Submission to the UN on the UN Interim Report on Governing AI for Humanity

1. Introduction

In December 2023, the UN’s Global Advisory Committee on AI released its Interim Report on Governing AI for Humanity. The purpose of the report was to analyse global risks and opportunities in relation to AI and to advance recommendations for the international governance thereof, in an effort to address the global governance deficit and to ensure AI technologies are developed and deployed in ways that benefit all of humanity. ​Accordingly, the UN’s Interim Report provides preliminary recommendations, outlines next steps for the development of a final report, and invites contributions from a wide range of stakeholders to make comments and suggestions on how to improve the report before it is finalised in August 2023.

2. Background to African Observatory’s Consultation with Regional Experts

In response to the UN’s Call for Submissions, the African Observatory on Responsible AI (African Observatory) facilitated a regional consultation with experts from across Africa to reflect on the content of the Interim Report and its implications for the region. The meeting took place virtually on 25 March, and was attended by a diverse range of experts from across the region, which included representatives from the following institutions: Université virtuelle du Sénégal (UVS), Bennett Institute for Public Policy, the Federal Ministry of Communications,Innovation & Digital Economy, MinoHealth AI Labs, South African Human Rights Commission, National Information Technology Development Agency, Dataphyte Foundation, African Observatory on Responsible AI, OpenUp, Global Index On Responsible AI, Information Regulator SA, Google Alphabet, Atlantic Council, Penplusbytes, Hawthorn Strategy, KaraAgro AI, Local Development Research Institute, Nigerian Nuclear Regulatory Authority, CIPIT, Strathmore University, Future of Privacy Forum, HSRC,Mozilla, London School of Economics & Political Science, Data For Development,Tech Hive Advisory, African Center for Economic Transformation (ACET), University of Johannesburg, Google, African Development Bank Group, Railway Safety Regulator, Policy - Uganda, as well as participants from the UN’s Geneva Office and the IDRC’s Office in Canada. 

The outcome of the African Observatory’s consultation was a comprehensive submission that outlined the priorities for the African region and provided recommendations for enhancement before finalisation in August 2024. The submission emphasised the significance of taking a differentiated approach to address the distinctive needs and priorities of Africa, and the critical need for equitable representation from the African continent in key leadership and decision-making positions concerning AI's global governance in the future. Furthermore, the submission underscored the importance of acknowledging Africa's distinct socio-economic, cultural, and political dynamics to ensure that global governance institutions are prepared to effectively address the region's specific needs, interests, and challenges.

3. Content of African Observatory’s Submission

Below is the content of the African Observatory's submission to the UN’s Advisory Body, which responds to the specific sections of the Interim Report, specifically in relation to opportunities and enablers, risks and challenges, guiding principles for the establishment of global governance institutions for AI, and institutional functions for a global governance region for AI. 

A. Opportunities and Enablers 

Although this section of the UN’s Interim Report highlights the unique challenges faced by the ‘Global South’, the report does not make a specific reference to Africa or the vast diversity within it. Since the purpose of the report is to analyse and advance recommendations for the international governance of AI, it is important for the report to propose a differentiated approach that explores the distinct opportunities present in different regions and the specific enablers required to seize them. 

In Africa, there are two key opportunities for harnessing AI to benefit humanity. The first is leveraging AI to advance the developmental priorities of the region. The second, which is closely linked with the first, is using AI to address inequalities in critical sectors, such as education, healthcare, agriculture, finance, and access to basic services. These opportunities are the most crucial for Africa given their potential to advance the region's broader developmental objectives, but also to alleviate a complex set of intersecting inequalities. Prioritising the responsible use of AI in these sectors can catalyse sustainable growth, improve the quality of life for individuals and communities, and drive inclusive economic prosperity across the region.

Accordingly, there are four key enablers required to leverage these opportunities. The first involves investing in AI infrastructure, such as hardware (servers, storage) and software (data processing, machine learning tools), systems for data handling and processing, as well as cloud services for supporting the development and operation of AI. Even more vital is the creation of extensive datasets that accurately reflect African populations, including data that is available in indigenous languages. Accordingly, this demands that African people become generators of data to ensure that AI is useful to them and to mitigate the impacts of unrepresentative datasets. The second requires building expertise in essential AI-related fields within critical sectors to empower local industries to leverage AI technologies, drive innovation, enhance competitiveness and sustainable economic growth. The third entails fostering partnerships between government, private sector [e.g. industry actors], civil society, and academia, to promote collaboration in addressing complex challenges, driving inclusive innovation, and advancing sustainable development goals for the region. The fourth necessitates establishing AI policies grounded in ethical principles, guidelines, and regulations for responsible AI development is critical to leveraging the power of AI to benefit humanity by guaranteeing transparency, fairness, accountability, and trust in AI systems, and fostering ethical innovation, and mitigating risks in AI applications.

Given the unique opportunities AI offers Africa and the set of enablers required, it is essential for the UN’s report to highlight differences amongst countries in the Global South.

B. Risks and Challenges 

The UN's Interim Report does not differentiate between risks affecting countries in the Global South, including Africa, and those affecting other regions. The same applies to the actions proposed to address those risks. As with the previous section, we consider this to be a limitation of the Report, and call on the Advisory Board to propose a differentiated approach that acknowledges the unique set of risks affecting Africa, and to differentiate between the actions required to address them.

In Africa, there are four major risks that threaten the potential of AI to benefit humanity. First, existing disparities in the distribution of benefits across the region are likely to hinder fair access to AI technologies, which not only worsens those inequalities but also jeopardises Africa's ability to guarantee the inclusive advancement of AI for the benefit of humanity. Second, the lack of AI infrastructure and resources across Africa limits the possibilities of using AI responsibly and equitably, limiting its potential to reach individuals, communities and countries that are most in need of the benefits, increasing the likelihood of introducing new risks. Third, insufficient attention to developing and retaining AI expertise presents challenges for Africa in leveraging the full potential of AI while safeguarding against risks, which increases the chance of individuals with essential skills leaving the region to explore more promising opportunities. Fourth, the absence of strong governance frameworks for AI and the high costs of implementation of such frameworks across most African countries creates an environment that is ripe for misuse, posing a potential threat to human rights and democracy. 

There are four proposed actions to address those risks. First, develop and implement inclusive AI policies and frameworks (such as impact assessments) to ensure the fair and equitable use of AI. Second, invest in AI infrastructure to enhance technical capabilities, improve data governance, and improve internet connectivity, as well as expand access to computers and other tools, such as centres for open data. Third, create skills development programs to build and retain talent across the region, fostering local expertise to govern AI for the benefit of Africa by facilitating partnerships between universities, government, and private companies to offer training courses and opportunities for start-ups. Fourth, institutionalise strong governance mechanisms to protect and uphold human rights and democracy by establishing regulatory bodies and ethical frameworks.

Considering the distinctive risks associated with AI in Africa, a distinct set of recommended actions is essential for the continent to harness the potential of AI for the betterment of humanity. Similar to the preceding section, it is crucial for the UN's report to emphasise these differences instead of applying a uniform set of risks and actions to the global community as a whole.

C. Guiding Principles to guide the formation of new global governance institutions for AI 

The guiding principles encompass many of the factors required to build strong governance institutions for AI. However, a deeper level of contextual and cultural understanding is required for a differentiated approach to ensure the relevance and applicability of these principles to the African region.

Accordingly, our suggestions are as follows:

• Principle 1: For AI to be governed by and for the benefit of all people, it is imperative to ensure fair and equitable representation from the entire African region. This requires a commitment to allocating key leadership positions to representatives from the region, and to ensuring African stakeholders are given equal power in key decision-making processes. Inclusion does not only mean participation, but requires the integration of African expertise, experiences, and perspectives to ensure global governance regimes are tailored to meet the needs, interests and values of the region.

  
  

• Principle 2: In order for AI to be governed in the public interest in Africa, it must be grounded in ethical and human-centred approaches, which promote the protection and well-being of individuals, communities and countries across Africa, specifically in areas where the benefits and impacts of AI may not be distributed equally. Without a commitment to ethical and human-centred approaches in the governance of AI, it will be very difficult to ensure it is used in the public interest.

  
  

• Principle 3: There must be continuous dedication to digitising all forms of data in Africa and to building robust mechanisms for data sharing and protection. This entails establishing strong quality assurance measures and legal frameworks to ensure the precision and inclusivity of local data sets from Africa, taking into account indigenous languages, thereby fostering a data commons accessible for public use and benefit.

  
  

• Principle 4: For AI governance to be universal, networked, and rooted in adaptive multi-stakeholder collaboration, there must be a commitment to localisation of capacities and context-sensitivity, which recognises the unique socio-economic, cultural, and political dynamics of Africa to ensure global governance institutions are equipped to meet the specific needs, interests and challenges of the entire region. Governance rules should be supported by a clear articulation of the underlying values they seek to protect, such as algorithmic transparency and accountability.

  
  

• Principle 5: For AI governance to be anchored in the UN Charter, International Human Rights Law, and other agreed international commitments such as the Sustainable Development Goals, it is crucial to emphasise the communal aspects of African human rights instruments, in contrast to the individualistic approaches commonly favoured by Western nations. In addition, there is a need to enhance mechanisms for overseeing and documenting the effects of AI on African nations, as well as enforcing human rights laws and ethical norms throughout the region.

D. Institutional Functions that an international governance regime for AI 

As with the previous section, it is necessary to take a differentiated approach that acknowledges the unique needs of Africa without undermining its capabilities to enable the applicability of these functions across the region.

Suggestions are provided below:

• Function 1: To effectively assess the trajectories and impacts of AI in Africa, sustained investment in the monitoring of national AI ecosystems is critical. Tools like the Global Index on Responsible AI can provide useful insights to governance institutions because it monitors progress towards responsible use of AI at a national level using primary data annually. 

  
  

• Function 2: To effectively reinforce the interoperability of governance efforts for AI, governance frameworks need to be adaptable to evolving technological advancements and flexible enough to accommodate unique regional characteristics and challenges across Africa. For example, the costs associated with compliance and governance are especially high for African nations, many of which already face insufficient levels of data protection, leading to restrictions in data sharing.

  
  

• Function 3: To effectively develop and harmonise standards, safety, and risk management frameworks that are relevant to Africa, governance institutions will need to meaningfully engage with multiple stakeholders, conduct comprehensive human rights impact assessments, develop guidelines that are relevant regional needs and interests, and implement tailored capacity-building initiatives.

  
  

• Function 4: To effectively facilitate the development, deployment, and use of AI for economic and societal benefit in Africa through international multi-stakeholder cooperation, governance institutions will need to work with both national and regional bodies to strengthen the legal, financial, and technical frameworks for AI. Capacity building for all sectors of society will be critical for the advancement of responsible use of AI across the region and ensuring that Africans contribute to AI innovation.

  
  

• Function 5: To effectively promote international collaboration on talent development, access to compute infrastructure, building of diverse high-quality datasets, responsible sharing of open-source models, and AI-enabled public goods for achievement of the SDGs, governance institutions to create and facilitate sustainable partnerships between government, private sector and academia in Africa.

  
  

• Function 6: To effectively monitor risks, report incidents, and coordinate emergency responses, governance institutions need to implement oversight systems, receive reports on misuse or abuse, facilitate coordinated responses, conduct thorough investigations, and have enforcement powers to collaborate with regulatory authorities in Africa to ensure compliance.

  
  

• Function 7: For effective compliance and accountability based on norms, there needs to be continuous engagement with both state and non-state actors on the relevance and applicability of these norms to the national and regional context. 

E. Other comments on the International Governance of AI section (aside from Principles and Functions, covered in above questions) 

The UN’s Interim Report should expand its scope to include a discussion on the proposed structure and institutional arrangements for a global governance regime on AI. This should involve considerations such as its location, membership composition, decision-making procedures, and interactions with other global governance entities, including regional bodies like the African Union, and international institutions such as National Human Rights Commissions. Further, given the absence of leadership from Global South nations in global structures on AI, the UN Advisory Board should leverage its position to enhance and broaden leadership in global AI governance for greater equity and to address existing imbalances on a global scale.

A fundamental gap in the UN Interim Report is the articulation of a global fund for AI. Without serious funding toward responsible AI R&D, skills development and learning, institutional capacity building, and compute and data resources for AI, the AI divide is set to increase at an exponential rate. A globally governed AI fund to attract, manage and disperse funding for the responsible development of AI in places that need it most, and to ensure that countries are capacitated to comply with the requirements of a global governance framework, is an essential function that the UN should be proposing. This forms a fundamental part of taking a differentiated approach to meet the diverse needs of the African region.

In addition, the UN’s Interim Report should call for the establishment of a Special Rapporteur on AI Governance which should be responsible for global oversight of AI, compliance with human rights law and international law, and mitigating risks associated with AI technologies. This role can promote transparency, protect human rights, and provide recommendations to ensure ethical and accountable use of AI at a national level. By monitoring compliance and setting international standards, the Special Rapporteur can play a vital role in guiding the responsible development and deployment of AI technologies worldwide, and ensuring that adequate attention is paid to the unique needs of African countries and that the interests and resources of the region are protected and prioritised.

F. Any other feedback on the Interim Report 

It is important for the UN’s Advisory Board to incorporate a section that defines key terms that are used throughout the report, such as ‘opportunity’, ‘enabler’, ‘governance’, ‘data commons’ etc. to establish a common understanding amongst its readers, and to standardise the style and language used throughout the document. This will ensure clarity, consistency, and effective communication of ideas.