Unconfuse Me with Bill Gates

Episode 5: Yejin Choi

Nov 16, 2023

Computer science professor and AI expert Yejin Choi discusses training language models, the challenges of robots picking tools, and the role of universities in AI research.

31:30

Creator website

Episode guests

Yejin Choi

AI Summary

Highlights

AI Chapters

Episode notes

Podcast summary created with Snipd AI

Quick takeaways

Training a large language model is a challenging task in artificial intelligence research.

The future of AI research depends on the active involvement of universities and the exploration of alternative, compute-efficient methods.

Deep dives

AI's Opaque Nature and Need for Understanding

The podcast episode explores the opaqueness of current artificial intelligence (AI) systems, which lack transparency regarding how knowledge is encoded. The guest, Dr. E. J. Genshoy, highlights the challenge of understanding why AI models perform well in some tasks while making surprising mistakes in others. This lack of understanding on both artificial and human intelligence presents new intellectual problems. Prompt engineering, where slight changes in input yield different results, is discussed as a way to improve AI performance. However, reactions to prompt engineering results vary, with some emphasizing success and others focusing more on failure cases.

The Challenges of Understanding Artificial and Human Intelligence

02:08

Moravac’s Paradox

01:01

The Importance of Making AI Models Accessible for Universities

00:46

Introduction

2min

The Opaque Nature of AI Models and Prompt Engineering

3min

Improvement in Language Models from GPT-3 to CHAT-GPT-4

9min

Building Systems from Scratch and the Possibilities of Explainable Knowledge

14min

The Role of Mathematics in Computer Science and the Importance of Specialized Data

3min

The Potential of AI to Improve Understanding and Reduce Conflict

3min

Few people are better at explaining the science of artificial intelligence than Yejin Choi. She’s a computer science professor at the University of Washington, senior resource manager at the Allen Institute for AI, and the recipient of a MacArthur Fellowship. I thought her recent TED talk was terrific, and I was thrilled to talk to her about how you train a large language model, why it’s so hard for robots to pick tools out of a box, and why universities must play a key role in the future of AI research.

Show notes:

Yejin’s TED talk: https://www.ted.com/talks/yejin_choi_why_ai_is_incredibly_smart_and_shockingly_stupid?language=en
Moravec’s paradox: https://en.wikipedia.org/wiki/Moravec%27s_paradox
Virtual Insanity by Jamiroquai: https://open.spotify.com/track/24SUWisv2lYQiB3bVpE1sn?si=b290674d2791460d

Remember Everything You Learn from Podcasts

Save insights instantly, chat with episodes, and build lasting knowledge - all powered by AI.

Unconfuse Me with Bill Gates

Episode 5: Yejin Choi

Episode guests

Podcast summary created with Snipd AI

Quick takeaways

Deep dives

AI's Opaque Nature and Need for Understanding

Scaling AI Models and Uncertainties

Concerns about Overreliance, Bias, and Misuse of AI

The Challenges of Understanding Artificial and Human Intelligence

Moravac’s Paradox

The Importance of Making AI Models Accessible for Universities

Remember Everything You Learn from Podcasts