Artificial Intelligence and Development in Africa

According to the African Union’s (AU) recent Continental AI Strategy, artificial intelligence (AI) is poised to play a crucial role in shaping a more sustainable global future. The AU is particularly optimistic that AI presents enormous opportunities for inclusive socio-economic development and cultural renaissance in Africa. The vision that technological innovation and scientific knowledge will fix problems of injustice, inequality and poverty in Africa is not new – and Yousif Hasan’s critical anatomy of this belief in the field of AI4Development (AI4D) helps us understand that better. Specifically, he calls for greater attention to the different and contested socio-political imaginaries attainable through and supportive of advances in science and technology, arguing that AI4D projects are not just about technology and innovation but also about desirable AI futures and the political imaginations sustaining them. Governing AI requires building institutions that support AI development and bring together diverse perspectives from different communities across the continent, ensuring that the needs and futures of marginalised and underrepresented groups are prioritised in policy discussions (Hassan 2023). 

What’s wrong with the prevailing discourse of AI4D in Africa?

In this webinar hosted by the EthicsLab on August 1st, 2024, Hassan reflects on the limits of this prevailing discourse of AI4D by foregrounding two coterminous questions. Amidst an international development crisis and prevailing techno-optimism, can (responsible) AI fix Africa’s development? Secondly, who gets to shape AI futures in Africa and the visions of a modern African state? He offers several observations and insights towards answering these questions. 

1. The AI innovation ecosystem in Africa is primarily configured and driven by International Development Agencies, Multinational Corporations, and Western philanthropic foundations. These actors aim to build capacity for AI research and innovation in Africa and enable the African AI Community to contribute locally and globally to developing responsible and ethical AI.
2. This ecosystem is underpinned, Hassan argues, by a deficit model of development that links development failures to a lack of human and technological capacity. Deficit models focus on what is missing, often leading to the imposition of external solutions, thereby fostering dependency and undermining local agency. In discourse around the uptake of AI technologies in Africa, the deficit model is premised on the lack of African innovations and datasets that can inform local AI applications and solutions. Deficit discourses can also lack a critical edge, removing the focus from questions about the aims, uses, and distribution of AI benefits. 
3. Interlocutors operating with this deficit model are mindful of the questions of power and inequality it elicits, consequently appropriating but unfortunately de-politicising ethical grammars like ‘decolonisation’. In the deficit model, decolonising AI becomes a technical practice of adapting technologies to a local context (e.g. low resource computing in response to historical infrastructure challenges). Once turned into a technical exercise of collecting local datasets and indigenous techno-solutionism, decolonisation is divorced from its historic articulations as a political project rooted in political, economic, and technological sovereignty. Put simply, it’s not always clear how ‘adaptation to local context’ is connected to a broader struggle against colonial and imperial hegemony.
4. AI4D lacks engagement with the political imaginations of marginalised and underrepresented AI communities on the continent. Hassan develops this argument by contrasting two Pan-African imaginaries. One vision is put forward by the African Union in its strategy for science, technology and innovation for 2024 as part of ‘Agenda 2063: The Africa We Want’. In this discourse, AI heralds an African Renaissance marked by inclusive socio-economic development in a globalised world. The crucial element here is a question of audience. The vision is extraverted, focused on attracting international development funding and investments from multinational corporations by aligning with Western ideas of social progress and economic competitiveness in a globalised world. This approach, Hassan contends, gives more control to international actors in shaping Africa’s development agenda. By contrast, civil society and local AI communities articulate a project of technological sovereignty (e.g. collective ownership of AI infrastructures) grounded in prosocial philosophies like ubuntu. One key difference between the two visions, for example, lies in contrasting approaches to data governance in Africa. The first sees data as a private resource governed by intellectual property laws, ready for commercial exploitation. The second views data as a communal resource and part of a digital commons with benefits that should be shared across the community. 

Ultimately, Hassan offers a valuable presentation that maps the diverse networks, projects, and actors involved in AI4D in Africa and illuminates the contested visions of desirable AI futures by peeling back the layers of an Africa encased by narratives of lack and deficit.

Concluding thoughts

Since the early days of political decolonisation in Africa, science and technology have been key components in the imaginaries of what Africa could be. For some, such as Amina Mama, academic scholarship is key to liberation and social justice in Africa. Yet, while African states initially invested heavily in African academies as a key component fostering political independence, structural adjustment programmes in the 1980s nipped Africa’s scientific and technological potential and entrenched the scientific and technological might of institutions in Europe and North America. Therefore, western dominance in AI is not accidental; it is the consequence of the historical and political processes that continue to shape the technological imaginaries and possibilities for scientists and innovators on the continent. Hassan’s analysis of the deficit model that guides the AI4D vision is a stark reminder of the enduring power of the technological imaginaries that continue to reduce African scientists and innovators to subservient roles.

This webinar was inspired by previous conversations with Amrita Pande (UCT) about how framing and representation impact the content and scope of ethics in debates around science and technological innovation in Africa. These discussions were driven by a curiosity about the dominant frames, like the deficit model described by Hassan, that inscribe Africa in conversations around the ethics of new and emerging technologies. The Africa which needs to embrace these technologies is often depicted as a place of material scarcity and developmental deficit. This means that conversations about ethics invariably focus on questions of distributive justice, aiming to facilitate or equalise access to technologies and the benefits they accrue. Africa is drawn into the discussion in terms of unequal access and the need for low-cost technologies that can reach diverse populations. This certainly aligns with Hassan’s observation about the depoliticisation of decolonisation, because decolonisation framed in terms of adaptation and relevance often speaks to this context of scarcity and deficit. 

This does not mean questions of distributive justice are unimportant. Of course, they are. The point is about the other conversations elided because of certain portrayals. Pande’s own work considers this question with repro-genetic technologies, looking beyond the framing of lack and (un)equal access to consider the kinds of individual desires legitimated, their embeddedness within and reproduction of racial hierarchies, and how assisted reproductive technologies often enjoin fertility experts and intended parents to co-produce the desirability of whiteness. Similarly, Hassan argues that narratives of lack and deficit not only shape the possibilities for certain interventions while excluding others but also divert attention from the AI futures envisioned by marginalised and underrepresented AI communities in Africa.