When Words Have No Root: A Conversation with Lisa Meeden

Large Language Models have a problem: The Symbol Grounding Problem. Even though they can write full essays and have conversations, they can’t communicate. The words that they produce have no root.

But they aren’t missing etymological roots - they are missing existential roots. If language exists as a reflection of experience, what happens when there is no experience to reflect? What are words when they reflect nothing?

These questions are the base of today’s episode of #MergingMinds, where Professor Lisa Meeden speaks with our host Gabriel Fairman. This episode is deeply philosophical, lightly technical, and a whole lot of fun.

They discuss:

- How the syntactical capabilities of #LLMs emulate semantic understanding

- What happens when words are only connected to each other

- The challenge of truth in Large Language Models

- Golfers, sheep and the challenge of ambiguity

- “Synthetic Texts” and “Stochastic Parrots”

- Grading students in the age of #ChatGPT

- Robotics as the path to general #AI

- The Symbol Grounding Problem

- “Nutrition labels” for data sets

- Constraints and lateral moves

- Watermarks for AI creations

- AI Springs and Winters

And more!

Click play to hear Lisa and Gabriel discuss!

Lisa Meeden is a professor of Computer Science at Swarthmore College who also participates in the interdisciplinary program of Cognitive Science. Her research focuses on developmental robotics and, she is part of a team that developed AITK: An artificial intelligence toolkit that includes open-source Python tools and computational essays for exploring and understanding Artificial Intelligence, Cognitive Science, Machine Learning, and Robotics.

You can learn more about Lisa Meeden's work here:

The syllabus for the new responsible AI course is available here:

https://www.cs.swarthmore.edu/~meeden/phil27/s24/

AITK is freely available via github:

https://github.com/ArtificialIntelligenceToolkit/aitk

A description of the #ResponsibleAI curriculum development project is here:

https://nationalhumanitiescenter.org/education-programs/faculty-programs/responsible-artificial-intelligence-undergraduate-curriculum-design-project/

Don't forget to leave a review and subscribe to Merging Minds Podcast powered by Bureau Works for more thought-provoking podcast episodes.

‍