Artificial Intelligence and Education in the Global South: A Critical Review
Book Details:
Reimers, F., Azim, Z., Palomo, M. R., & Thony, C. (2025). Artificial Intelligence and Education in the Global South: A Systems Perspective. Springer. https://doi.org/10.1007/978-3-032-11449-5
Reviewer:
Albert Schram, Ph.D.
Former Vice-Chancellor and Education Consultant
Introduction: The Context of Crisis
To evaluate the potential of Artificial Intelligence (AI) in the Global South, one must first confront the brutal reality of the canvas upon which this technology is being painted. We are not merely facing a "challenge"; we are witnessing a generational catastrophe, the infamous global learning crisis. By now boys and girls globally go to school, but there is one 'minor' details: they are not learning anything, not reading, not writing and no artythmatic. The targets of Sustainable Development Goal 4 (SDG4)—inclusive and equitable quality education for all—are not just being missed; in many regions, they are receding into the horizon.
According to the World Bank (2022), "learning poverty"—the percentage of 10-year-olds unable to read and understand a simple text—has soared to nearly 90% in Sub-Saharan Africa. Having spent over 18 years working in educational leadership and development economics in Central America, the Caribbean, and the South Pacific, including a tenure as Vice-Chancellor, I have walked through the classrooms that statistics merely describe. I have seen the hollowed-out infrastructure, the teacher absenteeism driven by unpaid wages, and the systemic inertia that deflates the most spirited reform efforts.
It is against this backdrop of severe systemic failure that Fernando Reimers and his colleagues from the Harvard Graduate School of Education present Artificial Intelligence and Education in the Global South: A Systems Perspective. The authors, secure in their tenures at Western institutions, approach this topic with what they term "open-minded agnosticism." While their scholarship is rigorous and their intentions noble, there is an inescapable distance between the view from Cambridge, Massachusetts, and the view from a rural school in the Papua New Guinean highlands, or a township in Gauteng. The reality on the ground is often far worse than academic frameworks can capture. However, this volume represents a critical, if occasionally overly optimistic, first step in mapping a terrain that is changing beneath our feet.
The Systems Perspective: Theory vs. Entropy
The central thesis of the book is that AI must be viewed not as a silver bullet, but through a "systems perspective." In fact, education has not changed, we only possess a more powerful tool that can enhance teaching and learning. Reimers et al. argue that education is a complex web of interactions involving curriculum, assessment, teacher capacity, and governance. They posit that introducing AI without aligning these components will merely result in "improvement"—automating existing inefficiencies—rather than the necessary "transformation" required to solve deep inequities.
The book is structured logically, moving from the core of learning (Chapter 3) outward to curriculum (Chapter 4), assessment (Chapter 5), teacher development (Chapter 6), school management (Chapter 7), and finally, governance (Chapter 8). This holistic approach is intellectually sound. The authors correctly identify that dropping tablets into a classroom without changing the curriculum or training teachers is a recipe for expensive failure—a lesson the development community has learned repeatedly with previous waves of EdTech.
However, the "systems perspective" assumes a functioning system exists to be optimized. In many contexts in the Global South, the system is not merely misaligned; it is broken or predatory. Per million inhabitants there may be 10.000 or more teachers missing, and thousands of schools that were never built. The authors discuss "feedback loops" and "coherence," concepts that presume a level of administrative competence and political will that is frequently absent. When a Ministry of Education is paralyzed by corruption or bureaucratic ossification, a "systems approach" can become a trap, waiting for a top-down alignment that will never arrive.
The Necessity of Bottom-Up Innovation
This leads to a critical observation that the volume touches upon but does not fully embrace: the necessity of bottom-up, village-level activities when systemic changes are unlikely.
The authors laudable focus on "AI Unplugged" (Chapter 3) and low-connectivity solutions like SMS-based learning (Chapter 4) hints at this. They acknowledge that high-tech solutions widen the digital divide. However, the review of governance (Chapter 8) still leans heavily on the idea of state capacity building. My own experience suggests that in the most fragile contexts, we cannot wait for the "system" to almost magically reform itself, unless one is willing to wait decades if not centuries. This resonates deeply with my own economic analysis of continuing almost complete illiteracy in 19th century Spain and Italy.
We must consider how AI can empower localized, decentralized education clusters. If the central curriculum is rigid and irrelevant (a common complaint noted in Chapter 4), can AI-enabled tools allow village elders and local teachers to generate context-specific learning materials that bypass the central bureaucracy? The book discusses the "localization" of content (e.g., translating textbooks into local dialects using AI), which is promising. Yet, true transformation might require bypassing the "system" entirely to deliver direct-to-learner or direct-to-community educational sovereignty. When the state fails to provide, the village must provide, and AI could be the lever that makes this feasible without requiring the massive infrastructure of traditional schooling.
The Teacher as the Fulcrum
Perhaps the most compelling argument in the book, is found in Chapter 6: AI and Teacher Development.
The authors rightly identify that teachers in the Global South are overwhelmed. They face large class sizes, administrative drudgery, and a lack of support. Here, the book shines by reframing AI not as a replacement for teachers, but as a mechanism to restore their humanity. The authors advocate for AI to handle the "robotic" tasks—grading, attendance, basic lesson planning—so that teachers can focus on the "human" tasks: mentorship, social-emotional support, and complex instruction. Like in the West, if AI can free up 5-6 hours per week per teacher, this would go a long way in rehumanizing the teachers.
This is the economic argument for AI in education: increasing the productivity of the scarcest resource—the qualified teacher. As the authors note, "AI can act as a coach," helping novice teachers practice through simulations or analyze their own teaching via dialogue analysis.
However, the implementation of this vision requires a pivot in investment. Governments and donors often prefer buying hardware (which offers photo opportunities) over the invisible, messy work of training. The book emphasizes that "teacher-in-the-loop" is essential. I would go further: investment in training teachers to use AI ethically and effectively is the single highest-yield investment available. If a teacher in a remote village can use a Large Language Model (LLM) to generate a high-quality lesson plan that incorporates local folklore and meets national standards in five minutes—a task that previously took two hours—we have fundamentally altered the economics of education delivery.
But this requires training that goes beyond technical literacy. It requires what the authors call "AI Pedagogical Content Knowledge" (PACK). Teachers must be trained to identify hallucinations, bias, and cultural irrelevance in AI outputs. Without this critical literacy, AI becomes another tool for colonization, importing Western epistemologies into Global South classrooms under the guise of efficiency.
Assessment and Governance: The Double-Edged Sword
Chapter 5 (Assessment and AI) and Chapter 8 (Governance) present the most signifcant ethical risks. The authors discuss the shift from high-stakes summative testing to continuous, AI-driven formative assessment. In theory, this is the Holy Grail of education: real-time feedback that allows for personalized learning paths. Several AI powered EdTech platforms are capable of providing quick, precise and useful feedback, somthing teachers simply can not do given the number of pupils they cater for, and the general lack of resources (e.g. paper, electrcitiy, computers, printers).
However, in the context of the Global South, where data privacy laws are often weak or non-existent, the potential for surveillance is terrifying. The authors acknowledge these ethical concerns, discussing "data privacy and trust" (p. 130). Yet, the danger of "algorithmic governance" in authoritarian or weak states is profound. If resource allocation (Chapter 8) is determined by opaque algorithms analyzing student data, we risk automating inequality. A rural school might be defunded because an algorithm predicts poor outcomes, creating a self-fulfilling prophecy.
The review of governance highlights the lack of capacity in ministries to manage these tools. This is an understatement. Placing powerful predictive tools in the hands of untrained or corrupt administrators is dangerous. The book’s call for "human-in-the-loop" oversight is vital, but it assumes the human in the loop is benevolent, accountable, and competent.
Conclusion
Artificial Intelligence and Education in the Global South is a necessary, rigorous, and comprehensive volume. Reimers and his team have provided a vocabulary and a framework that policymakers and researchers desperately need. Their insistence on moving beyond "improvement" to "transformation" is the correct strategic north star.
However, as a practitioner who has navigated the muddy waters of implementation in all cycles from early-years to tertiary, I offer a note of caution. The "systems perspective" is a luxury of stable states. In many parts of the Global South, we are operating in survival mode. The most immediate promise of AI lies not in optimizing the Ministry of Education’s dashboard, but in placing a powerful, localized, and offline-capable tutor in the pocket of a teacher in a remote village, and giving that teacher the training to wield it.
We must avoid the trap of "isomorphic mimicry"—building education systems that look like those in the West (now with AI!) but fail to function. Previous efforts have produced the "global learning crisis, where almost everybody is in school but not learning. Instead, we should use AI to empower the grassroots, reducing the administrative burden on our overworked teachers and allowing them to return to the sacred task of teaching. This book is a valuable map, but the territory is far rougher than the map suggests.
References
Reimers, F., Azim, Z., Palomo, M. R., & Thony, C. (2025). Artificial Intelligence and Education in the Global South: A Systems Perspective. Springer.
Schram, A. (1997). Railways and the formation of the Italian state in the nineteenth century. Cambridge University Press. (Reviewed by Frank Dobbin in The Economic History Review, 52(1), 1999).
World Bank. (2022). The State of Global Learning Poverty: 2022 Update. The World Bank. https://www.worldbank.org/en/topic/education/publication/state-of-global-learning-poverty
No comments:
Post a Comment
Note: only a member of this blog may post a comment.