The objective of this paper is to co-develop a ‘how-to’ pipeline for protecting and championing responsible AI among Sub-Saharan African online learners: from the conception of the idea for an online learning tool, through the iterative design and development, until the moment of scale. We address significant gaps with our research. In contrast to frameworks emerging from tech giants and leading thinkers in digital technology with responsible AI (e.g., Facebook, 2021; Google, 2023), our work specifies step-by-step guidance in the form of a pipeline to directly address design, delivery, and decision-making (de Laat, 2021). In contrast to existing AIED guidance (Khosravi et al., 2022) and communities (Holmes et al., 2022), our project goes beyond the crude distinction between the Western and non-Western boundaries to focus on one specific non-Western region: namely, Sub-Saharan Africa, with representation from three Sub-Saharan African countries within the research group itself—Kenya, the Democratic Republic of Congo (DRC), and South Africa. Frameworks will be kept in mind that already exist for sustainable integration of educational technology in Sub-Saharan African contexts (Mabila, van Biljon, et al. 2017). Principles for maturity model development will be garnered and integrated into the pipeline development, as a framework for evaluating the ‘capabilities’ required of online learning with responsible AI (Alsheiabni et al., 2019; Fukas et al., 2021). We are conducting a comprehensive and systematic programme of consultation and synergy across the edtech ecosystem in Sub-Saharan Africa, with methodological elements taken from public participation and the Grounded Delphi Method (e.g., Howard, 2018), as appropriate. Discussions started within the research group as a collaborative partnership of invited experts. Consultations have then extended to member groups in the public: learners, educators, community members, non-profits, policymakers, and academics from outside our international research partnership. As the systematic consultations have just concluded, we already find emphatic consensus around key areas for the pipeline for responsible AI in online learning in Sub-Saharan Africa. We find that equitable online learning design and delivery must be people-centred with affected actors at the centre: this entails a plurality of voices for strengthened representation to redress historical inequities. Deployment of AI must be ethical, with AI solutions being culturally responsive and adaptive to local requirements. Throughout, infrastructural obstacles must be kept in sight, and capacity building unforgotten. Meanwhile, adaptive learning systems are starting to make offerings available that are genuinely affordable and sustainable.

Alsheiabni, S., Cheung, Y., & Messom, C. (2019). Towards An Artificial Intelligence Maturity Model: From Science Fiction To Business Facts. de Laat, P. B. (2021). Companies Committed to Responsible AI: From Principles towards Implementation and Regulation? Philosophy & Technology, 34(4), 1135–1193. Facebook. (2021, June 22). Facebook’s five pillars of Responsible AI. Fukas, P., Rebstadt, J., Remark, F., & Thomas, O. (2021). Developing an Artificial Intelligence Maturity Model for Auditing. ECIS. Google. (2023). Google AI Principles. Google AI. Holmes, W., Porayska-Pomsta, K., Holstein, K., et al. (2022). Ethics of AI in Education: Towards a Community-Wide Framework. International Journal of Artificial Intelligence in Education, 32(3), 504–526. Howard, K. J. (2018). Emergence of a new method: The Grounded Delphi method. Library and Information Research, 42(126), 5–31. Khosravi, H., Shum, S. B., Chen, G., et al. (2022). Explainable Artificial Intelligence in education. Computers and Education: Artificial Intelligence, 100074. Mabila, J., Van Biljon, J., & Herselman, M. (2017). A sustainability framework for mobile technology integration in schools: The case of resource-constrained environments in South Africa. The Journal of Community Informatics, 13(2).