The 80000 Hours Podcast on Artificial Intelligence

Six: Richard Ngo on large language models, OpenAI, and striving to make the future go well

Sep 2, 2023
This podcast explores the understanding and potential risks of large language models like GPT-3 and ChatGPT. Richard Ngo from OpenAI discusses AI governance, concerns surrounding these models, and the challenges of AI behavior prediction. They also delve into the development of general AI, situational awareness in AI systems, and the need to study and modify goal formation in neural networks. The podcast concludes with discussions on the challenges of understanding AI behaviors, exploring utopia and the role of technology, and alternative history thought experiments.
Ask episode
Chapters
Transcript
Episode notes
1
Introduction
00:00 • 4min
2
AI Governance and the Alignment Problem
03:41 • 9min
3
Comparing Governance Arrangements for Nuclear Weapons and Computer Chips
12:20 • 2min
4
Concerns Surrounding Large Language Models
13:58 • 18min
5
Concerns and Misconceptions about AI Behavior
31:35 • 3min
6
OpenAI's Work and Goals
34:41 • 18min
7
Exploring the Vast Space of Language Model Prompts
53:08 • 16min
8
Misconceptions and Insights into Machine Learning Models
01:09:17 • 4min
9
Development of General AI and Lab Priorities
01:13:18 • 13min
10
Situational Awareness in AI
01:26:02 • 10min
11
Understanding and Modifying Goal Formation in Neural Networks
01:35:59 • 14min
12
The Challenges of Understanding and Predicting AI Behaviors
01:49:31 • 23min
13
Exploring Conceptual Alignment Research on Defining Goals in Neural Networks
02:12:29 • 2min
14
Discussing Existing Books on the Alignment Problem
02:14:08 • 4min
15
Taking Action and Transitioning to Producing Knowledge
02:17:43 • 11min
16
Exploring Utopia and the Role of Technology
02:28:37 • 9min
17
Alternative History Thought Experiments and Contingencies in Evolution
02:37:22 • 2min
18
Science Fiction Books, Thought Experiments, and AGI Safety Fundamentals
02:39:28 • 5min