The Science of Fundamental AI Research (SOFAIR) Lab will develop next generation open-source AI technologies. Modern AI relies heavily on a small number of popular architectures trained on vast data, requiring immense computing infrastructure. SOFAIR will bring together researchers from across computer science, mathematics, statistics and neuroscience to explore new AI architectures and those designed to run on widely available hardware. This will mean cutting-edge AI will be more widely accessible for everyone, including researchers and institutions.
Kanishka Narayan, the UK’s AI minister, stressed the role the lab will play in making the UK a global leader in AI-powered technologies. “AI is already doing things we could never have imagined just a few years ago, like helping to diagnose cancer,” he said. ” It can and will do even more - but if we want this technology to be a force for good, we need to make sure the next big AI breakthroughs are made in Britain.”
As part of this ambitious programme, we are building a vibrant, interdisciplinary research environment and are seeking outstanding PhD students and postdoctoral researchers to join us. Researchers will benefit from access to world-class expertise, high-performance computing infrastructure, and the opportunity to work on projects with the potential to push the boundaries of AI.
David Barber, SOFAIR Lab Director says “The SOFAIR Lab completes the missing piece in the UK AI ecosystem, bringing together the UK’s leading groups in NLP, vision modelling, probabilistic inference, agentic AI, distributed training and neuroscience with a laser focus to create and deliver next-generation foundation models. Our team from Edinburgh, Oxford, Cambridge and UCL have long been the drivers for fundamental AI training and research in the UK, and we know that students, postdocs and staff are intensely interested in foundation model research.”
Participating Universities
SOFAIR Lab unites researchers from UCL, Cambridge, Edinburgh and Oxford in a national collaborative effort to develop the next generation of open-source AI technologies. Through cross-institutional research and innovation, the lab aims to make artificial intelligence more accessible, efficient and transparent, while reducing dependence on proprietary systems and large-scale corporate computing infrastructure.
University College London (UCL): Hosts and leads the lab under Professor David Barber, focusing on overall coordination and foundational architecture.
University of Cambridge: Contributes via its Department of Computer Science and Technology (led by Professor Nicholas Lane), focusing on scalable distributed training systems.
University of Edinburgh: Involves researchers from the School of Informatics to rethink underlying AI learning paradigms. Led by Professor Amos Storkey
University of Oxford: Professor Yarin Gal and Professor Michael Wooldridge will collaborate across natural language processing, probabilistic inference, and agentic AI
