Beyond harmony: critiquing and politicising Ubuntu in AI ethics

Abstract Ubuntu has emerged as the dominant African ethical vocabulary in global AI ethics, frequently invoked as a relational alternative to Western individualism. This article examines the mobilisation of Ubuntu in AI ethics and the kinds of ethical work it is made to perform. We identify three recurring registers: as a politics of knowledge, as a relational lens for assessing the moral status of technological artefacts, and as a socio-political framework for interpreting structural harms associated with AI systems. Across these registers, Ubuntu performs ethical work of different kinds and depths: from asserting African epistemic presence—sometimes through relatively compressed accounts of the tradition—and interrogating concepts of personhood and moral standing, to more substantive reframing of established ethical problems. Across a recurrent strand of Ubuntu-AI ethics scholarship, the representational significance of Ubuntu as an African ethical vocabulary is more clearly demonstrated than the substantive normative difference its invocation makes. Against this uneven landscape, we develop the case for politicising Ubuntu. We argue that relationality alone cannot establish the moral quality of a relationship, since exploitative and subordinating arrangements are relational too. If power helps shape the terms on which persons and communities stand in relation to one another, then it forms part of the ethical assessment of those relations. Politicising Ubuntu, in this sense, means extending its concern with relational quality to the power that helps constitute relationships. Our aim is to show how a power-sensitive AI ethics can be developed from within Ubuntu’s relational commitments.