A Sovereignty Centred Model of AI Literacy for African Higher Education
This paper critiques the context-blind nature of prevailing AI literacy frameworks in African higher education and proposes the AI Literacy for Sovereign Learning (AIL-SL) model, which integrates cognitive-technical, critical-epistemic, ethical-normative, and socio-cultural capacities to foster epistemic sovereignty rather than mere technical dependency.
Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
In the modern university, a quiet revolution is reshaping how knowledge is created, shared, and judged. For centuries, education relied on human authors and physical texts, where a student could trace an idea back to its source and understand the context in which it was born. Today, artificial intelligence systems generate text, summarize research, and decide which information appears most credible. These systems act as powerful intermediaries, standing between the learner and the world of ideas. While schools are rushing to teach students how to use these tools, a critical question remains unanswered: does knowing how to operate a machine mean you truly understand the knowledge it produces? This is the heart of a new perspective emerging from African higher education, where scholars are asking whether learning to use imported technology without understanding its hidden assumptions might actually make students more dependent, rather than more free.
The researchers behind this perspective, a team of scholars from the Democratic Republic of the Congo and beyond, argue that the current way of teaching "AI literacy" is fundamentally flawed for their context. Most existing programs treat AI skills like a checklist of tasks: learn to write a prompt, learn to spot errors, learn to use the software. They assume that if a student gets better at these tasks, they become more independent thinkers. The authors suggest this is a dangerous mistake. They propose that in places where the technology is built elsewhere, trained on data from other cultures, and governed by distant corporations, simply getting better at using the tool can deepen a student's reliance on it. A student might become fluent in asking the machine for answers, yet lose the ability to question whether those answers are true or relevant to their own reality.
To address this, the team developed a new framework called the AI Literacy for Sovereign Learning model. Instead of viewing literacy as a pile of separate skills, they see it as a complex relationship between four different capacities. First is the technical skill to operate the system. Second is the critical ability to question what the system knows and where its knowledge comes from. Third is the ethical sense to judge how the system affects fairness and justice. Fourth is the cultural capacity to understand how the system fits—or fails to fit—local ways of thinking and community needs. The authors argue that true independence, or "epistemic sovereignty," only happens when these four capacities grow together. If a student becomes highly skilled at using the technology but lacks the critical or cultural skills to question it, they do not become autonomous; they become a more efficient servant to the machine.
The paper challenges the idea that you can simply add a lesson on "ethics" or "sovereignty" to a standard computer class and expect a good result. The researchers suggest that the order in which students learn matters deeply. If a curriculum teaches students how to use the tools before they learn how to critically analyze the system's biases, the students may develop a false sense of confidence. They might trust the machine's output simply because they know how to make it work well. The authors warn that this creates a specific kind of dependency where the learner is so comfortable with the technology that they stop noticing its limitations. They argue that for students in Africa, and indeed anywhere where the technology is not locally owned, critical thinking must be the foundation upon which technical skills are built, not an afterthought.
This perspective also highlights a gap in how we measure success. The authors point out that while there are sixteen different tests designed to measure AI literacy, none of them have been properly tested to see if they work across different cultures. A test designed in one part of the world might assume that "critical thinking" looks a certain way, missing the specific ways a student in a different culture might identify bias or error. The researchers suggest that without fixing this, universities risk importing tools that measure the wrong things, giving a false sense of security to institutions that are actually becoming more dependent on foreign systems.
The implications for teachers are significant. The paper argues that the biggest bottleneck in this new kind of learning is the teachers themselves. While a student might learn to type a prompt by reading a manual, learning to critically question the system's worldview requires a teacher who can model that behavior. If a teacher has not developed their own critical and cultural understanding of AI, they cannot guide students to do the same. The authors suggest that professional development for educators must shift focus from just teaching the software to helping teachers understand the deeper questions of knowledge, power, and culture that the software raises.
Finally, the researchers emphasize that individual learning cannot solve the problem alone. Even if a university produces brilliant, critical thinkers, those students cannot change the system if the university itself has no say in which technologies it buys or how they are used. The paper calls for institutions to create channels where students and staff can influence decisions about AI adoption. Without these channels, the hard work of developing critical minds may be wasted, as the institution continues to purchase and use systems that reinforce the very dependencies the students are learning to resist.
The authors are careful to note that their work is a conceptual proposal, a map for how to think about the problem rather than a report of finished experiments. They have not yet tested their specific predictions with new data, but they have built their argument on a careful review of existing research and the lived experience of African scholars. They offer a set of clear claims that can be tested in the future, inviting other researchers to check if their model holds true. By shifting the focus from simple technical proficiency to a deeper, culturally grounded form of independence, this work offers a vital new direction for education in an age of intelligent machines. It suggests that the goal of education should not just be to teach students how to use the tools of the future, but to ensure they remain the masters of their own minds while doing so.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.