The proliferation of Artificial Intelligence (AI) into the creative industries, especially in the context of Indigenous African music, is a deep-rooted political action that builds upon the historical dynamics of power of digital extraction and algorithmic colonialism. While existing scholarship predominantly critiques AI through Western intellectual property paradigms, it largely overlooks the training dataset as a primary site of epistemic violence. Based on a decolonial theoretical framework, this paper fills this gap by arguing that the AI training datasets are political artefacts that fix neo-colonial forms of oppression by freezing dynamic, living cultures to be consumed exogenously. The study utilises a triangulated research design and intentionally uses a multi-regional Pan-African scope of analysis, triangulating three analytical fields: a historical excavation of the 1939 Mbube (Solomon Linda) case in South Africa as an analogue blueprint of extraction; a netnographic study of five highly active industries. The findings reveal a ‘dual architecture’ of digital neo-colonialism, wherein external data colonialism via uncompensated scraping is facilitated by internal structural deficits and outdated post-colonial legislation. Empirically, the netnographic data reveals a phenomenon of ‘silent extraction,’ where grassroots resistance is actively overridden by a consumer-driven adapt or die techno-optimism. To address this perceived epistemic alienation and legal gap, the paper concludes that the way to realize ethical AI is by shifting the paradigm to Indigenous Data Sovereignty (IDSov) and the Collective Benefit, Authority to Control, Responsibility and Ethics (CARE Principles). In direct relation to the structural gaps that were revealed through the research, the paper proposes that the African Musicology Institute (AMI) should be created as a free, sovereign infrastructural centre to manage ethically sourced datasets. Finally, this study alters the academic debate of reactive copyright litigation to proactive infrastructural autonomy, which offers a critical map of computational self-determination in the Global South.
Kirui et al. (Sat,) studied this question.