Policy Significance Statement

This article highlights the potential clash between artificial intelligence (AI) integration and essential African ethical values, specifically Afro-communitarianism and human dignity. It underscores how corporate adoption of AI could jeopardize these core values ingrained in sub-Saharan ethics. By examining this intersection, it provides a pioneering perspective within AI ethics, a crucial consideration in policymaking. This unique exploration emphasizes the urgency of preserving these values amidst AI proliferation, offering a roadmap to prioritize these foundational ethics in the development of an African AI ethics framework. Policymakers on the African continent must heed this call to ensure the responsible adoption of AI that respects and aligns with ethical context, shaping inclusive and culturally sensitive AI policies.

2. Values at Risk

Two assumptions are required to pursue a discourse such as this; first, that ethical values could be at risk from AI proliferation, at least as used today, and second that they are worth defending. In this section, I discuss two values prone to the threat by the adoption, development, and deployment of AI given current business motivation. I also show how they may differ from predominantly Western ethics like classical utilitarianism and Kantian deontology. To do this, I first unpack these values, showing how integral they are to the notion of ethics in Africa. Furthermore, I show why these values are worth defending by demonstrating their criticality to sub-Saharan African ethical systems.

In analyzing African ethical values at risk, below are two subsections in which I identify the ethical values that are at risk – communitarianism and dignity (rooted in personhood). Each of the subsections is dedicated to unpacking one of these ethical values and then discussing select ethical issues as they apply to AI. I take this approach to better buttress my claims that AI systems, as developed today, stand as a threat to these selected African ethical values. I cite examples of AI use such as recommender systems, facial recognition technology, and AI in criminal justice – law enforcement and jury. I explore how the use of this technology could substantially affect the values of humanness and Afro-communitarianism.

3. An Ode to Afro-communitarianism

Afro-communitarianism is one of the most discussed ethical principles of characteristically sub-Saharan ethics, emphasizing community and communal relations. This is because relationships play a pivotal role in all conceptions of African ethics. According to some construal of African ethics, particularly those of Southern African extract, communal relations are considered the highest good and one worth pursuing (Mokgoro, Reference Mokgoro1998; Ramose, Reference Ramose1999; Shutte, Reference Shutte2001; Cornell and van Marle, Reference Cornell and Van Marle2005; Metz, Reference Metz2007). Having a good relationship with the community means that one is in connection with the very fiber that sustains the community. Communal relations are the building blocks of society, and Afro-communal ethic suggests that we prioritize them.

It could be argued that relationality as an ethical principle is not unique to Africa as other cultures demonstrate a sense of communitarianism. While true in a technical sense, relationality is salient in sub-Saharan ethical systems, and hence can be broadly described as African in much the same way gridiron football or American football signifies a distinct style that can be called American despite being played internationally. African relational ethics captures a core value central to African people. This claim has been reiterated by several scholars such as Nyerere, (Reference Nyerere1987), Mbiti (Reference Mbiti1990), Ramose (Reference Ramose1999), and Oruka (Reference Oruka1990). These scholars, though diverse, point to the primacy of relationality in African ethics.

There are at least two prominent ways to conceive of a relational ethic from an Afro-ethical standpoint. I will focus on my most favored version for the purpose of this paper; however, I will briefly discuss the other to show an awareness of it. The first, which is not my favored version, states that the pursuit of communal relations is a means to actualizing oneself (this position is shared by Menkiti (Reference Menkiti and Wright1984), (Reference Menkiti and Wiredu2004), Mokogoro (Reference Mokgoro1998), Shutte (Reference Shutte2001), and Molefe (Reference Molefe2019)). Under this version, the goal of being in a relationship with others is self-actualization; we become by being in a community with others. Ramose captured this more profoundly when he opined that “to be a human being is to affirm one’s humanity by recognising the humanity of others and, on that basis, establish humane relations with them…. One is enjoined, yes, commanded as it were, to actually become a human being” (Reference Ramose1999: 52). To put this version of relational morality into context, a right action becomes one that promotes communal relations insofar as it helps to actualize oneself; a wrong action is one that fails to actualize oneself.

In a relational version of African ethics or Afro-ethics, one becomes a person only as far as they can actualize certain communal values, of which relationality is most important. Indeed, Menkiti (Reference Menkiti and Wiredu2004) suggests that to attain personhood, some sort of group solidarity or group relationship is important, for the claim that one is a person through other persons or “I am because we are…” is simply the individual’s recognition of the fact that the source of their humanity is based on their engagement with others. And so, if no other exists as a source/ground on which this personhood is cultivated, then there would be no grounds for the conferment of personhood on the individual (Reference Menkiti and Wiredu2004; 324). Echoing the same sentiments, Augustine Shutte (Reference Shutte2001: Reference Metz and Gaie30) relays to us that “… the moral life is seen as a process of personal growth [because] our deepest moral obligation is to become more fully human. And this means entering more and more deeply into community with others”. Entering communion or community with others would entail that we share relationships, closeness, vulnerability, and intimacy with others, share in the joys and pains of others, look out for others, and be of assistance to others.

Even though traces of self-realization as a goal of morality can be seen in other ethical systems, like Kantian and Utilitarian ethics, this version of Afro-relational ethics has clear differences between its construal of self-realization and those of Kantian and utilitarian ethics. For Kant, self-realization is tied to autonomy (Kant, Reference Kant2008), the ability to make independent moral decisions as an agent. Hence, a Kantian would expect a moral agent to be able to make moral decisions free from desires or compulsion to do so (Louden, Reference Louden2002). By relying on the centrality of rational thinking, Kantian ethics would expect the agent to have the capacity to deliberate and heed oneself to the moral law rather than just heeding externally imposed injunctions (Paton, Reference Paton1948). The utilitarian, on the other hand, would understand self-realization as a process of actualizing one’s hedonic desires (Driver, Reference Driver and Zalta2014), preferences, or at least minimizing pain (Rawls, Reference Rawls and Bayles1968). When contrasted with these two – Kantian and utilitarianism – self-actualization in African ethics focuses less on the individual’s independence and more on the individual’s journey to becoming one with the community.

The second interpretation of African relational ethics, which is my more favored version and one that I find uniquely suited for this paper, dictates that advancing communal relations is morally good in itself since it shows respect for others based on their capacity to be human. I find this approach desirable largely because extensive work has been done to refine and apply it to several important fields such as bioethics, law, and political philosophy. More so, the emphasis on relationality as an intrinsic value allows us to de-emphasize the role of relationality as merely a means to some other end-goal – personhood/self-realization. The major proponent of this version of African ethics is Thaddeus Metz. Metz (Reference Metz2012) in his work on “African Conceptions of Human Dignity: Vitality and Community as the Ground of Human Rights” insists that there is a very appealing interpretation of African ethics, which takes a relational framework. He notes:

This interpretation in many ways is deontological in form. Our capacity to relate communally gives us dignity, and so we should pursue these relations as ends not merely because they help us actualize ourselves or maintain some sort of peace among members of society, even though these things are seen as appurtenances. This construal offers us the platform to ground individual rights, a vital matter in the ethics of AI. A closer look at this interpretation suggests to us two distinct elements; one, “identifying with others” and two, “exhibiting solidarity with others” (Metz and Gaie, Reference Metz and Gaie2010: 276; Metz, Reference Metz and Van Hooft2013: 278–279).

To identify with others involves seeing oneself as well as being accepted as a part of a relationship, bound by a sharing of values, worldviews, way of life, and feeling assimilated to others. By identifying with others, we take up a cooperative disposition, putting the well-being of others and group goals at the center stage, thereby ensuring that our dominant psychological posture is represented by pronouns such as ‘we’ and ‘us’, as opposed to ‘I’. To show solidarity, then, involves being sympathetic to others, prioritizing the common good, which is the good of members of a community/group, showing the willingness to serve others, demonstrating sympathetic altruism, serving others by celebrating with them when they celebrate and mourning with them when they mourn. By demonstrating solidarity with others, we seek to protect their interests.

According to Metz in his article “An African theory of social justice: Relationship as the ground of rights, resources and recognition,” “…one way to be party to a communal relationship would be to exhibit identity and solidarity with others, as a subject. However, one could also be a party to such a relationship by being identified with or exhibited solidarity towards, as an object” (Reference Metz, Camilla and Matthew2016: 179). Metz’s notion that “one could be party to a relationship by being identified with or exhibited solidarity towards” could grant an Afro-ethical justification to widen the moral circle to accommodate AI systems. In other words, given this patient-centered dimension of the relational view, the mere existence of an entity that is deserving of receiving some form of solidarity incorporates such an entity into the community. In this way, animals (pets, totems), trees, AI systems, and the like become part of the community for which relational behaviors are relevant. Using this (patient-centered) relational model, we can accommodate AI systems that we exhibit solidarity towards and identify with such as care robots, robot nannies, and possibly sex robots into our moral circle; but some types of AIS, especially given certain contexts, are unlikely to be considered objects we share identity and solidarity with such as autonomous weapons systems or a self-driving car. Notice that, unlike the Kantian view, emphasis is not placed on some intrinsic rational capacity but on the capacity of the object to be part of a relationship, whether as a subject-object or merely as an object. And so, questions about consciousness, rationality, and the like would not apply here. Similarly, and regarding the utilitarian view, the litmus test would not be whether AI systems are the type of objects/subjects that possesses the requisite amount of sentience to feel pleasure or pain or seek to maximize satisfaction but on their capacity to be subjects of a relationship.

4. How AI Undermines Afro-communitarianism

Before discussing how AI undermines Afro-communitarian values, it is important to offer a working definition of AI. AI encompasses a range of concepts, applications, and technologies that refer to the development of computer systems or machines to perform tasks that mimic human intelligence. These tasks range from learning, problem-solving, perception, natural language understanding, reasoning, and the like.

When I discuss AI broadly in this paper, I am focused on the sort of description Dan McQuillan offers in his book Resisting AI (Reference McQuillan2022) where the focus is on computational systems that not only produce effects through predictions but also shape how we perceive the world and interact with it. McQuillan argues that AI is more than just a set of machine learning methods; when we speak of AI “we can’t separate the calculations in the code from the social context of its application. AI is never separate from the assembly of institutional arrangements that need to be in place for it to make an impact in society” (Reference McQuillan2022: 1). Additionally, AI, like these institutions, is not value-neutral as it still embodies assumptions about how the world works.

Given what we now know about the African relational approach to ethics, as discussed in the previous section, I must now delineate what I take to be the shortfall of AI in terms of how it undermines communal-relational values. I begin this discussion with the claim that AI developed (for the most part) in the global north (un)intentionally globalizes the value systems of those societies and by implication imposes them on (individuals from) developing societies like Africa, who are consumers of these products (Segun, Reference Segun2021). More so, these AI systems still embody the implicit and explicit assumptions that grounds modern capitalism.

To unpack the above claim, I begin by tracing the historical context that foregrounds it – colonialism in Africa. Quite evidently, the approach of technological corporations to gain and entrench a presence on the continent follows a similar pattern as colonialism, a phenomenon Michael Kwet (Reference Kwet2019) and Abeba Birhane (Reference Birhane2020) refer to as “digital colonialism” and “algorithmic colonization” respectively. Birhane contends that while colonialism used brute force to drive economic exploitation in Africa, technological colonization uses sophisticated algorithms and AI to do so. Digital colonization is carried out through the extraction of data from Africa without informed consent. As data is not without context, using them to train AI models designed to solve problems in a different context could increase the likelihood of having algorithmic bias.

Just like historical colonialism, technology corporations – often big tech companies – pose as a panacea to Africa’s socioeconomic problems and as viaducts connecting the continent to technological infrastructure common in the global north. However, this is sometimes built on the backs of exploitation following the lack of robust regulatory requirements on the continent relative to the European Union. For example, through companies like Crowdflower and platforms like Amazon’s Mechanical Turk acting as intermediaries for managing cheap labor, many of the workers or contributors to the development of AI systems reside in the majority world; “it is these forms of globally distributed labor that make it economically viable to produce the required volumes of labelled data, whether that’s tagging images from social media or transcribing voice recordings from systems like Siri and Alexa” (McQuillan, Reference McQuillan2022: 24). By exploiting cheap labor, big tech companies can pay poor wages to workers at a click-farm in the majority world to help identify, annotate, and analyze data to teach self-driving cars (Lee, Reference Lee2018).

Exploitation seems to be a fundamental moral failing for many big tech companies. For instance, in Kenya, Facebook was identified as the platform where internet-based sexual exploitation of minors was most prevalent compared to other sites (Njanja, Reference Njanja2021). Beyond overt human rights exploitations like the one mentioned above, there are other subtle ones that are nonetheless still exploitative. Undoubtedly, all forms of exploitation conducted by technology companies leveraging AI are incongruous with the concept of Afro-communitarianism. Fundamentally, the idea of being one with the community and communing with others is not designed for exploitation but for mutual prosperity. On the contrary, most big tech companies operate with unrestrained capitalist ideals, which are built on exploitation of the workforce and profit maximization, a phenomenon that mirrors the motivation that grounds colonialism. In addition, under this capitalist cloak, the individual is seen as “a means to an end” and not the “end in itself,” which is the goal of African relational ethics.

The exploitative nature of big technology companies when carrying out critical AI development tasks like data labeling and annotation as shown above, often depended on “extractive labour practices” (McQuillan, Reference McQuillan2022: 25) motivated by Western individualism, which sits in opposition to Afro-communitarianism. As Sabelo Mhlambi explains, Western individualism:

Thus, while one might take for granted that much of our technologies, including AI, are developed in Euro-American tech hubs, and embody value systems/worldview, of that society the impact these technologies have on African ethical values could be profound. This is largely because if these technologies and the substructures upon which they are built are not properly interrogated, they will undermine such values as Afro-communitarianism, as they present a clash of values. As McQuillan notes, AI systems not only produce effects, but they also “shape how we perceive the world and interact with it” (Reference McQuillan2022: 1). For example, recommendation systems, as an AI tool used by social media platforms, continue to shape how we see the world by pushing and consequently globalizing western ideals, making ideologues of many young Africans, and creating tension as they juggle two different value systems. A Pew Research report showed that Africans younger than 30, as with smartphone ownership, were more likely than those aged 50 or older to go online (Silver and Johnson, Reference Silver and Johnson2018); this means continuous interaction with contents that highlight non-African values. With the proliferation of polarized Euro-American socio-political views on social media platforms like Twitter, Facebook, and TikTok across two extremes (conservative vs liberals), there has been a growing push and idolization of individualism and isolation. Perhaps the more ominous implication is that many of the users of social media in Africa are still impressionable and are exposed to incompatible ideological values that conflate with their lived experience.

AI tools carry with them implicit or even explicit values that play finely into the structural issues Mhlambi (Reference Mhlambi2020) alluded to such as hierarchical, discriminatory, and exclusionary values that are offshoot of Western individualism. In our example above, by globalizing the algorithm for recommender systems, social media platforms create room for ethical imperialism, where Western standards are often imposed on AI development globally, disregarding the unique cultural context of African communities. Consequently, this leads to AI systems that prioritize individual gain over collective good, weakening the strong sense of solidarity that is foundational to Afro-communitarianism.

There are also the phenomena of what I call negative added values and absent values. Artificial intelligent systems have grown in sophistication and adoption in the last 5 years – from basic computing of early computers to very advanced and complex large language models like the Beijing Academy of Artificial Intelligence WuDao 2.0, Open AI’s ChatGTP4, Google’s Gemini, and the like. Many aspects of human life – from healthcare, banking, education, policing, governance, business, to retail, etc. – are increasingly influenced by AI. While the growth and application of this technology in everyday life is quick, it seems clear that not a lot of time is afforded to developers to think through some of the wider implications of these systems as well as outlying problems that are of great importance to certain groups of people. Given these gaps, problems are bound to occur. Negative added values (NAVs) are an example of such problems. The focus on the business case or problem that an AIS is supposed to solve often creates the type of narrow-mindedness that ignores important issues like ethical and human rights considerations. For instance, social media platforms using AI, make content recommendations that centers the individual until they become addicted and detached from the real world. As recent research from the Huntsman Mental Health Institute, University of Utah shows, “young adults who use social media are three times as likely to suffer from depression, putting a large portion of the population at risk for suicidal thoughts and behaviors” (2023). This growing self-isolation and hyper-focus on the individual puts at risk values such as friendship, solidarity, common good, and fellowship with others, which are central to Afro-communitarianism. To show solidarity would involve being sympathetic to others, prioritizing the common good, which is the good of members of a community/group, showing the willingness to serving others, demonstrating sympathetic altruism, serving others by celebrating with them when they celebrate and mourning with them when they mourn.

Another example that buttresses the argument on NAVs is the use of AI in the human resource management and employee recruiting field. With the aid of AI-enabled tools known as applicant tracking systems (ATS), a vast number of resumes and applications are sorted using specific keywords or parameters that match the job description. While such systems are effective in sieving through thousands of applications to select the best group of applicants to proceed to other rounds of interviews, research has shown that the often selected groups do not always reflect the best of the bunch and can discriminate against qualified candidates who do not have certain prompts on their resume like an Ivy League education or experience with a popular previous employer (Hunkenschroer and Luetge, Reference Hunkenschroer and Luetge2022). One implication of using these systems is that individuals, who are qualified but come from underrepresented/disadvantaged groups or with ethnic-sounding names from the majority world, are often ignored since the AI tool employed often represents the bias from the training data regarding what constitutes a good candidate, further entrenching structural inequalities in society. This same argument can be made of the use of AI to predict criminal recidivism, where research by ProPublica indicated that individuals of African descent were (wrongly) profiled by these tools to be more likely to re-offend, as opposed to certain other groups (Larson et al., Reference Larson, Mattu, Kirchner and Angwin2016)Footnote ². These examples reflect negative added values in AI and run in contradistinction to the values of Afro-communitarianism where the sum of an individual within the relational value system is not reduced to numeric data predicted by an AI system based on proximate inputs, as it hinders individual and community economic growth, a vital aspect of Afro-communitarian well-being.

Absent values (AVs), on the other hand, reflect a situation where AI systems are developed in a way that fails to meet the socio-cultural realities or context-specificity of the region in which they are deployed. In addition, the algorithms that undergird some of these systems might show biases that undermine some cultural or context-specific issues. For instance, an AI system in healthcare developed for Western hospitals, with the goal of addressing unique challenges might not prioritize the needs of rural African communities. This could significantly undermine the ability of the community to collectively maintain health and well-being, a core element of Afro-communitarianism. Another good example of this is algorithms used for content moderation on social networking platforms. These algorithms often are encoded with a context in mind that is absent from areas they may be deployed. For instance, in October 2020, there was a series of protests against the Nigerian police following accusations of human rights violations. Videos and photo evidence emerged showing these abuses and were shared on Facebook. These images were flagged as false or disinformation and automatically deleted, inadvertently silencing the voice of protesters (Ilori, Reference Ilori2020; Edward-Ekup, Reference Edward-Ekpu2020; Kolawole, Reference Kolawole2020). These algorithms were created in Facebook’s bid to address the problem of disinformation that ensued following the 2016 US presidential election. Given the lack of appropriate context, an algorithm designed to address a specific issue that applied to the US and other countries with similar polity was transposed and hampered civil rights campaigns in Nigeria. This is perhaps one of the biggest challenges with globalizing values through AI.

Furthermore, while negative added values rile against what is acceptable within the context of the relational view, absent values fail to incorporate aspects of the relational view that remain relevant within the African context. Both circumstances reflect a lack of intentionality, in terms of incorporating context-specific values in the development of AI. The takeaway from the above arguments is that many AI systems or tools are built on the back of a sociocultural worldview that tends to be universalized with little care for cultural or ethical contexts outside the predominant worldview. Within the context of Africa, this creates the sort of enabling environment that allows for some non-context-specific values to be uncritically accepted, inevitably causing tensions and entrenching inequalities.

5. Much Ado About Dignity

In this Section I, discuss the concept of dignity within the context of African philosophical thought. Dignity is an important subject when conversations around the ethics of AI come up, and it is a central theme in African normative ethics. Collectivist cultures, like those common to Africa, are known to prioritize interdependence over independence and solidarity over autonomy. Since African ethics is relational in nature, there is a tendency to assume that the bias against autonomy somehow makes it difficult to ground human rights. This is far from true, as I will show how a relational model grounds individual rights. Some commendable work has been done in this area, but I hope to show how the concepts of dignity and least harm play a key role in how we must view the ethics of AI.

Discourses around human rights are ever so often grounded on the concept of dignity. Absent of this, many ethicists believe there is no real basis to talk about human rights and the need to protect them. Before rights are conceived of as legal concepts, they are first rooted in moral intuitions. Often, these intuitions are considered universal injunctions. For example, murder is considered a criminal offense because it robs the victim of the right to life. The right to life is preserved under constitutions that consider upholding the claim that there is some intrinsic worth in a human life, i.e., dignity. The predominant view of dignity rests on two notions that are grounded in Western ethics, particularly Kantianism. These are the concepts of autonomy and rationality.

The concept of dignity connotes a prized and non-tangential value that is worthy of respect and often attributed to humans. To speak of dignity is to speak of the worth possessed by a person. This intrinsic worth merits respect in virtue of the bearer’s capacity. For the Kantian, dignity is dependent on the capacity to be rational or exercise autonomy. For the Afro-ethicist, dignity is dependent on the capacity to commune with others and to exhibit some form of vitality (Metz, Reference Metz2012: 20). To be able to commune with others, in this respect, one must possess vital force, which is summarily exhibiting personhood or having the potentiality to become a person in much the same way Menkiti (Reference Menkiti and Wiredu2004) describes it. In simple terms, the concept of vitality takes a psychological look at persons, capturing the mental dimension of human life. The vitalist perspective suggests to us that if a person is depressed and unhealthy, for instance, what is bad about their lives is that they are not thriving and are unable to function (Metz, Reference Metz2012: 25). Alternatively, if a person shows signs of liveliness, happiness, and excitement, what is good about their lives is that they are strong, thriving and can function within a community. The vitalist perspective gives a strong account of what makes a life a good life or a bad life. The good life is one that is complex, sophisticated, growing, strong, and creative while the bad life is one that is simple, static, weak, or repetitive (Metz, Reference Metz2012: 28).

A deeper look at the vitality theory suggests that vital force is a necessary property of personhood and by extension for communality. Vitality is the quality that distinguishes a human being from other life forms and is the substructure upon which humans have the capacity to commune. For instance, at birth, humans possess an intrinsic property that merits respect, and this makes us place sanctity on life. Another argument for vital force as a grounding for dignity in humans as opposed to other life forms is that humans possess a unique quantity and quality of vital force than those found in other physical things or life forms in the world. The implication is that vitality in humans makes them special (Agada, Reference Agada2020). For this reason, dignity is not merely a respect for the rational nature of an agent but the respect for the capacity to commune and be communed with. Unlike the Kantian conception of dignity which solely internal, the African conception of dignity can be validated both internally and externally. Internally, it can be validated by an appeal to vital force and externally by demonstrating a capacity to be in communion with others by either identity or solidarity.

From the Afro-ethical standpoint, a person is said to have dignity if they are capable of being part of a community. By being part of a community, I mean being the subject or object of a communal relationship (see Metz, Reference Metz2015). Considering that African normative principles are community-centered, every individual is believed to have, by nature, the capacity to commune or be communed with, hence, meriting respect. The vitality human beings possess pushes us to relate with others, in doing so we become more human. Hence, you cannot successfully commune with others if you lack vital force.

Let us look at some marginal or edge cases. In many ways, a psychopath, within this context, will not be considered a person (if we follow Menkiti’s notion of personhood) and would lack the capacity to commune with others. The inability of persons with psychopathy to show remorse and empathy would mean that they are not living a good life because they do not exhibit the capacity necessary to commune with others. On the other hand, a baby, though limited in its capacity to commune, is said to have the potential to do so.

In looking at the concept of vital force and the capacity to commune, it is easy to notice that there are things with the capacity to commune but lacking in vitality to the same degree we speak of it in humans. For example, our dogs and cats have the capacity to commune, and for many African moral theorists, possess vital force, however, not to the degree of humans who are imbued with consciousness, creative power, and rationality. That essential property that distinguishes humans from these life forms is what makes us speak of inalienable rights.

Dignity can also be grounded in the notion of personhood. Recall in the first Section I showed that self-actualization in the sense of becoming more human was one of the conceptions of relational ethics; in the same vein, the pursuit of moral perfection, “where the chief moral goal of the agent is to perfect her own humanity,” as argued by Motsamai Molefe (Reference Molefe2020), is considered a cardinal part of personhood. Against this backdrop, we establish that a person is not considered human if they fail to affirm the humanity of others. This process is what gives birth to an egalitarian notion of dignity in African moral thought.

For most African ethicists, promoting dignity as a value indirectly prizes the principle of least harm, which is a respect for the communal nature of a person and the deliberate avoidance of degrading such a person. Hence, we are enjoined to avoid denigrating other humans because doing so is likely to affect their capacity to commune with others. An action that, no matter their net consequences, degrades the dignity of others affects their capacity to commune.

For most Kantians, a violation of dignity is seen as a degradation of either a person’s autonomy or rational nature; for African ethics, a violation of dignity is characterized by a degradation of relational nature, what Metz in his work, “Human Dignity, Capital Punishment, and an African Moral Theory” (Reference Metz2010), argues is our capacity to commune or be in friendly relationships. Under this principle, actions that violate the rights and dignity of others are those that treat others in a discordant, unfriendly, and violent way. Unlike Kant’s conception of dignity which attributes dignity based on rationality, the African ethical perspective contends that it is possible to lack the mental capacity to be rational and still be considered one with dignity because they can be a party to a relationship – as an object of that relationship.

As shown above, the principle of dignity is tied to that of least harm. Should an action have the potential to cause harm to others, we must ensure that it does the least harm and not degrade others. Aptly put, agents carrying out a decision must minimize the extent of degradation of others and their ability to commune with others.

It is easy to think the principle of dignity and least harm is conflated with negative utilitarianism. This is not the case for two reasons. One, utilitarianism does not make a defense for dignity; in fact, the concept is alien to utilitarians. Two, the African ethical conception of dignity and least harm appeals to a very different idea. For the utilitarians, the appeal is to some sort of hedonism on the grounds of the individual having the capacity to feel pain or the desire to maximize utility. From the African ethical viewpoint, the appeal is to the capacity of the individual to be in communion with others. Harm, here, refers to any action able to degrade the capacity to commune with others or cause injury. I reckon that the Afro-ethical perspective considers the dignity of the individual as a primary pillar in the notion of least harm. With this understanding of dignity, the problems associated with the non-contextualization of AI become apparent. I will discuss two examples that buttress this point – the use of facial recognition technology and targeted advertising using private data.

6. How AI Undermines the Notion of Dignity

Perhaps you are wondering why dignity is an important subject to discuss when we talk about AI. The answer is simple; a respect for human dignity would imply developing just, fair, and human-centred AI. It is self-evident that digital colonialism which has now been enabled by AI and manifests itself in data surveillance and mining will in no way prioritize or preserve the dignity of all; it is instead, likely to create more inequality.

As McQuillan (Reference McQuillan2022) notes, the focus on optimization and efficiency is one that prioritizes corporate interest and maximizing shareholder values over human well-being. The implication being the release of AI tools and systems that require significant upgrades to address issues of bias, security, and data privacy.

One example of how AI undermines the notion of dignity from a relational perspective is the use of facial recognition technology. The use of facial recognition technology (FRT) remains a highly debated subject in the ethics of AI. In most cases, its use, at least by the police, is strictly for the identification of criminals and often for biometric security, which could be instrumental in finding and identifying missing persons, preventing identity theft, and the like. However, two primary issues often crop up in the conversation – accuracy and privacy. Firstly, facial recognition tech has been proven over and again to not be very accurate and to misclassify people of darker skin tones and especially women by over 30% (Hardesty, Reference Hardesty2018; Buolamwini, Reference Buolamwini2020; Learned-Miller et al., Reference Learned-Miller, Ordóñez, Morgenstern and Buolamwini2020). Although the accuracy has improved on many facial recognition software, it remains a cause for concern. Secondly, there are concerns about the training dataset and the lack of consent in acquiring images used to train the models.

Considering our conversation on African ethics, the concept of facial recognition glosses over the acknowledgment of the intrinsic worth of an individual. In effect, the technology uses facial characteristics to generate mathematical representation to map out features like eye size or spacing, nose length and width, facial structure, mouth curvature, width, and openness, head size and shape, etc. Using these characteristics, human faces are often matched against large databases of images or videos. The obvious consequence is that individuals affected by the inaccuracies of the prediction of these systems are often not considered from the point of intrinsic worth but the statistical relevance of their facial features. The knock-on effect, especially in the use of facial recognition by the police, is that victims of this misclassification could end up in jail, truncating their ability to commune with family and friends, and upending their lives. However, a holistic approach of considering a person to be more than just the sum of their facial features and seeing them as persons with intrinsic worth and a capacity to commune is essential to implementing some criteria for the responsible use of this technology.

Another concern with the use of facial recognition technology is with respect to the issue of privacy. In the world’s first legal case (Sabbagh, Reference Sabbagh2020) against the police use of facial recognition technology, Ed Bridges challenged the South Wales Police at the High Court, citing that it has “used the tech on more than 60 occasions since May 2017 and may have taken sensitive facial biometric data from 500,000 people without their consent” (Liberty, n.d.). In September 2019, the High Court ruled that the South Wales Police’s use of FRT does not interfere with the privacy rights of individuals scanned. This decision was appealed, and the Court of Appeal found the South Wales Police’s use of the technology to be in breach of “privacy rights, data protection laws and equality laws”. When sensitive and personally identifiable information/data of individuals are collected without consent, it indicates a failure to view these persons as worthy of respect and respect for their human rights. From a relational ethics perspective, exploiting private data for economic gains or without informed consent demonstrates a lack of respect for individual dignity. If one takes dignity to involve the recognition of the capacity to commune, then one would imagine that such a capacity is identifiable by other sentient communal agents for whom such a capacity is apparent. This is because it is difficult for one to argue that the capacity for an individual to relate with others in positive ways is the sort of thing that can be accounted for by an AIS whose sentience and capacity to commune is not yet as robust as those of humans.

If this is true, then it becomes clear that AI systems like the Correctional Offender Management Profiling for Alternative Sanctions (COMPAS) cannot be used to make court judgments or assess the likelihood of a defendant becoming a recidivist. This is because tools like COMPAS cannot provide the necessary guidance, context, or nuance to decipher whether an individual possesses the capacity to relate positively with others – that is, possesses dignity. Beyond this, one can see how such an AI system could undermine the values of forgiveness, as well as the process of reconciliation and restitution, which are all significant values within the context of personhood as self-actualization. By placing undue emphasis on historical data as the groundwork for the AI supporting decisions about an individual’s likelihood to be a recidivist, that individual’s dignity as well as the possibility of revitalising her personhood is truncated.

Another notable example of how individual uniqueness and worth is glossed over and seen only for their statistical and financial relevance is in the use of AI for personalized programmatic advertising. With the high premium placed on data and data being regarded as the new “oil,” we have seen an uncanny uptick in mass surveillance by technology companies. There is a penchant for technology companies to gather data on users’ preferences and track every activity carried out on mobile phones, smartwatches, and computers to profile and sell such data to third-party for targeted advertisement. The implication in light of African ethics is that the individual is considered a mere commodity and their intrinsic worth is seen as less important or superficial at best. Companies now use AI to exploit users’ vulnerabilities through recommender systems by gathering as much data as can be gotten and nudging users to indulge their proclivities.

In many cases, impressionable persons can be exploited into shopping more, gambling more or even spending inordinate time on the internet, as the goal is to increase “stickiness” – the amount of time spent engaging in the use of a product, especially digital products. Indeed, as some have noted (see: Yuan et al., Reference Yuan2011), social media and other digital spaces have been created in ways that make it addictive to the user. Of course, by keeping the individual immersed in these digital experiences, the individual is set up for all forms of targeting and exploitation, which the individual must be a party to since s/he is psychologically addicted to the product. Again, in this case, the individual is seen, treated, and exploited as a means to a variety of ends that have little to do with the individual herself.

By viewing a user as a mere commodity, an individual’s private/personal information is constantly collected and sold for profit – often without their informed consent. Sometimes, private information, posted in the virtual domain, often finds its way to the public domain, and beyond the control of the individual involved. As scholars like Whitehouse (Reference Whitehouse2010, 310) note: “In a British Press Complaint Commission survey, more than three-quarters of adults online would change information that they digitally publish about themselves if they thought it would end up in the mainstream media” (Reference Kant2008). Never mind that social network users may fail to impose privacy settings offered to them and may fail to use good judgment in selecting photos and making status updates. Sometimes, it is not the spread of private information that is the problem but the spread of misleading information with the assistance of AIS (Chimakonam, Reference Chimakonam2020). Apart from problems related to defamation, the subtle nudge of individuals towards certain political lines (as well as the monetization of the coercive power) presents itself as another case in which the mindset of the individual is commodified (Confessore, Reference Confessore2018).

While the commodification of the individual, in this example, is problematic, also notice that the individual who is violated in this way, is not thought of as end that is valuable in themselves, but as merely a means to profit, or other end-goals like winning an election (Confessore, Reference Confessore2018). As a consumer, the individual’s preferences are harvested through AIS which collates browsing habits to target the individual with customized advertisements tailor-made to encourage the individual to make purchases based on an artificially whipped-up desire for a product.

In all these instances, the focus is not placed on improving the individual’s capacity to commune or be relational, nor is the focus on the development of the individual’s vitality/creative power. Individuals are not only treated as means to an end but are treated in ways that actively water down their dignity and their ability to pursue/acquire that dignity. In other words, the current trajectory of AIS fails to account for the value of human dignity from African personhood. Perspective, whether in terms of acknowledging that dignity or building it.