What We Do

Large language models are transforming social science research, but using them well requires specialised training that bridges machine learning and the social sciences. Oxford LLMs addresses this gap through three activities:

Research. We develop new methods for applying LLMs to social science problems, with a focus on synthetic data generation, opinion prediction, and model evaluation. Some of this work is ongoing; other projects are completed and are being peer-reviewed.

Teaching. We deliver courses on LLMs for social scientists at Oxford, Sciences Po Paris, and the European University Institute in Florence. All materials are open-source.

Community. Our annual summer school, now in its third year, brings together early-career researchers for an intensive week of lectures, coding, and collaborative research. Past participants have gone on to co-author publications.

Team

Oxford LLMs is led by Maksim Zubok (DPhil candidate, DPIR, Nuffield College) and Ilya Boytsov (NLP Lead, Wayfair). Rachel Bernhard (Associate Professor, Nuffield College) was co-organiser in 2025, and Humeyra Biricik (doctoral candidate, Pembroke College) in 2024.

Partners & Supporters

Oxford LLMs operates as an independent research lab. Our work has been supported by academic institutions and industry partners who share our commitment to open research and education.

Type Partners
Academic Nuffield College · University of Oxford · Oxford Van Houten Fund
Industry (2024) Ori Cloud (GPU support) · Qdrant · Arize · Google

Contact

For inquiries: maksim.zubok@nuffield.ox.ac.uk