While AI in education is rapidly developing into a potential global equalizer, offering individual learning support and access to high-quality learning resources, there are large disparities in digitalization between countries and within countries. In low- and middle-income countries in particular, there is a lack of adequate digital infrastructure and Internet access, while at the same time there is a severe shortage of teachers. As a result, the potential of AI in education is being held back by a variety of practical constraints that are especially pronounced in resource-poor environments. In this paper, we examine the potential of AI-powered educational technologies to narrow the global digital divide. We present a range of affordable educational technologies that are designed to work offline. Examples include computer laboratories based on Raspberry Pi devices, portable educational servers, and edge AI systems. To help reduce the global digital divide and enable affordable AI education in regions with limited internet and a lack of power, edge computing with AI can be deployed. Using compact AI models running on Raspberry Pi’s, this enables offline tutoring, coding assistance, personal learning, and multilingual support. Teacher training, a locally adapted curriculum, and low power/solar-powered infrastructure complete the sustainable solution to increase digital skills and educational opportunities in underserved communities. This framework of international cooperation can enhance online literacy and offer education to countries in need of development by reducing the gap in education and pursuing sustainable technological development.

References

[1] International Telecommunication Union, Facts and Figures 2024, Geneva, Switzerland: ITU, 2024.

[2] World Bank, Digital Progress and Trends Report 2023, Washington, DC, USA: World Bank, 2024.

[3] F. Miao and W. Holmes, Guidance for Generative AI in Education and Research. Paris, France: UNESCO, 2023.

[4] UNESCO, AI Competency Framework for Teachers. Paris, France: UNESCO, 2024.

[5] UNESCO, AI Competency Framework for Students. Paris, France: UNESCO, 2024.

[6] UNICEF, “Offline learning,” The Learning Passport, UNICEF.

[7] Raspberry Pi Ltd., “Raspberry Pi 5,” Raspberry Pi Documentation, Cambridge, U.K.

[8] UNICEF, Pathways: Digital Competency Framework for Educators in Sub-Saharan Africa. New York, NY, USA: UNICEF, 2026.

[9] Gemma Team et al., “Gemma: Open models based on Gemini research and technology,” arXiv preprint arXiv:2403.08295, 2024.

[10] B. Hui et al., “Qwen2.5-Coder technical report,” arXiv preprint arXiv:2409.12186, 2024.

[11] Microsoft Research, “Phi-4 technical report,” Microsoft, 2024.

[12] U. Kurt, “Which quantization should I use? A unified evaluation of llama.cpp quantization on Llama-3.1-8B-Instruct,” arXiv preprint, 2026

[13] Learning Equality, “About Kolibri,” Kolibri, Learning Equality. [Online]. Available: https://learningequality.org/kolibri/about-kolibri/

[14] Raspberry Pi Foundation, “Raspberry Pi reaches more schools in rural Togo,” Raspberry Pi News, Sep. 30, 2020. [Online]. Available: https://www.raspberrypi.com/news/raspberry-pi-reaches-more-schools-in-rural-togo/

[15] Microsoft, “Welcome to the new Phi-4 models—Microsoft Phi-4-mini & Phi-4-multimodal,” Microsoft Tech Community, Feb. 26, 2025. [Online]. Available: https://techcommunity.microsoft.com/blog/educatordeveloperblog/welcome-to-the-new-phi-4-models---microsoft-phi-4-mini--phi-4-multimodal/4386037

[16] Qwen Team, “Qwen2.5-Coder Series: Powerful, diverse, practical,” Nov. 12, 2024. [Online]. Available: https://qwenlm.github.io/blog/qwen2.5-coder-family/

[17] Microsoft Azure, “Empowering innovation: The next generation of the Phi family,” Feb. 26, 2025. [Online]. Available: https://azure.microsoft.com/en-us/blog/empowering-innovation-the-next-generation-of-the-phi-family/