Large language models (LLMs) such as Claude and Gemini have reached a high level of language processing. Based on linguistic input, they produce lines of reasoning, make summaries, and analyse arguments. For many people, AI chatbots are a daily conversational partner. LLMs thus process language and produce linguistic output, but do these systems actually understand what they 'read and say'?

According to linguist Emily Bender, LLMs are ultimately nothing more than 'stochastic parrots' that reproduce statistical patterns in language without access to meaning. In contrast, researchers such as Geoffrey Hinton, one of the founders of modern AI, believe that these models understand language like humans, and even attribute a form of consciousness to them. This workshop brings together researchers from cognitive psychology, linguistics, philosophy, and AI to investigate the conceptual and empirical foundations of this discussion. What do we mean when we talk about understanding and meaning? 

Do LLMs really understand language? By juxtaposing different theoretical perspectives and empirical findings, this workshop aims to create an interdisciplinary basis to answer these fundamental questions of understanding, meaning, and representation in computational models.

This is an invitational event.

Programme