Clearer Thinking with Spencer Greenberg

Taking pleasure in being wrong (with Buck Shlegeris)

Jun 8, 2022

01:16:10

Creator website

Highlights

AI Chapters

Episode notes

How to Get Fast at Small Skills

01:20

The Core Problem With Machine Learning

01:20

Introduction

2min

The Importance of Beliefs About the World

2min

The Importance of Argumentation in AI Arguments

2min

The Importance of Being Coherent

2min

How to Maintain Humility in Conversations

2min

The Importance of Pain Tolerance

3min

How to Write a Good Essay

2min

How to Validate a Theory

2min

The Importance of Factual Knowledge in Behavioral Economics

3min

10.

The Importance of Depth of Knowledge in Management

2min

11.

How to Get Feedback on Hard Cases

2min

12.

How to Learn Quickly and Effectively

4min

13.

The Importance of Asking the Simplest Questions

2min

14.

The Power Cord and the Air Conditioner

2min

15.

The Complexity of Simple Questions in Math

4min

16.

The Importance of a Decision Procedure in Physics

2min

17.

The Holding Problem and Math

3min

18.

The Importance of Having a Dry Diagram

2min

19.

The Importance of Aligning and Artificial Intelligence

2min

20.

The Importance of Using Futurism in AI Alignment Research

2min

21.

How to Make Neural Nets Suspect of Text

2min

22.

The Problem With Adversaries in Neural Nets

2min

23.

The No Free Lunch Theorem

2min

24.

The Core Problem With Machine Learning

3min

25.

How to Set Hyper Parameters Differently in Machine Learning

2min

26.

The Evolution of Machine Learning

2min

27.

The Unifying Themes of Machine Learning

3min

28.

How to Reduce Variance in Machine Learning

3min

29.

How to Reduce the Variance in the Learning Process

2min

30.

How to Reduce Existential Risk From AI Alignment

2min

31.

How to Scale Up Your Research on the Eve of Deploying AI

2min

32.

The Importance of Early Capacity Building

2min

33.

How to Pick the Type of Research You're Doing

2min

34.

How to Make Intuition Flooding Predictions

2min

Read the full transcript here.

How hard is it to arrive at true beliefs about the world? How can you find enjoyment in being wrong? When presenting claims that will be scrutinized by others, is it better to hedge and pad the claims in lots of caveats and uncertainty, or to strive for a tone that matches (or perhaps even exaggerates) the intensity with which you hold your beliefs? Why should you maybe focus on drilling small skills when learning a new skill set? What counts as a "simple" question? How can you tell when you actually understand something and when you don't? What is "cargo culting"? Which features of AI are likely in the future to become existential threats? What are the hardest parts of AI research? What skills will we probably really wish we had on the eve of deploying superintelligent AIs?

Buck Shlegeris is the CTO of Redwood Research, an independent AI alignment research organization. He currently leads their interpretability research. He previously worked on research and outreach at the Machine Intelligence Research Institute. His website is shlegeris.com.

Staff