Governing algorithms from the South: a case study of AI development in Africa

Abstract

AI technology is capturing the African imaginations as a gateway to progress and prosperity. There is a growing interest in AI by different actors across the continent including scientists, researchers, humanitarian and aid organizations, academic institutions, tech start-ups, and media organizations. Several African states are looking to adopt AI technology to capture economic growth and development opportunities. On the other hand, African researchers highlight the gap in regulatory frameworks and policies that govern the development of AI in the continent. They argue that this could lead to AI technology exacerbating problems of inequalities and injustice in the continent. However, most of the literature on AI ethics is biased toward Euro-American perspectives and lack the understanding of how AI development is apprehended in the Global South, and particularly Africa. Drawing on the case study of the first African Master’s in Machine Intelligence program, this paper argues for looking beyond the question of ethics in AI and examining AI governance issues through the analytical lens of the raciality of computing and the political economy of technoscience to understand AI development in Africa. By doing so, this paper seeks a different theorization for AI ethics from the South that is based on lived experiences of those in the margins and avoids the framings of technological futures that simplistically pathologize or celebrate Africa.

Full text

Vol.:(0123456789)
1 3
AI & SOCIETY (2023) 38:1429–1442
https://doi.org/10.1007/s00146-022-01527-7
OPEN FORUM
Governing algorithms fromtheSouth: acase study ofAI development
inAfrica
YousifHassan1
Received: 14 June 2021 / Accepted: 16 June 2022 / Published online: 13 July 2022
© The Author(s), under exclusive licence to Springer-Verlag London Ltd., part of Springer Nature 2022
Abstract
AI technology is capturing the African imaginations as a gateway to progress and prosperity. There is a growing interest in
AI by different actors across the continent including scientists, researchers, humanitarian and aid organizations, academic
institutions, tech start-ups, and media organizations. Several African states are looking to adopt AI technology to capture
economic growth and development opportunities. On the other hand, African researchers highlight the gap in regulatory
frameworks and policies that govern the development of AI in the continent. They argue that this could lead to AI technology
exacerbating problems of inequalities and injustice in the continent. However, most of the literature on AI ethics is biased
toward Euro-American perspectives and lack the understanding of how AI development is apprehended in the Global South,
and particularly Africa. Drawing on the case study of the first African Master’s in Machine Intelligence program, this paper
argues for looking beyond the question of ethics in AI and examining AI governance issues through the analytical lens of
the raciality of computing and the political economy of technoscience to understand AI development in Africa. By doing
so, this paper seeks a different theorization for AI ethics from the South that is based on lived experiences of those in the
margins and avoids the framings of technological futures that simplistically pathologize or celebrate Africa.
Keywords AI ethics· Decolonizing AI· Technoscientific capitalism· Political economy of AI· Science and technology
studies
1 Introduction
The main objective of this paper is to underscore the need
for a different approach based on the raciality of comput-
ing while incorporating political economy of technosci-
ence analysis when looking at the governance of artificial
intelligence (AI) in Africa, and more broadly the Global
South. The interest in AI has grown significantly in Africa
in recent years. For example, several AI conferences have
emerged in the content such as Data Science Africa, Deep
Learning Indaba, AI Expo Africa, and Machine Learning
Africa. The African data science1 project is established by
the 2030 Vision initiative to support the UN Sustainable
Development Goals (SDGs) and targets. Many African states
such as Ethiopia and Nigeria are looking at AI as a driver
for economic growth through technology entrepreneur-
ship and innovation (Kopf 2018; Mumbere 2018). African
news outlets such as the New African magazine, part of the
influential African IC publications group, have promoted
AI development in Africa “as a leapfrogging opportunity to
prosperity”.2 Humanitarian organizations have seized upon
AI technology as an opportunity to lift poverty, eradicate
diseases, and eliminate hunger in the continent. For example,
the Action Against Hunger’s Modelling Early Risk Indica-
tors to Anticipate Malnutrition (MERIAM)3 project claims
to use AI to tackle the problem of malnutrition in Africa
(Smith 2017). In short, AI technology has been promoted as
a gateway to progress and prosperity in Africa.
This recent development has raised growing concerns
about AI governance in the continent and the potential
for AI to exacerbate problems of inequality and injustice
(Birhane 2020; Bjola 2021; Truby 2020; Wall etal. 2021).
* Yousif Hassan
yousifh@yorku.ca
1 Department ofScience andTechnology Studies, York
University, 4700 Keele Street, Toronto, ONM3J1P3,
Canada
1 https:// www. 2030v ision. com/ proje cts/ data- scien ce- africa
2 https:// newaf rican magaz ine. com/ opini ons/ ai- and- africa- leapf rog-
to- prosp erity/
3 https:// www. actio nagai nsthu nger. org/ meriam

1430 AI & SOCIETY (2023) 38:1429–1442
1 3
However, most of the literature on AI governance (Introna
2016; Just and Latzer 2016; Saurwein etal. 2015; Ziewitz
2016) is biased toward Euro-American centric perspectives
and appears to be looking at AI knowledge production prac-
tices and governance frameworks as universal technology
policies that can govern AI technology globally (Adams
2021; Mohamed etal. 2020; Peña1 and Varon 2019). For
example, most of the literature on AI ethics is focused on
algorithmic biases in AI models and forms of discrimina-
tion in AI automated systems in the context of advanced
industrialized economies, mainly in Euro-America (Bilić
2018; Zarsky 2016; Ziewitz 2016). Thus, the social, politi-
cal, and economic implications of AI in the Global South
remain understudied. Additionally, most AI literature on the
Global South, and particularly Africa (Bjola 2021; Gwagwa
etal. 2021; Hilbert 2016; Mann and Hilbert 2020; Wall etal.
2021), appears to be making normative claims about the
negative and positive impacts of AI. However, many Afri-
can AI researchers and practitioners point out to what they
call the lack of “African context” in AI research, develop-
ment, and governance (Asemota 2018; Cisse 2018b, p. 461;
Wairegi etal. 2021).
From this perspective and underpinned by racial and
colonial understanding of technoscience (Fanon 2008;
Smith 2012; Anderson 2002; Baker 2009; Subramaniam
etal. 2016; Abraham 2006) and the political economy of
technoscience (Birch 2017; Birch and Tyfield 2013; Tyfield
etal. 2017), this paper asks the question of how should Afri-
can researchers and policy makers approach AI governance
in the continent? To answer this question, I examine how
AI development is understood by various actors including
researchers and practitioners in Africa. In other words, what
are some of the visions and assumptions about innovation
that are influencing the development of AI in Africa? What
do African researchers and practitioners mean when they
talk about the lack of African context in AI development?
The raciality of computing examines the relation between
race, coloniality, and computing (Ali 2014, 2016; Chun
2009; Coleman 2009; Donnor 2005; Irani etal. 2010;
Ogbonnaya-Ogburu etal. 2020; Philip etal. 2012). This
theoretical approach highlights computing knowledge pro-
duction practices that are influenced by racial past biases and
marginalization of social groups that outlived colonialism
and are still persistent and reproduced by social, economic,
and political structures and agencies of both human and non-
human influenced by a coloniality of knowledge and imagi-
nation articulated through modernity (Ali 2014; Anderson
2002; Bhambra 2014; Harding 2011; Quijano 2000). There-
fore, the raciality of computing illuminates understandings
of AI technology that are linked to future visions of tech-
nological progress in Africa that are influenced by past and
postcolonial imaginations of Africa. The political economy
of technoscience examines the ethical, social, and political
dimensions of economies that are constituted, organized,
and configured by technoscience and how in turn technosci-
ence is shaped by these processes (Birch 2013; Tyfield etal.
2017). For example, the political economy of technoscience
provides a framework to examine understandings of AI that
are linked to market formations, economic development, and
innovation practices in Africa.
Drawing on the case of the African Master’s in Machine
Intelligence (AMMI)4 program at the African Institute for
Mathematical Sciences (AIMS),5 this paper examines some
of the ideas and visions underpinning the development of
this program. The AMMI program is an illustrative case
study because it is the first African program that’s focused
on developing AI talent and building AI capacity within the
continent and brings together different actors to articulate
their visions of AI in Africa. The vision of the founders of
the program is to develop a pan-African AI program that is
rooted in the African context and able to propel innovation
and economic growth while enabling Africa to eliminate its
colonial legacy.
The findings from the analysis of this paper suggest that
there are two related ideas about the lack of African context
in AI. The first idea refers to the lack of African data that
can inform AI technological development in the continent.
The second idea refers to the lack of African AI innovations
that are rooted in the local context but has the potential to
compete at the global scale. However, this paper emphasizes
that in approaching AI governance in Africa, it is crucial to
simultaneously consider the political imaginations underpin-
ning these two articulated ideas about the lack of African
context in AI and the different futures they attempt to imag-
ine and construct with science, technology, and innovation.
The contribution of this paper is, then, twofold. First, it
begins to articulate the need for examining AI development
in Africa through the analytical lens of the raciality of com-
puting and political economy of technoscience for deeper
understanding of the political imaginations that influence
this development. Second, it attempts to complement critical
perspectives on the burgeoning area of AI for development
(AI4D) by showing the necessity of centering political dis-
course in AI governance approaches. AI4D can be under-
stood as an emerging subset of the larger area of Information
and Communication Technology for Development (ICT4D).
The goal of the paper is not to devise a specific AI ethics
proposal but rather argues for an approach that locates nar-
ratives of AI technological innovation in Africa within their
colonial and postcolonial continuum in terms of notions of
progress and modernity using science, technology, and inno-
vation and centers the political imaginations of the African
4 https:// aimsa mmi. org/
5 https:// www. aims. ac. za/

1431AI & SOCIETY (2023) 38:1429–1442
1 3
AI communities alongside the materialities of AI technology
as an imperative for creating an effective governance model
of AI in Africa.
The paper proceeds as follows: first, I provide synthe-
sis on the relation between race, coloniality, and comput-
ing and show the different approaches in this area as the
decolonizing discourse has become more salient within the
AI ethics and AI4D literature. My goal with this synthe-
sis is to show that while the discourse on race and AI has
become increasingly predominant in the field, it is important
for racial analysis to be geopolitically situated to capture
the specificities the margins demand of such examination.
Additionally, I provide synthesis on the political economy
of technoscience focusing on innovation practices as they
relate to AI. I attempt to situate my discussion of both topics
within the literature of AI4D. I then discuss the development
of AI in Africa in the context of the AMMI program to high-
light the different ideas and visions of what it means to do
AI and innovation from Africa. I follow with a discussion
of these visions within the broader theoretical literature on
race and technology while bringing in perspectives from the
political economy of AI and innovation. I then argue for the
necessity to center the political imaginations of technology
in the discourse of AI ethics as an approach to think about
AI governance in Africa.
2 Raciality ofcomputing
In broader theoretical terms, computing as a sociotechnical
practice is interrogated in the context of racial analysis from
different perspectives including postcolonial, decolonial,
and intersectional analysis (Anderson 2002; Noble 2016;
Philip etal. 2012). Racial perspectives of technoscientific
knowledge production challenge the historical narratives and
context of modernity, link modernity to race formation and
distribution of power, reject notions of objectivity, neutral-
ity and universality, understand race as a social construct,
and political and economic classification system, and work
toward the elimination of racial oppression with the goal of
ending all forms of oppression (Cooper 2016; Donnor 2005;
Haraway 1990; Harding 2011; Subramaniam etal., 2016).
These approaches of racial analysis have informed criti-
cal scholarship of technoscientific knowledge production in
many areas related to digital technology including AI. For
example, there is a growing body of literature examining
AI from intersectionality perspective focusing on oppres-
sive and discriminatory knowledge production practices
including issues of racialized digital surveillance, racial
and gender algorithmic biases, reinforcements of inequali-
ties and marginalization of unrepresented social groups, and
issues of participation and exclusion of racialized communi-
ties in technoscientific research and innovation (Benjamin
2019a, b; Noble 2016, 2018; Roberts 2011, 2013). These
approaches highlight concurrent existence of racism and sex-
ism, making them visible, and showing that they are part of
the social structures and economies of contemporary soci-
ety including those of digital cultures, platforms, infrastruc-
tures, and so forth. It is important to note that there are also
critiques by non-intersectional scholars of the exclusion of
the perspectives of racialized groups in the development of
digital technologies and the normalization of White-male
values in AI technology discourse and design (Crawford
2016; Crawford etal. 2014; Zarsky 2016). However, schol-
arship based on intersectional approaches interrogates prac-
tices such as color-blind racism in AI systems. For example,
Benjamin conceptualizes forms of color-blind racism as the
New Jim Code: “the employment of new technologies that
reflect and reproduce inequalities but are promoted and per-
ceived as more objective or progressive than the discrimina-
tory systems of a previous era” (Benjamin 2019b, pp. 5–6).
Similarly, Noble (2018) conceptualizes the forms of color-
blind racism in the context of automated decision systems
as technological redlining: the design of digital technologies
that enact new modes of racial profiling that are underpinned
by neoliberal logics, values, and assumptions in a way that
reinforces oppressive social relationships. Intersectional
approaches frame AI sociotechnical practices as part of
digital technology assemblages that operate at the inter-
section of race, gender, class, power, sexuality, and other
socially constructed categories to create matrix of relations
that reinforce inequality and make oppression possible in
digital technology (Benjamin 2019a; Howard 2021; Kanai
2021; Noble 2016).
However, approaches by intersectional scholars are rooted
in the Euro-American context and remains connected to their
interlocutors in the West. For example, the concept of the
New Jim Code comes from a long history of racial discrimi-
nation against Black people in the United States (US) with
the Jim Crow racial segregation laws. The concept of tech-
nological redlining comes from a long history of housing
discrimination against Black people in the US. Furthermore,
intersectionality has emerged as a Black feminist theory in
the US and was concerned with forms of activism and civil
rights. Legal scholar Crenshaw (1989), who coined the term
intersectionality, argues that Black female are discriminated
against in ways that don’t fit the US legal system definition
of sexism and racism. Prominent intersectionality scholar
Collins (2008) argues that Black women are subordinated
within intersecting oppressions of race, class, gender, sex-
uality, and nation. She further conceptualizes the specific
“lived experiences” of African-American women and their
distinctive ways of knowing and understanding the world as
Black Feminist Epistemology, arguing that the politics of
race and gender influence knowledge production (Collins
2008, p. 251). From this perspective, intersectionality seeks

1432 AI & SOCIETY (2023) 38:1429–1442
1 3
an alternative to dominant knowledge production systems in
the Euro-American context.
On the other hand, critical perspectives of postcolonial
and decolonial computing tend to focus more on the Global
South (Dourish and Mainwaring 2012; Irani etal. 2010;
Philip etal. 2012). Postcolonial and decolonial approaches
have been used by scholars in many areas including infor-
mation and communication studies (Ali 2014; Dourish
and Mainwaring 2012; Irani etal. 2010; Philip etal. 2012)
and science and technology studies (Anderson 2002; Bon-
neuil 2000; Harding 2011) to examine issues in the broader
ICT4D area including computing, human–computer inter-
action, software and hardware design, among many other
areas. Both postcolonial and decolonial literature contend
several critical questions and concerns influenced by the
conditions of coloniality that are relevant to ICT4D projects
in the Global South. However, there are some differences
in terms of epistemology between postcolonial and decolo-
nial approaches (Bhambra 2014). Postcolonial computing
tends to be more cultural focusing on situated knowledge
production practices of computing and trying to bring post-
colonial sensibilities into the design and development of
Information and Communication Technology (ICT) (Ali
2014). More specifically, postcolonial approaches attempt
to address issues in technology development as it relates
to global connectivity and movement by engaging with
generative models of culture, looking at development as a
historical program, examining uneven economic relations,
and considering cultural epistemologies in the design and
development of technology, as articulated by Irani etal.
(2010) and Philip etal. (2012). In this sense, postcolonial
computing looks at postcoloniality as a project about “the
historical transformation of conditions of cultural encounter”
and understands technology research, design, and practice as
“culturally located and power laden” (Irani etal. 2010, pp.
1311–1312). However, postcolonial computing approaches
have been criticized as being grounded in Western episte-
mologies including the critique of modernity from Eurocen-
tric perspective and only concerned with how postcolonial
theory can inform technology design and development (Ali
2014; Bhambra 2014).
On the other hand, decolonial computing looks at com-
puting as inherently colonial practice that is influenced by
existing economic asymmetries, uneven global structural
and institutional power, and colonial relations and episte-
mologies that continue to persist and inform contemporary
computing practices (Ali 2016; Dourish and Mainwaring
2012). Decolonial computing scholars are more con-
cerned with critiquing the historical origins of computing
areas and the epistemologies that inform their knowledge
production practices (Ali 2014). From this perspective,
decolonial computing tends to foreground the geopoliti-
cal and the political orientation and the positionality of
those practicing and researching computing. For example,
Ali (2016) argues that decolonial computing is a way to
think through what it means to design and develop com-
puting technology for and with those in the margins of the
world systems using epistemologies and ways of knowing
situated in the peripheries while attempting to decenter
Euro-American centric universals. Scholars (Ali 2014,
2016; Mohamed etal. 2020; Peña1 and Varon 2019) have
extended decoloniality to different areas of computing
practices; however, decolonial computing remains under-
theorized (Ali 2016) and lacks the geopolitical specificities
of the African context.
In response to the recent interest in AI in Africa, there has
been an increase in the literature on AI4D examining differ-
ent issues of AI by African scholars from different perspec-
tives including decoloniality. There are two widely popular
views that exist in this literature. One view challenges AI4D
projects and sees them as a new form of imperial domination
and Western hegemony (Birhane 2020; Kwet 2018). The
second view looks at the role of AI with less critical view on
the agency of the local population, and role of the state and
transnational corporations (Gwagwa etal. 2021; Mann and
Hilbert 2020; Siminyu etal. 2020). Both views include nor-
mative ideas about the implications of AI technology with
the main question being how AI impacts economic devel-
opment and social justice in Africa. Conversely, another
emerging literature tends to emphasize the racialized and
colonial nature of AI and engage more with decoloniality
(Mohamed etal. 2020; Peña1 and Varon 2019). For exam-
ple, Mohamed etal. (2020) argue for the use of decolonial
theory as critical science and to focus on values and power
in AI as its two critical pillars from which to establish ethical
principles while centering vulnerable communities in the
Global South and elsewhere. They contend that embedding
decolonial approach within technical practice of AI is cru-
cial to developing foresight and tactics that can reduce the
negative impact of technological innovation on marginalized
communities.
In short, while the discussed approaches overlap in terms
of their social projects, however, they differ in their episte-
mologies which influence the ways in which ethical issues
in AI can be approached across different geographies. For
example, intersectionality appears to be focused more on the
Euro-American context, while postcoloniality and decoloni-
ality are increasingly mobilized to examine AI ethics from
a Global South perspective. However, most of the existing
literature on both areas doesn’t seem to problematize the
normative notions of ethics and intelligence. Adams (2021)
argues that AI ethics is based on colonial logics of rational-
ity and Euro-American centric conceptions of ethics and
intelligence. There’s a dearth of literature that looks at the
historical origins of these concepts at the epistemological
level and attempts to read AI with histories of colonialism,

1433AI & SOCIETY (2023) 38:1429–1442
1 3
which is imperative for a critique of universalist approaches
to AI ethics.
2.1 The political economy ofAI
The political economy of technoscience literature examines
the impact of AI technology on several economic registers
including labor and market formation and reconfiguration.
From this perspective, the political economy of technosci-
ence focuses on how AI and data practices are impacted
by the ethical, social, and political dimensions of economic
practices and how in turn these economic practices are
impacted by AI technology including issues around the dis-
tribution of power and wealth in society (Birch 2013; Tyfield
etal. 2017). For example, the literature on this topic looks
at how AI is increasingly embedded in the structure of the
emerging digital economy where AI technology is weaved
into the fabric of modern digital systems that power markets
ranging from trading systems to assembly lines in factories
among many other applications of AI, data, and algorithmic
logic (Bilic 2016; Birch 2017; Leonelli 2016; MacKenzie
2017; Peters 2017; Srnicek 2016).
However, the study of the digital economies of the mar-
gins, and particularly Africa, is lacking (Graham 2019).
Most of the literature on the political economy of techno-
science appears to be focused more on advanced industrial
economies and lacks the theoretical and empirical specifici-
ties of the margins. An example of this literature includes
scholarly work to theorize and understand how economic
value is extracted from data (Bilić 2018; Birch 2020; Zuboff
2019). This work highlights processes of capitalization and
assetization of data in modern technoscientific formations
and places this technological shift within a modern capital-
ist system that is characterized by its increasing reliance on
technoscience, data, and algorithms as part of its scaling and
value extraction processes (Birch 2017, 2019; Zuboff 2019).
Another scholarly work in this area looks at how algorithms
leverage different technological innovations such as automa-
tion and data to operate at a global market scale and links
issues of AI ethics such as fairness, inclusion, and equality
to data ownership, loss of control, and regimes of intellectual
property that are tied to data appropriation, capitalization,
and financialization practices of technoscientific capital-
ism (Birch 2020; Leonelli 2016; Peters 2017). For example,
Srnicek (2016) argues that in platform capitalism digitaliza-
tion is changing capitalist mode of production using digital
platforms as intermediaries for digital economic circulation
using the data generated by these platforms. In addition, the
literature on this area looks at how capitalism has turned to
data, amid long decline in traditional manufacturing profit-
ability, as one way to maintain economic growth, and ensure
the vitality and increase the productivity of capitalist sys-
tems. Srnicek (2016) argues that digital platforms are a new
business model that is capable of extracting and controlling
immense amounts of data, resulting in new monopolies by
large tech firms. While this literature on the broader political
economy of technoscience might have some resonance to
audiences of the digital economy in the Global South, how-
ever, there is a dearth of literature that examines ICT as an
economic practice in Africa (Ojo 2018). Thus, the political
economy in the broader area of ICT4D and more specifically
AI4D remains understudied.
However, an area of interest that could be highly relevant
to a discussion of the political economy of AI in Africa is
concerning the emerging innovation ecosystem in the con-
tinent. The concept of innovation ecosystem is an emerging
and fragmented concept and does not have a widely accepted
definition (Kivimaa etal. 2017). Oksanen and Hautamäki
(2014) describe the innovation ecosystem as “an interactive
network that breeds innovation”. In practice, this network
is comprised of local innovation hubs, global networks, or
technology platforms (Oksanen and Hautamäki 2014, p.
4). An innovation hub is described as “a region or a place
with an extraordinary amount of accumulated knowledge
and innovativeness” (Oksanen and Hautamäki 2014, p.
4). A significant literature on this area by African scholars
discusses AI impact in the context of the fourth industrial
revolution and shows that it cannot be decoupled from the
wider trends and practices of technological innovation in AI
(Jegede and Ncube 2021; Madden 2020; Ndung’u and Signé
2020; Nyagadza etal. 2022). The fourth industrial revolution
is a term used to describe a paradigm shift in capitalist mode
of production that is caused by the deployment of cyber-
physical systems and the ubiquitous connectivity of billions
of people and things such as sensors and a plethora of data
sources and digital objects from mobile phones to cars and
so forth, also known as the “Internet of Things” (Schwab
2017). This shift employs current technological advances
in AI, connectivity, and data.
On the other hand, the broader literature on innova-
tion practices in Africa contains two widely discussed
views on the impact of AI and digital innovation in Africa
(Birhane 2020; Bjola 2021; Kwet 2018; Wairegi etal.
2021). Some scholars are concerned with the risks and
benefits associated with AI adoption and its application
for economic development (Bjola 2021; Mann and Hilbert
2020; Wairegi etal. 2021; Wall etal. 2021). This literature
provides more of a normative and instrumental view on
the social and economic impact of AI technology in the
continent. On the other side, critical literature from other
African scholars links innovation practices to new forms
of digital colonialism practiced by Western transnational
corporations in order to increase capital accumulation and
wealth concentration within bigtech and corporate monop-
olies (Birhane 2020; Kwet 2018; Madianou 2019). This lit-
erature is concerned with the deployment of technological

1434 AI & SOCIETY (2023) 38:1429–1442
1 3
innovations practices as a new form of domination, power,
and control using algorithmic logic for profit maximizing
at any cost including the appropriation of human soul,
behavior, and action (Birhane 2020). This literature over-
laps with a more general body of scholarly work linking
data practices to colonial practices around resource appro-
priation and subject formation (Couldry and Mejias 2019a,
b; Phan and Wark 2021). For example, Couldry and Mejias
(2019a) argue that through processes of data relations, as
an emerging social form of human relations enabled by
data as potential commodity, new mode of colonialism
(i.e., data colonialism) is enacted by big data and digital
cloud platforms that is dependent on the normalization and
exploitation of human beings through data in similar ways
to historical processes of colonialism in its appropriation
of territory, resources, and ruled subjects for profit. This
literature focuses on the social relations that are produced
by processes of data colonialism that configure a new con-
temporary data-colonized subject (Coleman 2019; Coul-
dry and Mejias 2019a; Mbembe 2017b). The notion of
data-colonized subject connects with other African stud-
ies literature that problematizes the relation between race,
colonialism, and data practices. This literature attempts
to show the constitutive nature of the relation between
race, colonialism, and old and new forms of capitalism
including the digital. For example, prominent African
scholar Mbembe (2017b) points out that early capitalism
is marked by earlier conceptions of race and Blackness
within the context of the transatlantic slave project where
colonialism and slavery constituted one of the most vio-
lent forms of resources and human labor appropriation in
early capitalism. He argues that the current moment of
neoliberalism, a wave characterized by the globalization of
markets and privatization of the world, and the domination
of powerful Silicon Valley and tech giants on the global
economy, represents another phase in the co-evolution of
race and capitalism which he calls the “becoming-black-
of-the-world”. Mbembe (2017a, b) argues that this marks
the globalization of Blackness, pointing out to the ten-
dencies of digital capitalism to turn everything into data
that can be appropriated and extracted. This literature
highlights the emergence of a new universal imagination
around subject formation and race conception in assem-
blages of digital technologies including AI.
While the discussed literature by African scholars inter-
sects with the broader literature on the political economy
of technoscience around data appropriation and exploita-
tion such as (Birch 2020; Madianou 2019; Zuboff 2019),
however, it is grounded on different epistemological
approaches such as race and colonial theories and their
connection with economic practices around data. The aim
of this literature is to open up analytical and theoretical
possibilities for a different theorization of data and its
economic practices in contemporary digital formations
from the South.
2.2 Artificial intelligence development inAfrica
In this section, I draw on the case of the AMMI program to
explore some of the ways in which AI is taken up in Africa.
This case study is not intended as a complete representation
of AI discourses across the continent; however, the analysis
of this case highlights some of the findings of an ongoing
project examining the issues surrounding the development
of an AI ecosystem in Africa. Therefore, I analyze a lim-
ited data set that includes online content and discourses,
interviews, and articles written by AMMI founders and
other actors connected to the development of this program.
I provide a brief background of AMMI and attempt to high-
light two aspects of its development related to the research
question. First, I try to illuminate the ideas and assump-
tions about AI technology and innovation underpinning the
development of this program. Second, I aim to underline
the understanding of the lack of African context in AI as
articulated by AMMI founders and others.
The AMMI program is established by AIMS, a teaching
and research institute originated in South Africa, and has a
network of associated institutes in Senegal, Ghana, Cam-
eroon, Tanzania, and Rwanda with interest in fostering AI
research and talent in Africa. AIMS launched the AMMI
program in Kigali, Rwanda, in partnership with Google and
Facebook (Lijadu 2018). This development may be viewed
skeptically as another way for Western companies to exploit
the continent or as a new form of digital colonialism and
Western hegemony given the asymmetrical power rela-
tion (Birhane 2020; Coleman 2019; Kwet 2018; Madianou
2019). Conversely, it may be viewed favorably as an oppor-
tunity to advance the agenda on technological development
in the continent and as a pathway toward social and eco-
nomic development (Bjola 2021; Gwagwa etal. 2021; Mann
and Hilbert 2020). These binary visions have always existed
in the debate about technology transfer and international
development programs in the continent; nevertheless, both
views contain normative claims about the benefits and risks
of technology transfer. However, the aim of this case study is
not to promote one view or the other but to rather highlight
what seems to be missing from this debate, which is a deeper
engagement with the political dimensions and messy reali-
ties of technology transfer in Africa.
The AMMI founders articulate a vision of building AI
capacity in the continent to address issues related to eco-
nomic development and AI governance in Africa. Similar to
other discourses of ICT4D in Africa, the development of AI
is shaped by lived experiences of key actors and narratives of
inequality and development disparities that long dominated
the debates about the role of ICT in propelling the continent

1435AI & SOCIETY (2023) 38:1429–1442
1 3
forward (Powell 2001; Unwin and Unwin 2009). One of the
key actors in the development of AMMI is Dr. Neil Turok,
a South African theoretical physicist and appointed Officer
of the Order of Canada,6 who founded AIMS in 2003. AIMS
launch was supported by the Global Outreach Initiative of
the Perimeter Institute for Theoretical Physics in Canada,7
where Dr. Turok was its president until February 2019. In
an interview with Sciences et Avenir, Dr. Turok indicated
that he was inspired by his parent’s, anti-apartheid white
activists, quest for social justice and equality in Africa. Dr.
Turok said:
“Once in the academic world, my father told me that it
was my turn to take action against inequality, I chose to
do it at my level with what I know best to do: to teach.”
(Sermondadaz 2018)
Inequality in the Global South, and particularly Africa,
has been the battle ground for technology transfer over the
years (Ojo 2018; Powell 2001). However, one of the dif-
ferent characteristics of inequality discourse in this case is
a tendency to go beyond issues of the digital divide that
dominated ICT4D (Parayil 2005) and shift more toward nar-
ratives of increased participation in the basic research and
innovation practices of AI technology globally. For example,
Dr. Turok emphasizes that his main objective is to try to
get rid of the colonial legacy once and for all in Africa by
establishing a pan-African scientific research agenda, and
develop African research with its own centers of excellence”
(Sermondadaz 2018). He describes AIMS vision as closing
the gap in local expertise. Dr. Turok said:
“This is an issue for basic research […]. This lack of
local expertise is critical, and Westerners have failed
to diagnose it correctly. We want to break the colo-
nial legacy and demonstrate that Africa has the same
potential as any other continent” (Sermondadaz 2018)
What Dr. Turok is referring to are Western actors such
as international development agencies and Western nation
funders. Historically, most development projects in Africa
related to ICT have been conceived, designed, and funded
by Western international development agencies. Recently,
there has been a noticeable involvement by transnational
corporations including bigtech. However, it is important to
contextualize what Dr. Turok is pointing at against the back-
drop of the current shift in development discourse toward
development ownership, albeit criticized by many develop-
ment studies scholars as being superficially applied in prac-
tice (Harper-Shipman 2019; Overton 2019). Development
ownership is a new approach that attempts to give more
agency to local actors and sovereign states in influencing
the development agenda, resources, and outcomes. This is
in response to increased critiques that postcolonial develop-
ment practices in international aid and development includ-
ing ICT4D are reinforcing the colonial legacy in the Global
South. However, what this paper tries to emphasize is that
the impact of the shift toward local ownership on AI4D may
need to be further examined at the confluence of the coloni-
ality of power (Quijano 2000) and the desire for grounding
AI innovations on the local context.
On the other hand, unlike dominant development dis-
course that continues to be more focused on vocational train-
ing in Africa (McGrath etal. 2020), the visions of AMMI
founders have explicit assumptions about creating a path of
technological future based on increased African capacity
for technoscientific research and development. For example,
Dr. Turok views the AMMI in Kigali, Rwanda, as the first
center of excellence focused on AI and big data with its first
Master’s program dedicated to AI questions, as AIMS vision
is to create similar centers across Africa. Dr. Turok’s vision
is to use this research to apply data processing to areas such
as astrophysics with the goal of preparing the continent for
quantum computing (Sermondadaz 2018). In this interview,
Dr. Turok said:
“Our challenge is to first train students for research.
Among our graduates, 50% will work in the academic
world, and 50% in the industry. By training them only
for the private sector, we would miss out on great sci-
entists” (Sermondadaz 2018)
While the AMMI program seems to be concerned with
the colonial legacy of the continent and histories of mar-
ginalization of Africa in the global economy, it attempts to
recast Africa as an equal global contributor to AI, signaling a
shift toward a globalized perspective in ICT4D and inequal-
ity discourse in the continent. There are explicit assumptions
about African contribution to the global development of AI
in order to address issues of inequality at the global level.
For example, Dr. Moustapha Cissé, another key actor and
cofounder of the AMMI program who worked for Facebook
Research before moving to Google AI where he was the
Head of Google AI Center in Accra, Ghana, said:
“The lack of Machine Intelligence researchers from
Africa means that many opportunities to use Machine
Intelligence to create a better and more stable world are
being missed. […] If Africa continues to be bypassed
by Machine Intelligence, a rare opportunity to alleviate
global social and economic disparities will be missed”
(Cisse 2018a).
Simultaneously, AMMI founders point out to the lack of
African context in AI research and development as a major
6 https:// www. idrc. ca/ en/ news/ neil- turok- appoi nted- offic er- order-
canada
7 https:// www. perim eteri nstit ute. ca/ outre ach/ global- outre ach

1436 AI & SOCIETY (2023) 38:1429–1442
1 3
challenge to materializing their visions. Dr. Cissé defines
the lack of African context in terms of deficit in local data
sets. In an interview during the AI for Good Global Sum-
mit in Geneva, Switzerland in 2017, Dr. Cissé articulated
this issue by pointing to the lack of AI solutions driven by
local data sets focused on problems that are not related to
dominant White male perspectives in Western societies.8 Dr.
Cissé argues that “the challenges we choose to work on are
strongly influenced by our backgrounds and our environ-
ment” (Cisse 2018a). AMMI attempts to position itself as a
means to confront this challenge. In another interview with
IT Web Africa, a South African tech news website, Dr. Cissé
said:
“We have already put the foundation of some aspects
of machine learning at this centre. We want to tackle
some important challenges that are critical to the Afri-
can context.” (Moyo 2019)
The concern about the lack of African context in AI is
also articulated by Victor Asemota, an African tech pioneer,
who describes his vision as harnessing the knowledge and
experience of African professionals globally and transform-
ing Africa into a global technology powerhouse.
“the African Context when added to Machine Learn-
ing and AI has the potential of bringing up new solu-
tions for humanity and create new markets” (Asemota
2018).
In an article for CNN Africa, Mr. Asemota stressed that
Africa can participate and make an impact on AI develop-
ment globally. Similarly, other African AI practitioners
such as Dr. Omoju Miller, who worked for GitHub in San
Francisco, look at the African context in terms of African
innovations. Dr. Miller said:
“The African context is very different from Mountain
View or Zurich. The kinds of innovation Africa needs
are similar to the innovative practices we have seen
coming out of China, with companies like Alibaba and
Ant financial. These are companies that are responsible
for inventing entire markets that serve the world. That
is the kind of innovative thinking we need in Africa.
Google AI can play a role in helping us get there”
(Asemota 2018)
Dr. Cissé cofounded AMMI under the banner of Google
AI research in collaboration with Facebook. The two bigtech
companies have been discreet in disclosing their funding
and financial commitment to support the program. However,
Digest Africa, a financial information service for investing
in African start-ups, reported that Facebook has committed
$4M to launch the AMMI program. On the other hand,
Google has committed its AI resources behind the AMMI
program including its research center in Accra. Jeff Dean,
the Head of Google AI, indicated that both the partnership
with AIMS and opening AI center in Accra are examples of
Google’s long-term commitment to advancing AI in Africa.
The visions articulated by the actors in this case study can
be understood within the context of the global tendencies in
innovation practices to follow the model of regional inno-
vation ecosystems (Pfotenhauer and Jasanoff 2017). While
the literature on innovation ecosystems is scarce, however,
innovation scholars such as Oksanen and Hautamäki (2014)
argue that the key components of innovation ecosystems are
a group of local actors, dynamic processes, entrepreneurial
culture, finance providers, large established companies, new
start-ups, customers, top-level universities, and research
institutions. On the other hand, Oksanen and Hautamäki
(2014) highlight the importance of the utilization of local
knowledge and competencies; however, they emphasize the
connectedness of the innovation hub to a global value net-
work and its ability to create value in the global economy.
In this sense, AMMI can be seen as a key component in a
network of local knowledge and expertise in this regional
innovation infrastructure. For example, Dr. Cissé said:
“A network of African institutes of artificial intelli-
gence, for example, could retain the best talents on
the continent, enlist world-class African scientists to
tackle AI challenges in the African context and col-
laborate with existing academic institutions” (Cisse
2018b, p. 461)
In summary, the different articulations of the lack of Afri-
can context in AI and their underlying visions and assump-
tions highlight the inherent tensions between the local and
the global and the desire for situatedness against a back-
drop of increased universality in AI innovation discourse.
While the visions of AMMI founders attempt to decenter the
past and move toward a future of decolonized AI, however,
there has to be a recognition that these are issues related to
political imaginations of the different AI communities in
the continent about what it means to do AI from Africa. In
the next section, I attempt to draw attention to the necessity
of centering political imaginaries of AI in the decolonizing
discourse and innovation practices in the African context as
an approach to think about AI governance in Africa.
2.3 From ethics toimaginaries
The discourse on ethics has dominated the field in recent
years in response to many discontents with AI including
algorithmic biases and their impact on reinforcing inequal-
ity and discrimination against marginalized and underrepre-
sented social groups. However, AI ethics has been criticized
8 https:// www. youtu be. com/ watch?v= seHc3 QyDDtc

1437AI & SOCIETY (2023) 38:1429–1442
1 3
for universalist approaches to AI ethical frameworks that are
grounded on Euro-American knowledge production prac-
tices (Ananny 2016; Hagerty and Rubinov 2019; Phan etal.
2021). As a result, there have been efforts by scholars from
the Global South to challenge Euro-American norms and
values, and attempts to decolonize the world of AI ethics
(Mohamed etal. 2020; Peña1 and Varon 2019). However,
with very few exceptions such (Adams 2021), the decolo-
nizing AI ethics discourse does not seem to problematize
the historical origins and epistemological underpinnings of
ideas such as ethics and intelligence. While some literature
such as Carman and Rosman 2021; Kiemde and Kora 2021;
Nandutu etal. 2021 attempts to challenge Euro-American
norms and values, however it reproduces another normative
and instrumental view of AI ethics in the African context.
More importantly, there is very little discussion about
what kind of futures these ethical frameworks are trying to
imagine and the performativity of their underlying political
imaginations in AI sociotechnical practices (Jasanoff 2016).
Nevertheless, these discussions are mobilized in the field
today to influence technology policy proposals and fix the
broken world of AI globally. However, Kim (2017) argues
that technology policies don’t get replicated between geog-
raphies or different countries because regulatory frameworks
are influenced by the imaginations of every nation about the
future and how social order ought to be. Kim (2017) refers to
how sociotechnical imaginaries influence technology poli-
cies and the need to situate these imaginaries in the local
context. Jasanoff and Kim (2015) describe sociotechnical
imaginaries as “collectively held, institutionally stabilized,
and publicly performed visions of desirable futures, ani-
mated by shared understandings of forms of social life and
social order attainable through, and supportive of, advances
in science and technology” (p. 6). From a theoretical per-
spective, the sociotechnical imaginaries framework analyzes
how political cultures shape the understandings of the risks
and benefits of technology and to what degree these visions
and ideas influence the fundamental ethical questions under-
pinning these technologies. In this sense, the sociotechni-
cal imaginaries framework analyzes the co-production of
technoscientific knowledge and the political materiality
of technoscience (Jasanoff 2004; Jasanoff and Kim 2009).
Co-production is the idea that social order is produced by
both material and discursive resources (Jasanoff 2004). The
sociotechnical imaginaries framework has been mobilized in
many areas including studies of smart cities (Miller 2020),
energy transitions (Ballo 2015), fourth industrial revolution
(Avis, 2018; Vicente and Dias-Trindade 2021), digital plat-
forms (Hassan 2020), and ICT in Africa (Bowman 2015) and
more to understand the political dimensions of technological
development and innovation. The use of the sociotechnical
imaginaries framework in this paper is meant to show its
usefulness in thinking about the political dimensions of AI
development when considering governance approaches to AI
in Africa, and more broadly the Global South.
From a politics of technology perspective, what seems
to be understudied in the literature of AI4D and the larger
ICT4D is an understanding of development as a state build-
ing project in postcolonial Africa (Bonneuil 2000). AI4D
development projects are not just about AI and innovation,
but they are equally about state building and the political
imaginations of Africans about the future in the continent.
Bonneuil (2000) argues that development regimes were
established by former colonial power as way of managing
the African environment and generating knowledge about
African societies. However, Western development ideologies
and practices played an import role in the construction of the
African state and ironically they fueled at later stages the
interest of colonial power in the values of indigenous knowl-
edge. Krige and Wang (2015) argue that the relation between
science, technology, and nation-building were laid down by
knowledge that was mobilized and adapted to local condi-
tions by established power structures. However, they point
out that opposition and resistance to the imposed knowledge
and values drew from local knowledge to construct alter-
native visions of the future. Krige and Wang (2015) argue
that knowledge was essential to the performance of power
in nation building, and the fashioning of legitimate and
legitimizing certain visions of the future. This included the
spectacular representation of advanced technologies and the
sophisticated exploitation of new political and legal sources
of legitimization (Krige and Wang 2015).
Integrating the racial and colonial understanding of com-
puting and the political economy of technoscience with the
political dimensions of AI, there are a couple of intertwined
threads in the discussion of AI in Africa that are worth pur-
suing to examine the development of AI in Africa from a
sociotechnical imaginaries perspective. They help in open-
ing up some possibilities to answer the question of how to
approach AI governance in the continent. The first one is
related to the notion of decolonizing AI and the lack of Afri-
can data sets. The second one is related to what innovation
practices mean in the local African context.
On the decolonizing register, Adams (2021) critiques the
way in which the notion of decoloniality is mobilized in AI
today and argues that decolonial theory has been applied
to only broaden the critique of AI. She asks the question of
whether AI can be decolonized given the way decolonization
discourse has been taken up in the field. For Adams (2021),
the important question is what does AI come to be because
of histories of colonialism. She argues this is critical to avoid
reproducing the same problematics that decoloniality set out
to disrupt in the first place. Previously, in a similar vein,
Tuck and Yang (2012) argue that decolonizing has become a
metaphor and point out to the need to decenter the narratives
by which colonial power romanticizes indigenous beliefs and

1438 AI & SOCIETY (2023) 38:1429–1442
1 3
instead deconstruct the colonial structures that continue to
oppress the former colonies. From this perspective, decol-
onizing AI needs to avoid recasting colonial futures and
instead seeks recentering the futures of indigenous, Black,
unrepresented and marginalized social groups. The founders
of the AMMI program call for increased and equal African
participation in the development of AI technology globally.
While this is expected given the language du jour in AI eth-
ics, however, what seems to be missing is an emphasis on
the situatedness of this demand and lack of explicit engage-
ment with African political imaginations that seek to narrate
Africa’s own versions of the futures from within. Access and
participation calls need to be grounded on situated knowl-
edge practices and epistemological critique of genealogies
of innovation, intelligence, and ethics that come to dominate
the AI field and share their roots in the colonial histories of
expansionist Western power that saw their military and tech-
nological supremacy as evidence of the efficacy of Western
sciences and technologies (Elshakry 2010). For example, the
ideas of the globalization of AI technology may have their
roots in the colonial history of science and technology and
the understanding of Western technology as universal and
global technology.
On the innovation register, the vision of the AMMI
program is to focus on capacity building and foster AI
research and talent in Africa through an expanding net-
work of regional innovation and centers of excellence.
Research from GSMA Ecosystem Accelerator in 2019
shows that the number of tech innovation hubs in Africa
is rapidly increasing with 618 active regional innovation
hubs across Africa, representing 40% increase from 442
hubs in 2018 (Giuliani and Ajadi 2019) and 314 hubs in
2016 (Boucher 2016). The development of these inno-
vation hubs in Africa can be understood through the
analytical lens of the sociotechnical imaginaries as the
“practice turn” in innovation (Pfotenhauer and Jasanoff
2017). This refers to a global shift in benchmark under-
standing of innovation and tendency to circulate innova-
tion globally based on a “best-practice model” (Silicon
Valley and MIT models of the future) on a regional basis.
Pfotenhauer and Jasanoff (2017) argue this is another way
sociotechnical imaginaries manifest themselves in a local
context as “traveling imaginaries of innovation,” where
“models in contemporary innovation discourse add a
dimension of global circulation to capture how innova-
tion policy simultaneously mobilizes local understand-
ings of what constitutes a desirable sociotechnical future
and a set of transnational practices that legitimize inno-
vation as a global policy imperative” (p. 417). From a
political economy perspective, AI development in Africa
is connected to Western efforts of globalizing AI. This
is evident by the involvement of global tech companies
such as Amazon, Facebook, and Google in advancing
AI in Africa, with the AMMI program being part of this
global interest (Benjy 2018). Western nation states such
as Canada, Sweden, and Germany are among the contribu-
tors to global AI research and AI4D programs in Africa.
Mbembe (2017a, b) argues that global capitalism is mov-
ing into two directions: increasing the exploitation of large
parts of the world, similar to what Marx conceptualizes
as primitive accumulation (Bonefeld 2011), and increas-
ing the rate of innovation and invention. Technoscientific
capitalism introduces new forms of value extraction and
capital accumulation through the appropriation of human
life and human experience through data underpinned by
technoeconomic assumptions and practices of neoliber-
alism forging new connections of inequalities of wealth
and power (Couldry and Mejias 2019b; Mbembe 2017a;
Zuboff 2019).
The understanding of these innovation practices through
the analytical lens of racial and colonial technoscience and
their impact on political economies of AI in the margins
is critical to understanding how the AI innovation story
impacts a sociotechnical imaginary in the continent that
attempts to fashion its own African futures. In another way,
what might be at stake here is the opportunity to engage
simultaneously with innovation discourse with all of its
flaws and tendencies to universalize technoscientific prac-
tices while seeking alternative narratives of what come to be
understood as modernity from its undersides as opposed to
its centers of power and affluence (Comaroff and Comaroff
2012). For example, the call for doing AI in the African
context while attending to the lack of African data sets
requires serious engagement with practices of data coloni-
alism. Couldry and Mejias (2019a) point out to how the
characteristics of European modernity such as absolute uni-
versal rationality are reproduced in data colonialism through
“its logics of universal data extraction and management of
human beings through data” (p. 346). An African imaginary
of AI demands what Couldry and Mejias (2019a) calls “epis-
temological decolonization,” that resists AI data colonialism
by rejecting the naturalization of data collections as a form
of social knowledge but understands it as a “commercially
motivated” form of extraction that advances particular eco-
nomic and governance interests.
In summary, what I wanted to show through this discus-
sion is the need to go beyond the ethics of AI and to look at
the imaginaries of AI in Africa to understand the entangle-
ment of race, economy, science, technology, and innova-
tion with development as a state building project, and the
political imaginations of Africans in the co-production of the
future. In this sense, the ideas and visions about AI develop-
ment in Africa become a sociotechnical imaginary of decol-
onizing AI that influences and shapes African’s conceptions
about the futures of Africa. In other words, decolonizing AI
may be understood as a sociotechnical imaginary in Africa.

1439AI & SOCIETY (2023) 38:1429–1442
1 3
3 Conclusion
Drawing on the discourse of AI in Africa, the notion of
the lack of African context in AI seems to influence many
of the ideas and visions of AI in the continent. While it
has no specific definition, however, two articulated ideas
about the African context emerge from the discussion. The
first is articulated by African researchers and refers to the
lack of African-specific data to inform AI research and
solutions to specific African problems. The second idea is
articulated by African entrepreneurs and refers to the lack
of African innovations that are originating from Africa and
done by African companies that can compete at the global
level and create new markets.
The articulated visions in the discourse of AI in Africa
heighten the tension between universalist approaches in
the globalization of innovation practices and the necessity
of situated knowledge production and local imaginations
of technoscientific futures. What at stake is how Africans
think about the different ways in which to produce their
own futures with AI technology and the desire to have an
equal voice in spaces built with legacies of colonial struc-
tures that are still persistent in the political economies of
global innovations with all their asymmetries of wealth
and power. In another word, the question of AI governance
in Africa, and more broadly the Global South, is in fact a
question about the different visions of desired futures and
the political imaginations that sustain them with science,
technology, and innovation.
In attending more closely to the question of how to
approach AI governance in Africa, AI researchers, prac-
titioners and policy makers need to resist normative and
instrumental views of how AI should fit into the Afri-
can context. They need to consider the political cultures
that infuse certain visions of AI in Africa. This requires
much problematization and destabilization of the idea of
the African context itself. The understanding of AI as an
imaginary centers political imagination of Africans and
highlights the contestation of different understandings
of the African context and what it means to do AI from
Africa. While imaginaries don’t represent policy propos-
als, nevertheless they help us make sense of their material
and organizational structure and resources (Jasanoff and
Kim 2009). Thinking through the sociotechnical imaginar-
ies invites the African AI community to pay attention to
the political imaginations that are underpinning visions
of AI, but it also highlights that an AI from Africa needs
institutional building more than anything. This requires
serious engagement with local political communities to
activate their imaginations and begin to ask serious ques-
tions about what kind of futures and social orders that
Africans desire out of AI. This also requires political
commitment by African nation states not only for fund-
ing but also for building democratic institutions that are
capable of bringing the multiplicity of visions of the citi-
zens and different AI communities to influence the kinds
of policies and governance frameworks that are required
to have an African AI project that engages seriously with
issues of social justice at the center of technology policy
discourse.
Funding The author did not receive any funding for this article.
Data availability All references and documents are available.
Code availability N/A.
Declarations
Conflict of interest There is no conflict of interest.
References
Abraham I (2006) The contradictory spaces of postcolonial techno-
science. Econ Pol Wkly 41(3):210–217
Adams R (2021) Can artificial intelligence be decolonized? Interdisc
Sci Rev 46(1–2):176–197. https:// doi. org/ 10. 1080/ 03080 188.
2020. 18402 25
Ali M (2014) Towards a decolonial computing. Ambiguous Technolo-
gies: Philosophical Issues, Practical Solutions, Human Nature,
28–35. http:// oro. open. ac. uk/ 41372/. Accessed 19 April 2019
Ali M (2016) A brief introduction to decolonial computing. XRDS
22(4):16–21. https:// doi. org/ 10. 1145/ 29308 86
Ananny M (2016) Toward an ethics of algorithms: convening, obser-
vation, probability, and timeliness. Sci Technol Human Values
41(1):93–117. https:// doi. org/ 10. 1177/ 01622 43915 606523
Anderson W (2002) Introduction: postcolonial technoscience. Soc Stud
Sci 32(5/6):643–658
Asemota V (2018) Ghana is the future of Africa: Why Google built an
AI lab in Accra. CNN. https:// www. cnn. com/ 2018/ 07/ 14/ africa/
google- ghana- ai/ index. html. Accessed 16 Oct 2018
Avis J (2018) Socio-technical imaginary of the fourth industrial revolu-
tion and its implications for vocational education and training:
a literature review. J Vocat Educ Train. https:// doi. org/ 10. 1080/
13636 820. 2018. 14989 07
Baker B (2009) CHAPTER 17: postcolonial technoscience, toleration
and anti-imperialism, and education and psychology. Counter-
points 369:289–308
Ballo IF (2015) Imagining energy futures: sociotechnical imaginaries
of the future Smart Grid in Norway. Energy Res Soc Sci 9:9–20.
https:// doi. org/ 10. 1016/j. erss. 2015. 08. 015
Benjamin R (2019a) Race after technology: abolitionist tools for the
new jim code (1 edition). Polity Press.
Benjamin R (ed) (2019b) Captivating technology: race, carceral tech-
noscience, and liberatory imagination in everyday life. Duke
University Press
Benjy B (2018) African Institute for Mathematical Sciences Launches
African Master’s in Machine Intelligence Program. African
Leadership  Magazine.https:// afric anlea dersh ip. co. uk/ afric
an- insti tute- for- mathe matic al- scien ces- launc hes- afric an- maste
rs- in- machi ne- intel ligen ce- progr am/. Accessed 19 April 2019

1440 AI & SOCIETY (2023) 38:1429–1442
1 3
Bhambra GK (2014) Postcolonial and decolonial dialogues. Post-
colon Stud 17(2):115–121. https:// doi. org/ 10. 1080/ 13688 790.
2014. 966414
Bilic P (2016) Search algorithms, hidden labour and information
control. Big Data Soci. https:// doi. org/ 10. 1177/ 20539 51716
652159
Bilić P (2018) A critique of the political economy of algorithms:
a brief history of google’s technological rationality. TripleC
16(1):315–331. https:// doi. org/ 10. 31269/ tripl ec. v16i1. 914
Birch K (2013) The political economy of technoscience: an emerging
research agenda. Spontaneous Generations 7(1):49–61. https://
doi. org/ 10. 4245/ sponge. v7i1. 19556
Birch K (2017) Rethinking value in the bio-economy: finance, asseti-
zation, and the management of value. Sci Technol Human
Values 42(3):460–490. https:// doi. org/ 10. 1177/ 01622 43916
661633
Birch K (2019) Technoscience rent: toward a theory of rentiership for
technoscientific capitalism. Sci Technol Human Values. https://
doi. org/ 10. 1177/ 01622 43919 829567
Birch K (2020) Technoscience rent: toward a theory of rentiership for
technoscientific capitalism. Sci Technol Human Values 45(1):3–
33. https:// doi. org/ 10. 1177/ 01622 43919 829567
Birch K, Muniesa F (eds) (2020) Assetization: turning things into
assets in technoscientific capitalism. The MIT Press
Birch K, Tyfield D (2013) Theorizing the bioeconomy: biovalue, bio-
capital, bioeconomics or … what? Sci Technol Human Values
38(3):299–327. https:// doi. org/ 10. 1177/ 01622 43912 442398
Birhane A (2020) Algorithmic colonization of Africa. Scripted
17(2):389–409. https:// doi. org/ 10. 2966/ scrip. 170220. 389
Bjola C (2021) AI for development: Implications for theory and prac-
tice. Oxf Dev Stud. https:// doi. org/ 10. 1080/ 13600 818. 2021.
19609 60
Bonefeld W (2011) Primitive accumulation and capitalist accumu-
lation: notes on social constitution and expropriation. Sci Soc
75(3):379–399
Bonneuil C (2000) Development as experiment: science and state
building in late colonial and postcolonial Africa, 1930–1970.
Osiris 15:258–281
Boucher VD (2016) A few things we learned about tech hubs in Africa
and Asia. Mobile for Development. https:// www. gsma. com/ mobil
eford evelo pment/ progr amme/ ecosy stem- accel erator/ things- learn
ed- tech- hubs- africa- asia. Accessed 08 Dec 2018
Bowman W (2015) Imagining a modern rwanda: sociotechnological
imaginaries, information technology, and the postgenocide state.
Dreamscapes of modernity: sociotechnical imaginaries and the
fabrication of power. University of Chicago Press, London
Carman M, Rosman B (2021) Applying a principle of explicability
to AI research in Africa: should we do it? Ethics Inf Technol
23(2):107–117. https:// doi. org/ 10. 1007/ s10676- 020- 09534-2
Chun WHK (2009) Introduction: race and/as technology; or, how to
do things to race. Camera Obscura 24:7–35. https:// doi. org/ 10.
1215/ 02705 346- 2008- 013
Cisse M (2018a) Introducing the African Master’s of Machine Intel-
ligence at AIMS. Medium. https:// medium. com/@ moust aphac
isse/ intro ducing- the- afric an- maste rs- of- machi ne- intel ligen ce- at-
aims- 9441b 4346a 55. Accessed 06 Feb 2019
Cisse M (2018b) Look to Africa to advance artificial intelligence.
Nature 562:461. https:// doi. org/ 10. 1038/ d41586- 018- 07104-7
Coleman B (2009) Race as technology. Camera Obscura. https:// doi.
org/ 10. 1215/ 02705 346- 2008- 018
Coleman D (2019) Digital colonialism: the 21st century scramble
for africa through the extraction and control of user data and
the limitations of data protection laws. Michigan J Race Law
24(2):417–439
Collins PH (2008) Black feminist thought: knowledge, consciousness,
and the politics of empowerment, 1st edn. Routledge
Comaroff J, Comaroff JL (2012) Theory from the south: or, how euro-
america is evolving toward Africa. Anthropol Forum 22(2):113–
131. https:// doi. org/ 10. 1080/ 00664 677. 2012. 694169
Cooper B (2016) Intersectionality. The Oxford handbook of feminist
theory. Oxford University Press
Couldry N, Mejias UA (2019a) Data colonialism: rethinking Big
Data’s relation to the contemporary subject. Telev New Media
20(4):336–349. https:// doi. org/ 10. 1177/ 15274 76418 796632
Couldry N, Mejias UA (2019b) The costs of connection: how data is
colonizing human life and appropriating it for capitalism, 1st edn.
Stanford University Press
Crawford K, Gray ML, Miltner K (2014) Big Data| critiquing big data:
politics, ethics, epistemology | special section introduction. Int
J Commun 8:10
Crawford K (2016) Opinion | Artificial Intelligence’s White Guy Prob-
lem. The New York Times. https:// www. nytim es. com/ 2016/ 06/
26/ opini on/ sunday/ artifi cial- intel ligen ces- white- guy- probl em.
html. Accessed 16 Oct 2018
Crenshaw K (1989) Demarginalizing the Intersection of Race and
Sex: A Black Feminist Critique of Antidiscrimination Doctrine,
Feminist Theory and Antiracist Politics. University of Chicago
Legal Forum, 1989(6). https:// chica gounb ound. uchic ago. edu/
uclf/ vol19 89/ iss1/8. Accessed 05 July 2021
Donnor JK (2005) Racialized technology: computers, commodification,
and “cyber-race”. Counterpoints, 131, 91–102. http:// www. jstor.
org/ stable/ 42977 285
Dourish P, Mainwaring SD (2012) Ubicomp’s colonial impulse. Pro-
ceedings of the 2012 ACM Conference on Ubiquitous Comput-
ing, 133–142. https:// doi. org/ 10. 1145/ 23702 16. 23702 38
Elshakry M (2010) When science became western: historiographical
reflections. Isis 101(1):98–109. https:// doi. org/ 10. 1086/ 652691
Fanon F (2008) Black skin, white masks (R. Philcox, Trans.), 1st edn.
Grove Press
Graham M (ed) (2019) Digital economies at global margins. The MIT
Press
Giuliani D, Ajadi S (2019) 618 active tech hubs: the backbone of Afri-
ca’s tech ecosystem. Mobile for Development. https:// www. gsma.
com/ mobil eford evelo pment/ blog/ 618- active- tech- hubs- the- backb
one- of- afric as- tech- ecosy stem/. Accessed 13 Oct 2020
Gwagwa A, Kazim E, Kachidza P, Hilliard A, Siminyu K, Smith M,
Shawe-Taylor J (2021) Road map for research on responsible arti-
ficial intelligence for development (AI4D) in African countries:
the case study of agriculture. Patterns 2(12):100381. https:// doi.
org/ 10. 1016/j. patter. 2021. 100381
Hagerty A, Rubinov I (2019) Global AI ethics: a review of the social
impacts and ethical implications of artificial ntelligence. http://
arxiv. org/ abs/ 1907. 07892
Haraway D (1990) Primate visions: gender, race, and nature in the
world of modern science, 1st edn. Routledge
Harding S (ed) (2011) The postcolonial science and technology studies
reader. Duke University Press Books
Harper-Shipman TD (2019) Rethinking ownership of development in
Africa, 1st edn. Routledge
Hassan Y (2020) The politics of sharing: sociotechnical imaginaries
of digital platforms. Inform Polity 25(2):159–176. https:// doi.
org/ 10. 3233/ IP- 190203
Hilbert M (2016) Big data for development: a review of promises and
challenges. Dev Policy Rev 34(1):135–174. https:// doi. org/ 10.
1111/ dpr. 12142
Howard A (2021) Real Talk: Intersectionality and AI. MIT Sloan
Blogs; Massachusetts Institute of Technology, Cambridge,
MA.https:// www. proqu est. com/ docvi ew/ 25638 39816/ citat ion/
E7160 93124 934A5 9PQ/1. Accessed 05 June 2022
Introna LD (2016) Algorithms, governance, and governmentality:
on governing academic writing. Sci Technol Human Values
41(1):17–49. https:// doi. org/ 10. 1177/ 01622 43915 587360

1441AI & SOCIETY (2023) 38:1429–1442
1 3
Irani L, Vertesi J, Dourish P, Philip K, Grinter RE (2010) Postcolo-
nial computing: A lens on design and development. Proceed-
ings of the 28th International Conference on Human Factors
in Computing Systems-CHI ’10, 1311. https:// doi. org/ 10. 1145/
17533 26. 17535 22
Jasanoff S (ed) (2004) States of knowledge: the co-production of
science and the social order, 1st edn. Routledge
Jasanoff S, Kim S-H (eds) (2015) Dreamscapes of modernity: socio-
technical imaginaries and the fabrication of power, 1st edn.
University of Chicago Press
Jasanoff S (2016) The ethics of invention: technology and the human
future. W. W. Norton & Company
Jasanoff S, Kim S-H (2009) Containing the atom: sociotechnical
imaginaries and nuclear power in the United States and South
Korea. Minerva 47(2):119–146
Jegede O, Ncube C (2021) Science, technology, innovation manage-
ment for industrial development in South Africa: implications
for the fourth industrial revolution. Int J Innov 15(9):23
Just N, Latzer M (2016) Governance by algorithms: reality construc-
tion by algorithmic selection on the Internet. Med Cult Soc.
https:// doi. org/ 10. 1177/ 01634 43716 643157
Kanai A (2021) Intersectionality in digital feminist knowledge cul-
tures: the practices and politics of a travelling theory. Fem
Theory 22(4):518–535. https:// doi. org/ 10. 1177/ 14647 00120
975701
Kiemde SMA, Kora AD (2021) Towards an ethics of AI in Africa:
rule of education. AI Ethics. https:// doi. org/ 10. 1007/
s43681- 021- 00106-8
Kim E-S (2017) Sociotechnical Imaginaries and the globalization of
converging technology policy: technological developmentalism
in South Korea. Science as Culture. https:// doi. org/ 10. 1080/
09505 431. 2017. 13548 44
Kivimaa P, Boon W, Antikainen R (2017) Commercialising univer-
sity inventions for sustainability—a case study of (non-)inter-
mediating ‘cleantech’ at Aalto University. Sci Public Policy
44(5):631–644. https:// doi. org/ 10. 1093/ scipol/ scw090
Kopf D (2018) Giving poor people cash is a good idea. Giving entre-
preneurs cash might be a great one. Quartz Africa. https:// qz.
com/ africa/ 13278 33/ kenya- is- follo wing- niger ia- in- runni ng-a-
massi ve- busin ess- plan- compe tition/. Accessed 16 Oct 2018
Krige J, Wang J (2015) Nation, Knowledge, and imagined futures:
science, technology, and nation-building, post-1945. Hist
Technol 31(3):171–179. https:// doi. org/ 10. 1080/ 07341 512.
2015. 11260 22
Kwet M (2018) Digital colonialism: US empire and the new imperial-
ism in the global south (SSRN Scholarly Paper ID 3232297). Soc
Sci Res Netw. https:// doi. org/ 10. 2139/ ssrn. 32322 97
Leonelli S (2016) Data-centric biology: a philosophical study. Univer-
sity of Chicago Press
Lijadu K (2018) How a master’s program in machine intelligence is
trying to close an African tech gap. Quartz Africa. https:// qz.
com/ africa/ 13445 52/ google- faceb ook- back- afr ic an- machi ne-
intel ligen ce- progr am/
MacKenzie D (2017) A material political economy: automated trading
desk and price prediction in high-frequency trading. Soc Stud
Sci 47(2):172–194. https:// doi. org/ 10. 1177/ 03063 12716 676900
Madden P (2020) Figures of the week: Africa’s preparedness for the
Fourth Industrial Revolution. The Brookings Institution. https://
www. brook ings. edu/ blog/ africa- in- focus/ 2020/ 03/ 11/ figur es- of-
the- week- afric as- prepa redne ss- for- the- fourth- indus trial- revol
ution/
Madianou M (2019) Technocolonialism: digital innovation and data
practices in the humanitarian response to refugee crises. Soc Med
Soc. https:// doi. org/ 10. 1177/ 20563 05119 863146
Mann S, Hilbert M (2020) AI4D: artificial intelligence for develop-
ment. Int J Commun 14:21
Mbembe A (2017) Africa in theory. African futures: essays on cri-
sis, emergence, and possibility, 1st edn. University of Chicago
Press, pp 211–230
Mbembe A (2017b) Critique of black reason (L. DuBois,Trans.).
Duke University Press
McGrath S, Ramsarup P, Zeelen J, Wedekind V, Allais S, Lotz-
Sisitka H, Monk D, Openjuru G, Russon J-A (2020) Vocational
education and training for African development: a literature
review. J Vocat Educ Train 72(4):465–487. https:// doi. org/ 10.
1080/ 13636 820. 2019. 16799 69
Miller TR (2020) Imaginaries of sustainability: the techno-politics of
smart cities. Sci Cult 29(3):365–387. https:// doi. org/ 10. 1080/
09505 431. 2019. 17052 73
Mohamed S, Png M-T, Isaac W (2020) Decolonial AI: decolonial
theory as sociotechnical foresight in artificial intelligence.
Philos Technol 33(4):659–684. https:// doi. org/ 10. 1007/
s13347- 020- 00405-8
Moyo A (2019) How Google is using AI to solve African problems.
ITWeb Africa. http:// www. itweb africa. com/ unifi ed- commu
nicat ions/ 656- ghana/ 245764- how- google- is- using- ai- to- solve-
afric an- probl ems. Accessed 20 April 2019
Mumbere D (2018) Sophia the robot misses dinner with Ethiopia PM
after losing some parts at German airport. Africanews. http://
www. afric anew s. com/ 2018/ 06/ 30/ sophia- the- robo t- misses- din-
ner- with- ethio pia- pm- after- losing- some- parts- at/. Accessed 15
Oct 2018
Nandutu I, Atemkeng M, Okouma P (2021) Integrating AI ethics in
wildlife conservation AI systems in South Africa: a review,
challenges, and future research agenda. AI Soc. https:// doi. org/
10. 1007/ s00146- 021- 01285-y
Ndung’u N, Signé L (2020) The Fourth Industrial Revolution and
digitization will transform Africa into a global powerhouse
(Foresight Africa 2020). The Brookings Institution. https://
www. brook ings. edu/ resea rch/ the- fourth- indus trial- revol ution-
and- digit izati on- will- trans form- africa- into-a- global- power
house/. Accessed 09 Feb 2022
Noble SU (2016) A Future for intersectional black feminist technol-
ogy studies. S&F Online, 14.1(13.3). http:// sfonl ine. barna rd.
edu/ trave rsing- techn ologi es/ safiya- umoja- noble-a- future- for-
inter secti onal- black- femin ist- techn ology- studi es/. Accessed
30 Oct 2018
Noble SU (2018) Algorithms of oppression: how search engines rein-
force racism. NYU Press
Nyagadza B, Pashapa R, Chare A, Mazuruse G, Hove PK (2022) Digi-
tal technologies, fourth industrial revolution (4IR) & global value
chains (GVCs) nexus with emerging economies’ future indus-
trial innovation dynamics. Cogent Econ Finance 10(1):2014654.
https:// doi. org/ 10. 1080/ 23322 039. 2021. 20146 54
Ogbonnaya-Ogburu IF, Smith ADR, To A, Toyama K (2020. Critical
Race Theory for HCI. Proceedings of the 2020 CHI Conference
on Human Factors in Computing Systems, 1–16. https:// doi. org/
10. 1145/ 33138 31. 33763 92
Ojo T (2018) Political economy of ICT4D and Africa. In: Servaes J
(ed) Handbook of communication for development and social
change. Springer, pp 1–14. https:// doi. org/ 10. 1007/ 978- 981- 10-
7035-8_ 64-1
Oksanen K, Hautamäki A (2014) Transforming regions into innova-
tion ecosystems: a model for renewing local industrial structures.
Innov J 19(2):16
Overton J (2019) Aid, ownership and development: the inverse sover-
eignty effect in the Pacific Islands. Routledge, London
Parayil G (2005). In: Parayil G (ed) Political economy and informa-
tion capitalism in india: digital divide, development and equity,
2005th edn. Palgrave Macmillan
Peña P, Varon J (2019) Decolonising AI: A transfeminist approach to
data and social justice. GLOBAL INFORMATION SOCIETY

1442 AI & SOCIETY (2023) 38:1429–1442
1 3
WATCH. https:// giswa tch. org/ sites/ defau lt/ files/ gisw2 019_ web_
th4. pdf
Peters MA (2017) Technological unemployment: educating for the
fourth industrial revolution. Educ Philos Theory 49(1):1–6.
https:// doi. org/ 10. 1080/ 00131 857. 2016. 11774 12
Pfotenhauer S, Jasanoff S (2017) Traveling Imaginaries: the “practice
turn” in innovation policy and the global circulation of innova-
tion models. In: Tyfield D, Lave R, Randalls S, Thorpe C (eds)
The Routledge handbook of the political economy of science, 1st
edn. Routledge, pp 416–428
Phan T, Wark S (2021) Racial formations as data formations. Big Data
Soc 8(2):20539517211046376. https:// doi. org/ 10. 1177/ 20539
51721 10463 77
Phan T, Goldenfein J, Mann M, Kuch D (2021) Economies of virtue:
the circulation of ‘ethics’ in big tech. Sci Culture. https:// doi. org/
10. 1080/ 09505 431. 2021. 19908 75
Philip K, Irani L, Dourish P (2012) Postcolonial computing: a tactical
survey. Sci Technol Human Values 37(1):3–29. https:// doi. org/
10. 1177/ 01622 43910 389594
Powell M (2001) Knowledge, culture & the internet in africa: a
challenge for political economists. Rev Afr Political Econ
28(88):241–260
Quijano A (2000) Coloniality of power and eurocentrism in latin Amer-
ica—Aníbal Quijano, 2000. Int Sociol. https:// doi. org/ 10. 1177/
02685 80900 01500 2005
Roberts D (2011) Fatal invention: how science, politics, and big busi-
ness re-create race in the twenty-first century. The New Press
Roberts D (2013) Law, race, and biotechnology: toward a biopolitical
and transdisciplinary paradigm. Ann Rev Law Soc Sci 9(1):149–
166. https:// doi. org/ 10. 1146/ annur ev- lawso csci- 102612- 134009
Saurwein F, Just N, Latzer M (2015) Governance of algorithms:
options and limitations. Info 17(6):35–49. https:// doi. org/ 10.
1108/ info- 05- 2015- 0025
Schwab K (2017) The fourth industrial revolution. Penguin Group
Sermondadaz S (2018) Neil Turok: “Nous souhaitons créer le premier
master Big data et IA en Afrique.” Sciences et Avenir. https://
www. scien ceset avenir. fr/ fonda mental/ neil- turok- nous- souha
itons- creer- le- premi er- master- big- data- et- ia- en- afriq ue_ 122595
Siminyu K, Freshia S, Abbott J, Marivate V (2020) AI4D -- African
Language Dataset Challenge. [Cs]. http:// arxiv. org/ abs/ 2007.
11865
Smith LT (2012) Decolonizing methodologies: research and indigenous
peoples, 2nd edn. Zed Books
Smith G (2017) Artificial Intelligence and Big Data to fast-track the
fight against malnutrition in Africa|CIAT Blog. https:// blog. ciat.
cgiar. org/ artifi cial- intel ligen ce- and- big- data- to- fast- track- the-
fight- again st- malnu triti on- in- africa/
Srnicek N (2016) Platform capitalism. Polity Press.
Subramaniam B, Foster L, Harding S, Roy D, TallBear K (2016) Femi-
nism, postcolonialism, and technoscience. In: Felt U, Fouche R,
Miller CA, Smith-Doerr L (eds) The handbook of science and
technology studies, 4th edn. MIT Press, pp 407–433
Truby J (2020) Governing artificial intelligence to benefit the UN sus-
tainable development goals. Sustain Dev 28(4):946–959. https://
doi. org/ 10. 1002/ sd. 2048
Tuck E, Yang KW (2012) Decolonization is not a metaphor. Decolo-
nization 1(1):1–40
Tyfield D, Lave R, Randalls S, Thorpe C (2017) The Routledge hand-
book of the political economy of science. Routledge
Unwin PTH, Unwin T (2009) ICT4D: information and communication
technology for development. Cambridge University Press
Vicente PN, Dias-Trindade S (2021) Reframing sociotechnial imaginar-
ies: the case of the Fourth Industrial Revolution. Public Underst
Sci 30(6):708–723. https:// doi. org/ 10. 1177/ 09636 62521 10135 13
Wairegi A, Omino M, Rutenberg I (2021) AI in Africa: framing AI
through an African Lens. Commun Technol Dév. https:// doi. org/
10. 4000/ ctd. 4775
Wall PJ, Saxena D, Brown S (2021) Artificial Intelligence in the Global
South (AI4D): Potential and Risks. [Cs]. http:// arxiv. org/ abs/
2108. 10093
Zarsky T (2016) The trouble with algorithmic decisions: an analytic
road map to examine efficiency and fairness in automated and
opaque decision making. Sci Technol Human Values 41(1):118–
132. https:// doi. org/ 10. 1177/ 01622 43915 605575
Ziewitz M (2016) Governing algorithms: myth, mess, and methods.
Sci Technol Human Values 41(1):3–16. https:// doi. org/ 10. 1177/
01622 43915 608948
Zuboff S (2019) The age of surveillance capitalism: the fight for a
human future at the new frontier of power, 1st edn. PublicAffairs
Publisher's Note Springer Nature remains neutral with regard to
jurisdictional claims in published maps and institutional affiliations.