Moving the Eiffel Tower: Opening LLMs with Explainable AI, Jelle Zuidema, UvA

Deep Dives with Iman, hosted by Iman Mossavat and featured by Radio 4 Brainport, features Associate Professor Jelle Zuidema from the University of Amsterdam's Institute for Logic, Language, and Computation (ILLC).Zuidema analyzes large language models (LLMs) using explainable AI, including striking examples to show how model knowledge can be surgically edited. He briefly discusses how researchers understand LLMs, fix biases, and analyze how they reason. Listeners gain insight into how these models represent knowledge, interact with data, and how analysis can clarify complex AI systems. This episode gives you an excellent view into the inner workings of modern language models.