My research combines themes from artificial intelligence, linguistics, and philosophy. My central area of focus is AI interpretability: how can the internal workings of models be explained in a way that humans can understand? In the AIdemoc project, I also focus on the challenges posed by AI model hallucinations.