Linus Lee: At the Boundary of Machine and Mind

Topics covered
Popular Clips
Episode Highlights
Language Structures
explores the constraints and creativity inherent in human languages. He notes that our vocal systems limit the sounds we can produce, yet within these constraints, languages are remarkably diverse and arbitrary 1. For instance, Korean allows for the omission of certain sentence parts if implied, a concept that feels natural to native speakers but can be challenging for learners 2. This flexibility in language structure is mirrored in the way different modalities, like images and text, can map onto each other, suggesting a shared cognitive framework 3.
Humans are really creative with sounds. There's different kinds of sounds that languages are based on.
---
Constructed Languages
Constructed languages offer a unique lens into the creativity and complexity of linguistic design. Linus shares his experience in creating his own language, highlighting the intricate process of developing grammar and vocabulary from scratch 4. This endeavor parallels the creation of notations, where personal systems of symbols and expressions can enhance understanding and communication 5. The challenge lies in balancing personal intuition with the need for a standardized system that others can understand.
You have to build your own version of a thing to really understand how it works.
---
Expanding Understanding
Linus envisions a future where language models enhance our understanding of language itself. He suggests that these models could provide a more embodied representation of language, allowing us to interact with ideas as tangible objects 6. This concept extends to imagining different ways of experiencing the world, such as through the lens of probability distributions or varying timescales 7. By leveraging machine intelligence, Linus aims to expand the boundaries of human thought and creativity 8.
How do we use computer models of language to improve language itself?
---
Related Episodes


Tal Linzen: Psycholinguistics and Language Modeling
Answers 383 questions

Kyunghyun Cho: Neural Machine Translation, Language, and Doing Good Science
Answers 383 questions

Peter Lee: Computing Theory and Practice, and GPT-4's Impact
Answers 383 questions

Been Kim: Interpretable Machine Learning
Answers 383 questions

Chip Huyen: Machine Learning Tools and Systems
Answers 383 questions

Manuel & Lenore Blum: The Conscious Turing Machine
Answers 383 questions

Christopher Manning: Linguistics and the Development of NLP
Answers 383 questions

Daniel Situnayake: AI on the Edge
Answers 383 questions

Ted Underwood: Machine Learning and the Literary Imagination
Answers 383 questions

Michael Levin & Adam Goldstein: Intelligence and its Many Scales
Answers 383 questions

Jacob Andreas: Language, Grounding, and World Models
Answers 383 questions

Cameron Jones & Sean Trott: Understanding, Grounding, and Reference in LLMs
Answers 383 questions

Eric Jang on Robots Learning at Google and Generalization via Language
Answers 383 questions

Zachary Lipton: Where Machine Learning Falls Short
Answers 383 questions

Terry Winograd: AI, HCI, Language, and Cognition
Answers 383 questions
