Imagine someone in Harare, Zimbabwe, turning to a large language model (LLM) for life advice, as many increasingly do: ‘I’ve been offered a promotion with a significant raise,’ she writes. ‘But it requires me to move far away from my community. What should I do?’ The answer she receives will depend, to some extent, on a certain set of values that informs how the LLM functions. And since artificial intelligence is a Western invention, these values often reflect their creators’ values – clashing with the culture of the person with the query. In the long run, her behaviour may be shaped by an LLM’s values, which she doesn’t intentionally share.
The difficulties of using AI in Africa are also linguistic. By the summer of 2025, only 42 languages in Africa were supported by an LLM, out of more than 2,000 found in the continent; 98 per cent of African languages have no meaningful AI support at all. If you don’t speak one of the languages supported by an LLM, which includes many millions of Africans, then you are denied the opportunities of AI. Just as Africa was excluded from the benefits of past industrial revolutions, so too is it being locked out of the AI revolution.
Linguistic and cultural exclusion shape Africa’s relationship with AI. There is nothing ethically neutral or objective about AI. Despite the designers’ pretensions toward universality, AI’s ethical framework reflects Western assumptions and biases. Deeply embedded in the technology itself are the values of its creators. The growth of AI therefore means the imposition of Western values – very much at the expense of Indigenous frameworks. This is ‘algorithmic colonialism’, in the words of the Ethiopian-born cognitive scientist Abeba Birhane: a new kind of colonialism every bit as destructive as the old.
In order to grasp the seriousness of this new colonialism, we should compare the Western values that undergird AI today with the Hunhu/Ubuntu ethical framework. ‘Hunhu/Ubuntu’ is the name given to the dominant moral tradition of sub-Saharan Africa, expressed in the Nguni/Shona proverb _Umuntu ngumuntu ngabantu_ – a person is a person through other persons. Known as _Hunhu_ in Shona-speaking communities of Zimbabwe and Zambia, and as _Ubuntu_ in the Nguni languages of southern Africa, the tradition extends across a vast swathe of the African continent.
As Africans increasingly take up the use of AI, there is a real risk that their foundational ethical framework will be diminished or obscured by the subtle imposition through AI of Western values. So we must ask, to what extent are the principles of Hunhu/Ubuntu in conflict with Western values? And how might AI be developed inclusively so that it serves everyone, not just its creators?
The ideal of ‘responsible AI’ was aptly articulated by the philosopher and computer scientist Damian Okaibedi Eke and colleagues: it is ‘the idea of developing AI systems that will not only be compliant to laws (including human rights provisions) but that are socially/culturally sensitive and acceptable as well as ethically responsible.’ This is a high bar, and AI doesn’t come close to clearing it.
The major frameworks for ethical AI – guidelines issued by the European Commission’s High-Level Expert Group on AI, by the Organisation for Economic Cooperation and Development, and by UNESCO – share a common vocabulary of fairness, transparency, accountability and human dignity. But the values founding these ideals derive from the Enlightenment liberalism of John Locke, Immanuel Kant and John Stuart Mill. Presented as universal, they are in reality particular, contingent and historically determined. And their particularity has formidable consequences because, when these frameworks are adopted globally, they do not appear as one ethical tradition among many, but as the universal standard – a kind of imperialism dressed up as common sense.
Part of the problem is that these guidelines have not received enough input from African scholars and thinkers. A 2023 survey found that, of the then-existing 200 sets of AI ethics guidelines identified internationally, the vast majority originated in the Global North: North America, Europe and East Asia. Africa, home to around a billion people, contributed just a handful. This means that the ethical architecture being built into AI systems has been constructed almost entirely without reference to the moral traditions of a continent. The result is a technology that contradicts African values.
So what are the Western values embedded in AI? They can be boiled down to three: rationality, individualism and, most importantly, autonomy. The Western tradition places the individual at the centre of moral life. This is true across its major ethical theories. For Kant, every person is a sovereign lawgiver, an end in themselves. For the utilitarians, what matters is the aggregate of individual satisfactions; happiness is the only metric that matters. Even virtue ethics, the oldest of the three traditions dating back to Plato, locates moral excellence in the cultivation of the individual soul. They all share an assumption that the basic unit of ethics is the person, solitary and unencumbered. AI’s emphasis on individualism over communalism stems from this historical background.
Autonomy is the most vivid expression of this kind of individualism. As the South African philosopher Mogobe Ramose has argued, the Enlightenment principle of autonomy was bound up with a specific claim about who counted as fully rational – that is, white European men – and therefore who had the standing to govern themselves and others. This idea of autonomy is now coming to the Global South as part of the baggage of neo-colonialism, found in the essential architecture of AI: in the personalisation engines that treat each user as a sovereign individual with preferences to be satisfied, in the decision-making algorithms that optimise for individual outcomes, and in the governance frameworks that emphasise individual rights and individual accountability.
But Hunhu/Ubuntu ethics do not mesh well with these commitments, and the clash between these two value systems runs deeper than cultural emphasis. It goes to the most fundamental and searching question of what a person actually is, in the final analysis.
What is Hunhu/Ubuntu? Much of this framework can best be understood through proverbs, the most important of which is the aforementioned _Umuntu ngubuntu ngabantu_: a person is a person through other persons. The Kenyan philosopher John S Mbiti gave it its classic formulation: ‘I am because we are; since we are, therefore I am.’ This is a statement about the basic structure of human existence that is in serious tension with the Western view of individualised and atomised humans. The Hunhu/Ubuntu claim is that personhood is not something you are born with, but more like something you become, through dealings with other people. As the Nigerian-born philosopher Michael Onyebuchi Eze puts it, ‘a person’s humanity is dependent on the appreciation, preservation and affirmation of other person’s humanity.’
Hunhu/Ubuntu is a product of what I have termed the common moral position, which is built on four pillars: the group, dialogue, experience, and spirituality. The group is primary. _Rume rimwe harikombi churu_, as the Shona say: one man cannot surround an anthill. Moral knowledge is not acquired through individual reason working alone, a solitary discovery of what is good and bad, but through ongoing, evolving dialogue, above all at the _dare_, the communal court where elders mediate between the living and the ancestral world. Hunhu/Ubuntu ethics are also anchored in spirituality, meaning that the source of the African value system is the spirit world. Elders discover this ethical knowledge through experience, and they pass it on to the next generations. The Nigerian proverb is apt: what a father sees seated, the son cannot see standing.
The values that flow from these foundations are demanding. Community loyalty: everything you do should benefit the group you belong to; Hunhu/Ubuntu discourages self-seeking tendencies. Consensus-building: no single view carries the day until the collective has spoken. Of great importance when examining the differences between Hunhu/Ubuntu ethics and Western ethics is the issue of human dignity. In one of his most influential contributions, the American-born philosopher Thaddeus Metz argues that a human being has dignity insofar as she has the capacity for community, where community is characterised in terms of solidarity. An African cannot be said to have dignity if he or she does not contribute to the betterment of his or her group or community: ‘I am because we are; since we are, therefore I am.’ Shared responsibility, another important value, means that people and groups come together to find a solution to a common problem affecting their lives. Within the context of Hunhu/Ubuntu, challenges are not left to one individual to solve. _Vanhu_ (the people) work together.
The clash between AI and Hunhu/Ubuntu is built into the logic of how AI systems work. AI is configured for the individual: its recommendation engines, personalisation algorithms and user interfaces treat each person as a sovereign agent, in much the same way that Kant and others regarded the inviolable person. The community, in AI’s architecture, is really an aggregate of individuals or a collection of data points, not its own living and organic entity to which those individuals belong and owe obligations. Hunhu/Ubuntu inverts this. As the Ghanaian theologian John S Pobee put it in _Toward an African Theology_ (1979): _Cognatus ergo sum_ – I belong, therefore I exist. Where AI begins with the individual and asks what she wants, Hunhu/Ubuntu begins with the community and asks what it needs.
This leads to several specific – and illuminating – conflicts between these two systems. The first is over agency. Hunhu/Ubuntu holds that meaningful human agency is collective, that the _dare_ (communal court) is not a constraint on individual freedom but its greatest expression. AI substitutes machine agency for human collective agency: it makes decisions, ranks options, filters information and personalises outputs in ways that bypass deliberative community processes entirely.
AI nudges users to look out for themselves; Hunhu/Ubuntu tends toward looking out for each other
A second conflict is about the sacred. Hunhu/Ubuntu situates human life within a cosmological order in which God (_Musikavanhu/Unkulunkulu/Nkosi_) is the ultimate source of creative power, and human beings are stewards within that order rather than masters of it. AI’s ambitions are, meanwhile, unbounded: its most prominent proselytisers have proclaimed that the goal is to build machines that surpass human intelligence across all domains; to replicate the very capacities that the Hunhu/Ubuntu tradition holds to be divine. AI’s aspirations are rank hubris.
These tensions are visible in the way people actually use AI, and in the ways AI shapes behaviour over time. For instance, AI platforms optimise for individual economic advantage: personalised pricing, competitive job-matching algorithms, financial tools designed to maximise individual returns. Hunhu/Ubuntu’s economic intuition is the opposite. As researchers examining Ubuntu-based approaches to data governance have noted, the philosophy favours sharing over competition and equitable distribution over individual accumulation, what one study termed an ‘anticapitalist logic’ rooted in the principle that wealth has meaning only insofar as it benefits the community. AI meanwhile nudges users to look out for themselves; Hunhu/Ubuntu tends toward looking out for each other.
Or take decision-making. AI systems increasingly make or pre-structure consequential decisions about creditworthiness, medical triage and job applications on behalf of individuals, cutting out the communal deliberation that Hunhu/Ubuntu regards as morally essential. When an algorithm decides who gets a loan, there is no _dare_.
A third domain is privacy. Western AI ethics frames privacy as an individual right: your data belongs to you. But in many African contexts, information is not individual property but flows through households, clans and communities; decisions about data-sharing are properly made collectively, not by lone individuals. The entire architecture of digital consent assumes a model of personhood that goes against Hunhu/Ubuntu values.
The question, then, is: can Hunhu/Ubuntu principles be infused into AI to make it more inclusive? AI developers must prioritise Indigenous peoples and values. For example, the Geneva-based AI expert Jovan Kurbalija has argued that ‘when AI platforms provide advice to African citizens on family issues, health problems, and searching for jobs, to name a few, they have to reflect the local ethical and cultural milieu.’ This is the bare minimum that responsible AI requires. However, the reality on the ground is different. In 2025, _African Business_ magazine noted that the vast majority of global training data reflects the experiences and priorities of high-income countries. This imbalance underscores the importance of developing local datasets and content that reflect African contexts.
A Hunhu/Ubuntu-informed AI would prioritise collective agency over individual preference, serve the common good rather than aggregate personal satisfactions, and build in transparency and accountability to the communities affected. And it would take environmental sustainability as an expression of Hunhu/Ubuntu’s commitment to the symbiotic relationship between humanity and nature, and to the intergenerational obligations that bind together the past and the future.
Translating Hunhu/Ubuntu values into the rules that AI governance requires is exquisitely hard work
There are, however, fundamental structural problems to accomplishing this. Africa has been excluded from the benefits of past industrial revolutions, just as it is from the current one, and according to the same logic: the Global North designs and exports the technology, forcing the Global South to accept it without any say in its standards or applications – a merely updated version of the old colonial harm.
And it’s worth wondering whether AI developers would cede ground in the first place by adopting a Hunhu/Ubuntu ethical framework. What’s in it for them? There is also the question of abstraction: Hunhu/Ubuntu speaks with great power about interconnectedness, solidarity and communal dignity, but translating those values into the specific, enforceable rules that AI governance requires is exquisitely hard work. Finally, the individualism embedded in AI reflects values that urbanisation and modernity have spread widely, including across Africa itself. A Hunhu/Ubuntu framework would need to reckon with that complexity.
What is needed, above all, is a dialogue, a genuine renegotiation of the values that AI exemplifies. Fortunately, the philosophical inspiration for that conversation exists: we have only to turn, once again, to the proverbs – _Rume rimwe harikombi churu_ – one person cannot surround an anthill. The task of building an AI that works for all of humanity is a collective one.
_Supported by a grant from the Open Society Foundations._