Edmund Mills

As an intern at CHAI, Edmund is working with Scott Emmons on a benchmark for methods for understanding large language model (LLM) behavior. Numerous methods, including those in the areas of mechanistic interpretability and explainable AI, have the goal of understanding and explaining the behavior of neural models; this work aims to create a standard for their evaluation and comparison. Edmund’s research interests include the emergent capabilities of LLMs, including the dynamics of chain of thought reasoning and their behavior as agents. He earned his Ph.D. in Materials Science and Engineering from the University of California, Davis in 2017, where his research focused on the electrochemical properties of solid interfaces in fuel cell electrolyte materials.