GAZELL.iO is thrilled to showcase Maximum Activations, the latest series by our December 2024 artist-in-residence, Tom White. The latest results in AI mechanistic interpretability research reveal intricate proto-concepts hiding inside Large Language Models like ChatGPT. The series "Maximum Activations" reveals this hidden structure as networked collections of activating contexts. Separated conceptual building blocks such as "supplies taken on journeys" and "references to social media and digital culture (learned when trained to be a friendly chatbot)" are presented as visual collages of relevant computer memories.