Hear This Idea

#63 – Ben Garfinkel on AI Governance

98 snips
May 13, 2023
Ask episode
Chapters
Transcript
Episode notes
1
Introduction
00:00 • 2min
2
The Impacts of Artificial Intelligence on the World
01:42 • 2min
3
The Potential Impacts of AI on Human Well-Being
03:36 • 5min
4
The Role of Technology in Conflict
08:20 • 4min
5
The Effects of the Industrial Revolution on the Nature of Productivity
12:25 • 5min
6
The Importance of International Coordination
17:20 • 2min
7
The Risks of Out of Control AI Systems
19:41 • 5min
8
The Story of Contingency in AI
24:29 • 2min
9
The Contingency Story of the Cold War
26:29 • 3min
10
The Noise of Nuclear War in the 1950s
29:42 • 3min
11
The Tradeoff Between Commercial Pressure and Safety
33:08 • 6min
12
The Probability of a Global Catastrophe
38:46 • 3min
13
The Differences Between International Competitive Dynamics and Domestic Relations
41:40 • 5min
14
The Problem With AI Governance
46:11 • 2min
15
The Future of AI Governance
47:49 • 4min
16
The Differences Between the Harms and the Benefits of AI
51:45 • 2min
17
The Transition to Advanced AI
53:23 • 3min
18
The Differences Between People Working in Governments Motivated by Long-Term Concerns
56:05 • 2min
19
The Importance of Paying Attention to Long-Term Impacts of AI
58:17 • 4min
20
The Perception of Unsafe AI Systems
01:01:52 • 3min
21
The Differentiation Effects of AI Views
01:05:00 • 3min
22
How Financial Markets Respond to AGI Catastrophes
01:07:35 • 2min
23
The Selection Effect of Financial Markets
01:09:28 • 3min
24
The Range of Views on Financial Markets
01:11:58 • 2min
25
How Much Weight Should You Give to Observations About Present Day Air Systems?
01:14:14 • 3min
26
The Importance of Skepticism in AI Arguments
01:17:31 • 3min
27
The Probability of Different Harms Caused by Unsafe AI Systems
01:20:11 • 3min
28
The Multiple Equilibriums of Frontier Systems
01:22:44 • 3min
29
The Mechanistic Story of Risk
01:25:58 • 5min
30
The Winner's Curse of Randomness
01:31:00 • 2min
31
The Propensity for Catastrophes in AI Systems
01:33:10 • 2min
32
The Argument for a Single AI System
01:35:29 • 5min
33
The Importance of a Skeptical Prior
01:40:04 • 3min
34
The Discontinuity of AI Development Processes
01:43:00 • 2min
35
The Orthogonality Thesis and the Argument for Instrumental Convergence
01:44:36 • 2min
36
The Abstract Instrumental Convergence Argument for Machine Learning
01:46:30 • 4min
37
The Limits of Artificial Intelligence Progress
01:50:11 • 3min
38
Doom Is High in Hindsight
01:52:51 • 3min
39
The Future of AI
01:55:23 • 3min
40
The Pros and Cons of Negative Feedback From AI Systems
01:58:34 • 3min
41
How to Handle a Situation Where You Have Deceptively Misaligned Systems
02:01:06 • 3min
42
Human Evolution History and the Septic Misalignment
02:03:37 • 5min
43
The Bi-Nology of AI Systems
02:08:29 • 4min
44
The Importance of Perception in AI Systems
02:12:01 • 3min
45
Evolutionary History and the Preferences People Have Today
02:14:58 • 4min
46
The Evolution of Large Language Models
02:18:49 • 3min
47
The Strange Position of the Brother World in the ML Space
02:21:24 • 3min
48
The Consequences of Estimating the Chance of Human Extinction
02:24:20 • 3min
49
The Future of AI Governance
02:27:02 • 6min
50
The Role of AGI Systems in Legislation
02:32:43 • 2min
51
The Importance of a Basket of Skills
02:34:27 • 5min
52
How to Be a Successful Biosecurity Policy Expert
02:39:38 • 5min
53
The Frontier Isn't Moved Very Far Forward
02:44:42 • 2min
54
The Limits of Risks From Advanced AI Systems
02:46:32 • 2min
55
Gavey I: A Venture Fellow's Perspective on the Future of Training
02:48:28 • 2min
56
The Future of AI Governance
02:50:14 • 4min
57
How to Find You in Gaveye Online
02:54:07 • 2min
58
Ben Garfinkel on AI Governance
02:56:08 • 2min