About me

Hi, I’m Guillaume.

Currently, I am a researcher interested in applying multivariate information theory to deep neural networks for AI interpretability.

Previously, I was a research affiliate at the Principle of Intelligence and Behavior in Biological and Social systems (PIBBSS), where I worked on singular learning theory and explored the application of information theory for lie detection in LLMs.

I have also conducted AI strategy research in collaboration with the Center on Long Term Risk.

My PhD thesis, completed under the supervision of Lionel Barnett and Anil Seth at the University of Sussex, focused on estimating information flow between different brain regions during visual perception.

I enjoy teaching and have worked as a teaching assistant at the african institute for mathematical sciences (AIMS) in Sénégal for ten months.

Outside of my areas of expertise, I have a keen interest in the fields of moral and political philosophy, as well as economics.