Harry is a final-year DPhil student at the OII, working in the Reasoning with Machines Lab (OxRML). His research focuses on explainability and interpretability techniques for LLMs. Specifically, he investigates whether LLMs can produce reliable natural language explanations of their own decision-making (self-explanations), a key requirement for trustworthy AI and effective human-computer interaction. He is interested in new ways to evaluate self-explanations and in developing training methods to improve them. At the centre of his work is the challenge of mitigating the risks of future AI systems.
Alongside his primary research, Harry works on capability and safety evaluations for LLMs. He is also a member of the British Open-Ended Learning and Discovery Lab (BOLD). During his DPhil, he was an Astra Fellow at Truthful AI, working with Owain Evans on discovering unexpected generalisation in LLMs.
LLMs, explainability, interpretability, LLM evaluations and benchmarks, AI safety, alignment.
22 April 2026
OII researchers and DPhil students will attend the 14th International Conference on Learning Representations in Rio de Janeiro from 23–27 April 2026.
25 November 2025
Researchers from the Oxford Internet Institute at the University of Oxford will be at NeurIPS 2025 in San Diego from 1- 7 December, 2025, contributing to one of the world’s leading AI conferences.
31 October 2025
Researchers, including those from the Oxford Internet Institute’s Reasoning with Machine’s Lab (OxRML), will attend the 2025 Conference on Empirical Methods in Natural Language Processing (EMNLP) in Suzhou, China from 4-9 November.
6 December 2024
Several researchers and DPhil students from the Oxford Internet Institute, University of Oxford, will head to Vancouver for the Thirty-Eighth annual Conference on Neural Information Processing Systems (NeurIPS) from 10-15 December 2024.
