With the world increasingly looking toward the potential opportunities and threats presented by artificial intelligence (AI), Africa finds itself at a critical juncture: the continent must develop its own policy, ethics and governance approach to the development and use of AI.
On the one hand, Africans have become acutely aware of big tech’s search for cheap data labourers in West Africa following the SAMA scandal; and generative AI testing in African contexts has fuelled disinformation long before Cambridge Analytica admitted to manipulating elections in Kenya. On the other hand, AI-based systems have been deployed by Google AI Center in Ghana to improve the detection of locust outbreaks and the Kenyan e-banking service M-Shwari has helped many non-urban residents get access to loans.
So in the wake of Security Council dialogues on achieving a UN watchdog operating under an international framework for the ethical governance of AI, what philosophical traditions can an African framework pull from in its search?
In one of his more famous papers, “How Not to Compare African Traditional Thought with Western Thought”, the Ghanaian philosopher Kwasi Wiredu showed that it was important to embrace a process of refinement of African philosophical tradition as foundational to future endeavours within the continent that required solutions to new social questions.
There are two major questions that have undergirded my own inquiry into AI ethics: (1), How do healthy and constructive human-machine relations look like? (2), How do we safeguard historically marginalised populations in the expansion of this new global sector?
As I have written elsewhere, “New technologies develop their own axes of exploitation that are systematically incorporated into social, geographical and economic relationships. Vigilance is always required.” These two questions resound because any technological advancement as unprecedented as AI must neither be left unguarded nor be deployed in a manner that allows it to replicate old global patterns of asymmetry. Vigilance on the continent begins with the agreement to and embracing of an African position and approach to AI ethical governance.
The various traditions of African humanism offer substantial inputs into both questions. As a Southern African, Ubuntu is one such humanistic tradition that I am most familiar with. Ubuntu is a relational philosophy, much like Ujamaa, that articulates values of collective and communal approaches to life and work.
These values stand in stark juxtaposition with the utilitarianism that guides Western normative approaches to AI governance, an approach that aims to construct AI that maximises what is useful and beneficial for the majority of human beings and, in the long run, minimises what is not. Although this is generally desirable, humanistic traditions offer a more comprehensive contribution.
Under humanistic tenets, we achieve two main things. Firstly, we recognise the collective ownership that all peoples have of AI. Secondly, inclusion. These values can be implemented in technological development and functioning, as we consider the ethical aspect of the development and use of AI.
Collective ownership allows acknowledgement of the truth that all tech is human: that AI technology depends on human beings as much as human beings are growing to depend on them. Therefore, all labour and human inputs, despite originating from the subaltern or so-called ‘non-skilled labourers’, is regarded as essential to functional AI systems. The tenet of collective ownership can be subsumed into ethical governance frameworks that assist with the implementation of universal values of justice. If all tech is human, a governance structure underpinned by justice motivates ethical data sourcing and penalises big tech entities and data brokers for misusing personal data for profit or disinformation.
Moreover, the tenet of inclusion helps us understand that equitable inclusion achieves the highest good for all peoples, not just the majority and the most privileged. African humanism, in all its iterations, embraces the rule that we do unto others what we believe they ought to do to us. It destroys older relations of domination and subjugation through the acknowledgement of the common humanity of all human beings. It reconciles conflicting economic and social justice imperatives. It also achieves solidarity through global communal relations. Ethical governance frameworks that are underpinned by humanistic inclusion can empower the position of historically marginalised populations like those in Africa. It prevents AI outputs from disproportionately disadvantaging said populations, be it through the use of AI-controlled weapons systems in conflict zones or the proliferation of racial, sexual and ability biases in tech systems.
These two tenets show us that African philosophical traditions are profound sources of modern ethical questions. The splendour is also in the fact that the relational emphasis of humanistic traditions like Ubuntu is not unique to Africa. It is also emphasised in various indigenous philosophies all over the world. Therefore, a framework emerging from these values can be widely applied across the globe, especially in the Global South.
As observed in the gatherings under the UN Global Pulse held in Tunisia and Ghana, Africa stands to benefit from a proactive rather than reactive stance to AI ethical governance. A process of ethics development that precedes further technological advancement is a worthwhile investment that can buttress sustainable technological advancement across the world, that is defining and refining values and subsequently aligning AI technologies.