveraging AI technologies.

Funding and Research - It is imperative to increase funding and research to encourage innovation and to explore the design and development of responsible AI applications. Research institutions should establish clear data policies to maintain the quality and integrity of data.

Support start-ups and Innovation Hubs - Establish and support innovation hubs and incubators within universities to encourage students and researchers to develop their own AI solutions and startups. This can help to bridge the gap between theoretical knowledge and practical skills.

What will the private sector's role be in AI R&D and innovation? What lessons can we learn from the existing AI startup ecosystem?

The private sector's role in AI R&D and innovation include but are not limited to the following. 

Investment and funding - They can provide the financial means to support research, design,  development, and scaling of AI technologies. 

Market Trends - Private companies are usually adept in identifying market trends in AI. In response, they quickly and skilfully develop AI solutions to meet these needs.

Talent Development - Many companies partner with academic institutions to nature talent through internships. This helps to advance AI knowledge and create a skilled workforce.

Lessons 
Agility and Speed - Start-ups are known to pivot and are skilful in quickly adapting to market needs.

Niche focus - Start-ups are usually focused on specific problems, which enable them to become experts in their domain. 

How can the basic education and TVET curriculum be improved to prepare the youth for AI?

Basic education and the TVET curriculum can be improved to prepare the youth for AI in the following ways.

1. Introduce fundamental AI concepts including what is AI? How it works and applications of AI at the primary school level. This should be done using appropriate language and explained in a simplified manner.

2. Strengthening critical thinking and problem solving is key to equip students with the analytical skills needed to make better decisions regarding the use of AI.

3. The ethical use of AI should be embedded in lessons with a focus on privacy, security, safety, and the societal implications. This encourages students to think of how AI impacts individuals and society and avoids harm.

4. Incorporate real-world practical projects that enable students to apply their learning on AI tools, techniques, and methods.

What strategies should be implemented to improve African infrastructure for AI?
E.g. broadband networks, data centres, high-performance computers

Addressing areas such as internet connectivity, digital literacy, access to funding/capital (mainly for start-ups), energy infrastructure, and the development of data centers and cloud infrastructure to handle the vast amount of data produced by AI systems. The AU should encourage the creation and use of AI solutions that support renewable energy, resource conservation, and waste reduction.

How can AU Member States improve the reliability of AI-related data?

AU Member States can improve the reliability of AI-related data by creating audit processes to assess data quality, model suitability, and efficient performance, while ensuring AI systems act as intended. Tools such as Error Analysis (https://erroranalysis.ai/) are relevant to diagnose and recognise errors in AI applications. 

As a respresentative of TeachSomebody, I thank the African Union for this opportunity to provide our recommendations  for this Multistakeholder Consultative Sessions on the Development of a Continental AI Strategy on Artficial Intelligence (AI). We are eager to continue our involvement in the process.