UCL and AIMS Release Free Toolkit for Teaching AI Foundations

University College London (UCL) and the African Institute for Mathematical Sciences (AIMS) have released a free teaching toolkit for introducing artificial intelligence at the university level. The AI Research Foundations Toolkit, funded by Google.org, is designed to help instructors turn Google DeepMind's self-paced online course into a full academic class.

The AI Research Foundations course itself was launched earlier by UCL and Google DeepMind as an open learning path for students and self-learners. It consists of eight modules, including a six-hour introduction to language models, a unit on data representation, a course on transformer architecture, an eight-hour model fine-tuning module, and a fifteen-hour capstone project. The new toolkit doesn't replace the course, it adds the layer needed to run it as a structured university class.

What the toolkit includes

The creators stress that the materials are modular, letting instructors pick and adapt individual pieces rather than adopting the whole program unchanged. The toolkit covers technical topics such as language models, text data preparation, neural networks, transformers, model fine-tuning and computing hardware, but weaves in ethics, sociology and philosophy instead of treating them as a separate, bolted-on module.

That design is meant to prevent responsible AI development from becoming an afterthought tacked onto technical coursework rather than an integral part of it. The toolkit's authors describe the approach as a response to growing demand from universities for materials that combine engineering skills with an understanding of the technology's social impact.

The role of African educators

The toolkit was built in partnership with AIMS, a network of mathematical science institutes operating across several African countries, with African university educators helping co-create the lesson plans. The earlier version of the AI Research Foundations program, launched by UCL and Google DeepMind in October 2025, was built with a similar emphasis on accessibility for students outside traditional AI research hubs, with particular attention to Africa.

Those involved in developing the program include professor John Mitchell, co-director of the UCL Centre for Engineering Education, and Ulrich Paquet, a Google DeepMind scientist and director of AIMS South Africa, who co-founded the Deep Learning Indaba initiative supporting AI research across the African continent.

This project reflects UCL's commitment to widening access to world-class AI research and innovation - prof. John Mitchell, co-director of the UCL Centre for Engineering Education

Why it matters for universities

Free, ready-to-use teaching materials lower the barrier for universities that want to introduce AI foundations courses but lack the resources to build a program from scratch. The modular structure lets institutions tailor the scope to local conditions, from a single-semester introduction to a deeper program for technical students.

The program targets students from early undergraduate years through early-career researchers in mathematics, computer science, physics and engineering, assuming familiarity with Python. Beyond technical foundations, the earlier version of the course also included practical applications, including projects on climate resilience and disease research.

For Polish universities, which have spent months grappling with the lack of a coherent AI teaching offering, free resources like this could be a starting point for launching courses faster without building a program from the ground up. That still requires translating and adapting the materials to the local context, something the toolkit does not provide on its own.