Decolonizing AI Ethics and Governance: Understanding the Impact of AI on Personhood and Identity From Within Indigenous Epistemologies

This article argues that mainstream artificial intelligence (AI) ethics and governance initiatives, often framed as “for people” or “for good,” do not meet their stated goals as grounded primarily in Western values and conceptions of personhood and identity flourishing. These initiatives often overlook non-Western and Indigenous epistemologies and ethical accounts, leading to AI ethics principles and practices that fail to address harms to Indigenous peoples, values, and identities. Consequently, prevailing approaches are not only epistemically and ethically inadequate but also unjust and harmful, perpetuating patterns and phenomena of Western cognitive imperialism, neocolonialism, global injustice, and epistemicide, at the expense of the most vulnerable and historically marginalized. To address these issues, this article carries out an ethical inquiry, drawing on decolonial and Indigenous scholarship, to highlight what personhood and identity flourishing mean and entail within three comprehensive epistemological and ethical frameworks: Ubuntu epistemology and ethos, Mapuche epistemology and Az Mapu ethics, and Mātauranga Māori and ethics. The analysis shows that the concepts and core relational values in these frameworks are largely missing from mainstream approaches and that current dominant AI design, deployment, and use practices significantly undermine them, with real consequences for affected communities. Finally, the article argues for the substantive integration of these concepts and values to decolonize mainstream AI ethics and governance discourse. To this end, it proposes the Indigenous “Two-Eyed Seeing” approach to bridge Indigenous and Western epistemologies and establish a conciliatory ground for a global AI ethics and governance framework that is epistemically plural, socially and ecologically sustainable, and ethically more inclusive and just.