Erlangen AI Hub: A year of momentum at the intersection of mathematics and AI

Launched in 2024 with EPSRC support, the Erlangen AI Hub has rapidly become a major force in UK artificial intelligence research. Inspired by Felix Klein’s Erlangen Programme, the hub places mathematics at the centre of AI, bringing together geometry, topology, and computation to shape the foundations of trustworthy and effective AI systems.

Now entering its second year, the hub continues to grow in reach and ambition. Originally conceived by Professor Michael Bronstein and based at the department, the hub has expanded to include co-directors Dr Anthea Monod (Imperial) and Professor Jeffery Giansiracusa (Durham). Between them they have supervised more than 200 PhDs, secured over £100 million in research funding, launched 17 tech spinouts, and received four Whitehead, two Adams, and three Leverhulme prizes.

The hub spans institutions across the UK, with strong links to industry. Collaborations in its first year include work with Microsoft Research using topology to study the shape of LLM activations under adversarial influences, and with Oxford Drug Design to enhance drug discovery pipelines using geometric data analysis. The hub’s first year also established major inroads into government, seeing the delivery of an AI training programme to civil servants including those at DSIT and the AI Security Institute, bringing in an additional £1.2 million in funding and three new postdoc positions at the Imperial Policy Forum.

With core research organised around four themes - Understanding Data, Understanding ML Models, Understanding Learning, and Understanding Decision-Making - the hub’s postdoctoral researchers and PhD students are already presenting their work at major AI conferences such as NeurIPS, ICML, and ICLR. In May, Oxford hosted a hub-wide gathering for early-career researchers, followed by the hub’s inaugural public conference at Queen Mary University of London, which brought together over 100 researchers, policy makers and industry representatives.

The hub also continues to strengthen its public and policy-facing work. A new taxonomy for AI technologies, led by Professor Peter Grindrod, is in development to help demystify AI for non-experts and support better decision-making across sectors.

As its work continues, the Erlangen AI Hub is consolidating its role in the UK’s AI landscape. Find out more about the hub's work at https://erlangenhub.ox.ac.uk/.