Explore the intersection of advanced mathematics and deep neural networks in this illuminating conference talk. Delve into the application of topos theory and stack theory to understand the semantic functioning of deep neural networks. Gain insights from Daniel Bennequim of the Institut de Mathématiques de Jussieu as he presents his research via Zoom at the AI and Pure Mathematics Conference. Discover how these complex mathematical concepts can shed light on the inner workings of artificial intelligence systems and potentially advance the field of machine learning.