By Enock Sithole
A new educational toolkit developed by University College London (UCL) and the African Institute of Mathematical Sciences (AIMS) is seeking to expand access to advanced artificial intelligence education across Africa while placing ethics, governance and social responsibility at the centre of AI development.
The AI Research Foundations Toolkit is designed to complement Google DeepMind’s free online curriculum on building and fine-tuning large language models (LLMs). The new resources transform the online material into a university-level blended learning module that lecturers can adapt to the circumstances of their own institutions.
Developed with support from Google.org, the toolkit was co-designed through workshops involving educators, students and researchers from several African countries, including Mali, Zambia, South Africa, Kenya, Somalia, Nigeria, Ghana and Namibia.
The initiative comes amid a significant imbalance in global AI research. According to the project partners, Africa accounts for roughly 20% of the world’s population but produces less than 2% of global AI research.
The project is intended to help address that gap by giving African academics access to teaching materials that can be adapted and delivered locally, rather than relying exclusively on imported approaches to AI education.
The toolkit builds on the AI Research Foundations programme, a collection of free online courses developed by Google DeepMind that covers generative AI and large language models.
Rather than focusing only on the technical capabilities of AI, however, the new educational resources combine machine learning instruction with questions about ethics, governance, environmental justice and the wider social consequences of artificial intelligence.
Professor Ulrich Paquet, project sponsor and AI Director at AIMS, said teaching AI required an interdisciplinary approach because the subject spans technical disciplines and questions of social responsibility.
“Teaching AI is a wonderful challenge, as it is truly interdisciplinary,” Paquet said. “It stretches from mathematical rigor to social responsibility, and from technical details to applied solutions.”
He said the collaboration with UCL had produced a toolkit designed to give educators flexibility while enabling students to engage with AI research in their own cultural and educational contexts.
The toolkit contains more than 50 hours of adaptable classroom resources, including lesson plans, coding notebooks, lecture slides and locally relevant case studies.
It also uses a blended-learning model that connects Google DeepMind’s online courses with classroom activities such as group discussion, collaboration and problem-solving.
The developers say the resources were structured around feedback from African university teachers and students, with particular attention to differences in class sizes, internet bandwidth and physical teaching environments.
For Matthew Seren Smith, project director at UCL Engineering Sciences, incorporating African contexts was central to making the resources accessible.
“By grounding the toolkit in local contexts through our collaboration with AIMS, we decentralized it,” Smith said.
He said regional case studies were not intended to narrow the application of the materials but to provide a foundation through which educators could introduce their own languages, cultures and local challenges.
Michael Alummoottil, project manager at AIMS, said in a statement that the collaboration had sought to build on forms of community responsibility already present in many African societies.
“African students often arrive with a deep, intuitive understanding of responsibility, growing up in societies where community values dominate,” he said.
According to Alummoottil, the involvement of specialists from across the AIMS network helped combine technical expertise with an understanding of the practical realities facing African STEM classrooms.
The responsible-AI component is a defining feature of the toolkit. Technical concepts in machine learning are taught alongside material from sociology, philosophy and ethics, with the aim of helping students understand not only how AI systems work but also how they affect societies.
Professor John Mitchell, project sponsor at UCL Electronic and Electrical Engineering, said the collaboration demonstrated the value of combining technical expertise with local knowledge and responsible-innovation principles.
“This AI Research Foundations Toolkit isn’t just about advancing technology; it’s about empowering the next generation of researchers to build AI safely, ethically, and with deep community context,” Mitchell said.
Other UCL academics involved in the project also highlighted the importance of flexibility.
Dr Stephen Hughes of UCL Science and Technology Studies added that the discussions with African educators, researchers and students showed that many already had strong community-centred perspectives on technological responsibility.
Professor Benjamin Guedj of UCL Computer Science said machine learning was rapidly becoming a foundational skill across science, engineering and other disciplines. He said teaching materials therefore needed to be technically rigorous while remaining accessible and practical for classroom use.
Professor Eileen Kennedy of the UCL Institute of Education said the co-design workshops were intended to ensure that the toolkit responded to what educators actually needed.
She said teachers did not want a rigid textbook imposed on them but wanted “dynamic, customizable materials” that would allow them to develop their own lessons and incorporate local cultural realities.
The project brings together academics from several parts of UCL, including Engineering, Computer Science, the Institute of Education, Science and Technology Studies and UCL Consultants.
The toolkit is freely available to educators worldwide and will be distributed through AIMS’s academic networks.
For Africa, where universities and educators face differing levels of infrastructure, resources and connectivity, the emphasis on adaptability could be important. Rather than prescribing a single model of AI education, the initiative seeks to give educators a foundation that can be modified to suit local conditions.
The broader ambition is to increase African participation not only as consumers of artificial intelligence but as researchers, developers, educators and contributors to debates about how the technology should be governed.
The project partners say that by combining technical AI education with ethics, governance and local knowledge, the toolkit can help students develop the skills needed to understand, question and shape rapidly evolving AI technologies.
AIMS, whose centres are located in South Africa, Senegal, Cameroon, Ghana and Rwanda, describes its mission as advancing mathematical sciences through education, research and public engagement across Africa.
The AI Research Foundations Toolkit adds another dimension to that mission by seeking to expand access to advanced AI education while encouraging a generation of African researchers to approach artificial intelligence as both a technical field and a societal responsibility.