Published Mar 13, 2023

#107 - Dr. RAPHAËL MILLIÈRE - Linguistics, Theory of Mind, Grounding

Dr. Raphaël Millière delves into the potential and limitations of AI in understanding human cognition, examining theory of mind, symbol grounding, and semantic competence, while contemplating the ethical implications of AI advancements and the challenges of achieving human-like language capabilities.

Episode Highlights

Topics covered

Theory of Mind
01
Symbol Grounding
02
Semantic Competence
03
Compositionality in AI
04

#107 - Dr. RAPHAËL MILLIÈRE - Linguistics, Theory of Mind, Grounding

Topics covered

Popular Clips

Color Space and Language Models

Simulating Agents

Neural Networks Compositionality Debate

Uncovering Model Competence

Neural Networks and Cognition

Language Model Capacities

Symbol Grounding Discussion

Intrinsic Grounding Debate

Novel Conceptual Combinations

Language Model Generalization

Unraveling Language Models

Language Models' Grounding

Theory of Mind

Language Models and Real World Representation

Episode Highlights

Theory of Mind

Symbol Grounding

Semantic Competence

Compositionality in AI

#107 - Dr. RAPHAËL MILLIÈRE - Linguistics, Theory of Mind, Grounding

Topics covered

Popular Clips

Episode Highlights

Theory of MindDr. Raphaël Millière examines the potential for AI, particularly language models, to develop a Theory of Mind. He questions whether these models can truly understand human cognition or if they merely mimic it, highlighting the ethical implications of such advancements.

Theory of Mind

Symbol GroundingThe discussion explores the complexities of grounding in AI, focusing on referential and sensory-motor grounding. Raphaël Millière examines how these concepts apply to AI models and the challenges they face in achieving meaningful grounding.

Symbol Grounding

Semantic Competence

Compositionality in AIThe challenges of compositionality in AI models and their ability to generate novel conceptual combinations are explored by Raphaël Millière. He examines how recent advancements in deep learning challenge traditional views on AI's capacity for human-like cognition.

Compositionality in AI