By Beatrice Bonami
Whether Artificial Intelligence (AI) can be decolonial is a question that may never have a definitive answer. It echoes a broader dilemma: can former imperial powers truly be decolonial while still benefiting – politically, economically, and socially – from the empire structures created? So far, what we know about AI suggests the answer may be no. The ways AI is produced, funded, governed, and deployed follow familiar patterns of data capitalism and digital colonialism. These systems disproportionately extract from racialised and vulnerable communities so that wealthy nations remain competitive in the global digital economy.
This theme was central to the keynote delivered by Dr. Tshilidzi Marwala at the High-Level Meetings for Internet Digital Governance (HMDICE GOV), hosted by the United Nations University in Abuja, Nigeria. Dr. Marwala emphasised that as any field matures, it requires critical reflection to keep human rights at the forefront. Governance itself only gained global prominence after World War II, when the world sought unity and peace. Yet in the last three decades, this drive toward uniformity has increasingly given way to the recognition that countries and regions are diverse, and governance models must reflect this plurality. Marwala’s question, therefore, is not only about AI – it is about the relevance and legitimacy of global frameworks in a world that insists on diversity rather than uniformity.
AI played a central role at HMDICE GOV, where diverse stakeholders interrogated the assumption that every region must join the so-called “AI race.” I joined Geci Karuri-Sebina, Farai Mlambo, Amy Mutua, Kholiswa Malindini, and Angella Ndaka on a panel titled “AI Beyond Consumption”. We explored what it would mean for Africa to participate in the AI ecosystem on its own terms. We framed our session through three provocations – data, economy, and politics – which drew more than 100 multi stakeholders into a lively and critical discussion.
- Data: Africa’s persistent data poverty – limited capacity to generate, govern, and use high-quality, localised data that reflects African realities.
- Economy: Africa’s low-value position within the AI economy, despite its rapidly expanding digital labour force, markets, and demographic potential.
- Politics: Africa remains unlikely to meaningfully benefit from the projected $15.7 trillion AI economy by 2030 (data from World Bank).
These provocations sparked energised debates. Many participants expressed frustration that Africa remains primarily an extractor of raw resources and consumer of imported technologies. The analogy that resonated most came from Nigeria itself – Dangote’s oil refinery, a powerful symbol of how long it has taken for one African industry to fully benefit from its own resources. Others warned that challenging entrenched global structures may be risky or unrealistic, especially if it jeopardises short-term economic gains.
Among many considerations, I would like to bring here the focus on AI frugality and feminism as a topic to move forward in the context of digital governance. Together, they offer a path to move from extraction to agency by rethinking who controls data, digital infrastructure, and AI narratives. The Dangote refinery analogy shows what is possible when conviction meets vision: sovereignty is built, not given. The same must apply to AI. Rather than assuming progress requires resource-hungry large language models, we can advance through frugal AI – models that are context-specific, fair, low-resource, and environmentally responsible. Frugal AI reduces computational and natural resource demands while still achieving high levels of accuracy, making it a strategic response to data poverty and infrastructural constraints.
Such an approach supports technological sovereignty and economic viability. But it also requires non-extractive, equity-driven partnerships, grounded in shared values, mutual respect, and models like federated learning, which preserve local data control. A feminist perspective strengthens this shift. Feminist digital governance foregrounds relationality, inclusion, and collective well-being – challenging the patriarchal and imperial logics that currently shape global AI production. As Geci Karuri-Sebina notes, the Global South cannot achieve a decolonial AI future without stepping into a leadership role and critically interrogating dominant development models.
Ultimately, building alternative digital futures demands more than infrastructure. It requires imagination – the ability to design technologies that serve societies instead of markets, and that stay within planetary limits. HMDICE GOV concluded with a commitment to draft a Resolution for UN acknowledgment, calling for stronger South–South leadership in multilateral digital governance. It is a small but necessary step toward a fairer digital order.