My research evaluates LLM self-explanations: the natural language explanations LLMs give to justify their own decision-making. I measure whether these explanations are faithful to models' true internal reasoning, and develop new training incentives to improve faithfulness. I'm motivated by AI safety, and also work on LLM evals and the science of evals, including the LingOly and LingOly-TOO reasoning benchmarks. My work has been published at leading venues including NeurIPS, ICML, and ICLR. I'm supervised by Prof. Adam Mahdi and Prof. Jakob Foerster.

I mainly write about AI safety, explainability, and evals.

I'm a final year PhD researcher. I've had an unusual path, having originally studied economics.

I've held several teaching positions including TA-ing the Social Data Science MSc at Oxford and tutoring Stanford computer science students. Previous students have gone on to the CS Masters at Stanford and various PhD positions at Oxford.