Meaning in Language
Raphaël discusses the relationship between form and meaning in language, highlighting the limitations of correspondence and the need for a historical learning process to fix representational content. He explores how large language models acquire meaningful representations through data learning and RLHF fine tuning, which embeds human preferences about truthfulness and accuracy. This process provides the missing ingredient for securing world involving lexical representational content.In this clip
From this podcast
The Gradient
Raphaël Millière: The Vector Grounding Problem and Self-Consciousness
Related Questions