Artificial intelligence (AI) is entering our daily lives and is quickly changing how we do things. It disrupts socio-material relations and induces new power relations. The average user’s understanding of these relationships is filtered through visible interfaces like computers, wires, robots, assistive devices, phones, etc.  

Yet beyond these physical things are virtual substructures such as servers, the internet, clouds, algorithms and other software — essential for AI functioning. They connect us to physical infrastructure and entities elsewhere. It is in these systems that massive individual and corporate data are being analysed, classified and categorised. 

Dr Angella Ndaka’s work on sustainable AI techno-futures uncovers interesting dynamics around these issues. Where are these physical infrastructures located? Who controls them, and for what purpose? What are embedded political and corporate interests? What are the social and environmental costs of maintaining them? And what is the transparency of the massive data the AI industry are continuously collecting?  

These questions create uncertainty as we spiral towards even more unanswered questions. Similar to author Timothy Morton’s hyperobjects description, AI to most people seems incredibly complex and hard to comprehend even though we interact with it every day. 

We are excited by its magnificence, yet we are scared of its depth. We all want to use it, no one wants to be left behind. Yet as we wilfully fall into its free space, the concept itself is frightening. AI’s complexity is hidden in the feel-good experiences that keep attracting us to it. Some scholars refer to this kind of technology as pervasive; as it loses its physicality, it disappears into invisibility and covertly weaves itself into our daily lives. Think of the cellphone or the cloud in our current society. 

In the case of AI, the more it disappears into invisibility, the further it entangles us in its underlying machinations, alongside the corporate and political interests whose agenda it serves. The question we ask is this: Who are we in this game? If we are not sure about our role in this hyperobject, have little information about the space, and the definition and apportionment of benefits are unclear — then how do we know whether “being included” in the game, as currently framed, means the inclusion that we might be seeking?  

Data capitalism

In the modern age, Africa has been locked into using imported technology from the Global North. Predictably, this has not produced the expected development heralded by local adoption. Instead, the dominant political economy has ensured that Africa continues to lag technologically. In the era of data-driven technologies and economies, AI will potentially produce worse consequences than the previous technological eras. AI technology, as it is currently, is an expression of extraction and data capitalism, colonialism and knowledge imperialism in Africa.  

Recent events on the continent that relate to the AI industry bring these issues to light. For instance, the massive and unjustified data collection and data appetite from the AI industry were highlighted in Fourcade and Healy’s Moral Views of Market Society. It exposes the extraction of cheap labour to power AI (to annotate, moderate and shape data) in the Global South (case of Meta in Kenya, 2023), and the recent unauthorised and unregulated data collection for AI in public places in Kenya (case of Worldcoin, 2023). 

The current invisible digital surveillance and the recent rhetoric about digital inclusion that justifies the unlimited amassing of data from unsuspecting populations and governments are a reflection of how powerful the opaque AI agendas in Africa are.  

This is coupled with increased funding that is targeting the development of AI technologies for Africa which mirror ideologies of the mother companies. The insidious consequences of these practices are masked by techno-optimistic messaging about AI. 

With the aid of badly formulated government policy, large amounts of funding are directed towards AI development, with less investment focused on its governance as proposed in McLennan’s paper Techno-optimism or Information Imperialism: Paradoxes in Online Networking, Social Media and Development.  

The government policy and tech development space therefore become dance floors for economic and corporate interests as well as the economic and political interests of government actors — neglecting the public interest. This has created restricted and narrow spaces for other well-meaning tech actors as well as alternative knowledge-holders to exercise agency on matters that compromise the wellbeing of humans and their environments.   

Agendas of power

The superficial narrative about digital inclusion and algorithmic bias where “the excluded” (or underrepresented) in AI/LLM training data must simply be adequately represented needs to be addressed critically and cautiously. History has shown all too well how science and technology can be deployed through dominant power systems to exploit black bodies or bodies of the South. From eugenics to the current data-driven classification and commodification, it has been shown that the agendas of power and technologists can be coupled to classify, subjugate and destroy those at the power margins.  

The question must then be posed: When we volunteer our data and ourselves in the name of digital inclusion, where are we being included? Whose agendas dominate in the technology being developed? And whose technologies are being produced anyway? Should we be so quick to volunteer for it?  

Answering these questions is not feasible when AI features remain silently hidden, invisible from users. This opacity directly affects how users assess the functionality, capability and possibilities of incorporating AI in their lives and the impact it has on wider society and the environment, as shown in Diefenbach’s Technology Invisibility and Transparency.  

The hyped “inclusion” rhetoric is a campaign for amassing data from the “unincluded” and unsuspecting groups, while African governments and local organisations are lured into the mythical call to the AI “hyperobject”, without questioning the underlying power structures and agendas. 

It should be noted that while the goal of invisibility might ostensibly be to make technology sophisticated, this opacity hides corporate and political interests, and the agencies that shape it, keeping AI users ignorant, as highlighted in Ndaka’s Sustainable AI Techno-futures. 

Until we can empower ourselves by gaining a better sense and control of these silences and hidden agendas, we might just consider whether we are better off not rushing blindly into submitting our data to AI in pursuit of inclusion. DM  

Dr Angella Ndaka is the executive director of the Centre for African Epistemic Justice, and a national consultant on gender and youth integration in digitisation at the Food and Agriculture Organization of the UN Kenya. Professor Geci Karuri-Sebina is an associate professor in digital governance at the Wits School of Governance. Both write in their own capacity. 

The Digital Afrikan is a journalism organisation with a mission to drive digital transformation in Africa. Visit our website or contact us on [email protected]