A groundbreaking randomized trial in Edo State, Nigeria, has found that senior secondary school students who used Microsoft Copilot in dedicated computer labs for just six weeks achieved learning gains that experts describe as remarkably large—roughly equivalent to those typically seen from intensive small-group tutoring. The study, supported by the World Bank, is one of the first rigorous evaluations of generative AI in real-world classrooms in Africa, and its results are sending ripples through Nigeria’s education community. While some universities are moving to ban AI tools outright, the Edo trial suggests that the smarter path may be to embed them directly into teaching.
Conducted in 2024, the trial involved hundreds of students in Edo State’s public secondary schools. Participants were given access to Microsoft Copilot, the AI assistant integrated into Office 365 applications and Windows, and were guided through structured exercises in computer labs. The intervention was modest: six weeks of lab sessions where Copilot assisted with writing, research, and problem-solving across subjects. But the impact was anything but modest. According to preliminary data shared by researchers, the students using Copilot gained the equivalent of roughly an extra year of learning in areas like written composition and information synthesis, compared to control groups who followed the standard curriculum without AI support.
The trial’s design was robust. It employed a randomized controlled methodology, the gold standard for establishing causal impact. Students were randomly assigned to either treatment (Copilot access) or control (no Copilot) groups within the same schools. Both groups had the same teachers, same lab time, and same baseline assessments. The only difference was the AI tool. Post-intervention tests captured not just factual recall but higher-order skills such as structuring an argument, editing drafts, and synthesizing multiple sources—precisely the competencies that educators worry AI might erode.
While the full academic paper is still under peer review, early briefings have ignited a fierce debate in Nigeria’s higher education circles. Over the past two years, a growing number of Nigerian universities have issued blanket bans on generative AI, warning students against using tools like ChatGPT or Copilot for assignments and exams. The University of Lagos and Ahmadu Bello University are among those that have updated their academic integrity policies to classify AI-generated content as a form of plagiarism. The rationale is understandable: in a system already plagued by contract cheating and exam malpractices, administrators fear that AI will make it too easy for students to bypass learning altogether.
But the Edo trial challenges that fear. It suggests that when AI is deployed in a structured, teacher-guided environment, it acts as a learning accelerator, not a shortcut. Students didn’t just copy and paste; they used Copilot to brainstorm ideas, receive instant feedback on grammar and logic, and explore alternative ways of expressing complex concepts. The tool became a tireless tutor, available at the point of need, while the physical presence of teachers and peers kept the experience grounded in collaborative learning. This hybrid model—human instruction augmented by AI—may be the blueprint that Nigerian universities urgently need to adopt.
“Banning AI outright is a disservice to students who will enter a job market where these tools are already standard,” said Dr. Adaobi Nwosu, an education technology researcher at the University of Nigeria, Nsukka, who was not involved in the trial. “The Edo study shows that we can teach with AI, not against it. The key is pedagogy: we must redesign courses to assume AI is available, just as we assumed calculators were available after the 1980s.” Nwosu’s point echoes a growing global consensus. Organizations like UNESCO have called for AI literacy to be integrated into curricula, and countries from Singapore to Estonia are already doing so. Yet many Nigerian institutions remain hesitant, citing infrastructure gaps and ethical risks.
The infrastructure challenge is real. Edo State’s trial succeeded partly because the government had already invested in a network of well-equipped computer labs under the EdoBEST basic education program. Most Nigerian tertiary institutions lack such facilities. Rolling out a similar AI-augmented approach nationally would require substantial investment in hardware, internet connectivity, and teacher training. But advocates argue that the cost of inaction is higher. Without AI fluency, graduates will be less competitive internationally and less capable of driving local innovation. The World Bank’s involvement in the Edo pilot hints at potential financing pathways for scaling.
Ethical concerns also need careful handling. The trial’s researchers monitored for misuse and found that with proper instruction and active supervision, incidents of outright plagiarism were negligible. Students were taught to cite AI assistance and to critically evaluate its outputs—skills that are fast becoming essential digital literacy. Universities could build on this by developing honor codes that distinguish between appropriate AI collaboration and academic dishonesty. For instance, a student might be required to submit their chat logs along with a final essay, showing how they prompted the AI and revised its suggestions. This turns AI from a cheating tool into a transparency tool.
Another significant finding from the Edo trial was equity: the learning gains were largest among lower-performing students. This suggests that AI can help close achievement gaps by providing personalized support that overstretched teachers cannot always offer. In a country where public university classes often exceed 200 students, such personalized attention at scale could be transformative. Rather than penalizing students who use AI to level the playing field, institutions should be exploring how to provide it to all.
The trial also underscores the importance of local content. Microsoft Copilot performed well because it can draw on general web knowledge, but the researchers noted that its effectiveness was limited in subjects requiring deep cultural context, like Nigerian history or indigenous languages. This highlights the need for African AI systems trained on local data—an opportunity for Nigerian universities to lead research and development, rather than merely consume foreign tools. Teaching AI means producing graduates who can build the next generation of tools, not just use them.
As the Edo results circulate, some universities are already reconsidering their bans. The National Universities Commission (NUC), Nigeria’s regulatory body for degree-awarding institutions, has begun consultations on a national AI in education policy. A draft framework expected later this year may urge institutions to move from prohibition to pedagogy, setting standards for how AI can be used responsibly in teaching and assessment. Such a policy wouldn’t be without precedent. In March 2024, the Tertiary Education Trust Fund (TETFund) announced a special intervention to train 10,000 lecturers in digital and AI skills—a signal that the federal government sees AI as an integral part of higher education’s future.
Yet resistance persists. Many lecturers worry that AI will make it impossible to assess genuine student ability, and some fear it could ultimately threaten their jobs. These concerns are not unfounded. But the Edo trial suggests a different future: one where the lecturer’s role evolves from sole dispenser of knowledge to facilitator of critical thinking and creativity. AI can handle the routine aspects of feedback and content delivery, freeing educators to engage students in higher-order discussions, mentorship, and hands-on projects—the very things that university education was meant to foster.
The conversation is also being shaped by the global job market. Nigeria’s tech sector, which attracted over $1 billion in funding in recent years, is already demanding AI-proficient graduates. Companies like Flutterwave, Andela, and Paystack build products that leverage machine learning; they need talent that can work alongside AI rather than be replaced by it. If universities produce graduates who are only trained to operate without AI, they will find themselves unemployable in a marketplace where AI augmentation is becoming the norm.
The Edo trial’s most immediate legacy may be in secondary education, where the state government plans to expand the Copilot program to all senior secondary schools by 2026, contingent on funding. But the larger lesson for universities is clear: banning AI is a losing strategy. Students will access these tools anyway, whether through smartphones or cybercafés. The choice is between unguided, furtive use and guided, productive integration. The latter not only preserves academic integrity but also prepares students for a world where AI is omnipresent.
Looking ahead, Nigerian universities have an opportunity to leapfrog decades of pedagogical stagnation. By embedding AI into their curricula now, they can graduate students who are not just consumers of Western technology but creators of African solutions. The Copilot trial in Edo may well be remembered as the spark that lit this fuse. As one researcher put it, “We’ve proven that AI can add two years of learning in six weeks. Imagine what a four-year degree program could do if we redesign it around this potential.”
For Windows enthusiasts and education leaders alike, the implications are profound. Microsoft’s investments in AI tools like Copilot are not just about productivity in offices; they are beginning to reshape how learning happens in some of the most resource-constrained environments on earth. The next step is for policymakers, university chancellors, and lecturers to move past their fears and start building the AI-enabled classrooms that their students deserve.
The Edo trial is not the final word, but it is a powerful first chapter. As the full dataset becomes public and replication studies launch in other Nigerian states, the evidence base for AI in African education will only grow. Universities that continue to ban AI may soon find themselves on the wrong side of history, holding back the very students they aim to empower. The message from Edo is clear: don’t ban the bot—teach with it.