Southern Nigeria isn’t exactly where you’d expect a revolution in education to start. But in the bustling schools of Benin City, a new era may be dawning.

“AI helps us to learn, it can serve as a tutor, it can be anything you want it to be, depending on the prompt you write,” says Omorogbe Uyiosa, known as “Uyi” by his friends, a student from the Edo Boys High School. Uyi’s school was one of the participants in groundbreaking pilot program that integrates generative AI into after-school learning. And the results? Transformative.

The problems Nigerian schools are facing are common in many parts of the world: overcrowded classrooms, too few teachers, scarce resources for personalized education. The six-week intervention, held during the rainy season of 2024, tested generative AI as a virtual tutor for students aged 12–17 to address those issues.

The main focus of the program was improving English language skills, but the project also ventured into teaching AI knowledge and digital literacy. The results, soon to be published, exceeded expectations.

Students were split into two groups: one that continued as normal, and another one who participated in the AI program. Those who participated significantly outperformed their peers in English, AI skills, and digital literacy. More strikingly, these skills translated into better performance on end-of-year exams in subjects beyond those explicitly covered in the program.

This success suggests that students didn’t just absorb material; they learned how to learn. Perhaps even more encouragingly, everyone benefitted from the program, not just the higher-performing students. In fact, girls who were initially performing worse than boys gained more from the intervention.

The more AI lessons, the better

Attendance was a critical factor in the program’s success. Each additional session attended resulted in more improvement. Despite challenges such as flooding, teacher strikes, and after-school work commitments, students who persevered gained disproportionately more benefits from the program.

The program also revealed something important: the gains didn’t plateau. This suggests that extending such interventions could lead to even greater outcomes. The “dose-response” relationship underscores the importance of consistent, prolonged engagement for maximizing impact.

To put the results in perspective, the program achieved a 0.3 standard deviation improvement in learning outcomes — a gain equivalent to nearly two years of typical schooling in just six weeks. Compared to other educational interventions in the developing world, this ranks among the top 20% for effectiveness.

The project was spearheaded by Martín E. De Simone, an Education Specialist at the World Bank. In a recent article, De Simone and colleagues explain that the findings are in line with similar projects in other parts of the world:

“To our knowledge, this is the first study to assess the impact of generative AI as a virtual tutor in such settings, building on promising evidence from other contexts and formats; for example, on AI in coding classes, AI and learning in one school in Turkey, teaching math with AI (an example through WhatsApp in Ghana), and AI as a homework tutor.”

AI should empower learning, not be a shortcut

These global examples reinforce a key insight: AI’s adaptability makes it uniquely positioned to tackle diverse educational challenges. AI is often discussed as a problem in education as students can use it to cheat. Efforts like this show that AI can also be extremely empowering, especially in schools with fewer resources.

The success of this pilot program is a call to action for educators, policymakers, and technologists. The potential of AI in education is too great to ignore, but its implementation must be thoughtful. For now, it’s not clear whether the findings carry out in the long term and whether the efforts can be scaled.

But one thing seems clear: the journey from chalkboards to chatbots has begun. And for students like Uyi, the possibilities have never looked brighter.