Making Sense with Sam Harris - Subscriber Content

#312 - The Trouble with AI

Mar 7, 2023
Ask episode
Chapters
Transcript
Episode notes
1
Introduction
00:00 • 3min
2
The Importance of Free Speech
03:10 • 3min
3
The Importance of Asking Experts
06:13 • 2min
4
The Long-Term Risks of Artificial General Intelligence
08:13 • 3min
5
Stuart Russell and Gary Marcus on the Future of AI
11:10 • 2min
6
The Problem of Control
12:51 • 5min
7
The Importance of Motivation in Chat GPT
17:43 • 6min
8
The Inexpressive Power of Computers
23:27 • 2min
9
The Imminence of AGI
25:51 • 5min
10
The Illusion of Language Generation
30:40 • 2min
11
The Importance of Narrow AI
32:50 • 6min
12
The Problems With Narrow AI
38:20 • 5min
13
The Problems and Potential Problems With Narrow AI
43:33 • 6min
14
The Failure to Use Symbols in the Human Mind
49:31 • 2min
15
The Existential Risk of AGI
51:29 • 4min
16
Probabilistic Programming and the Future of Social Media
55:50 • 2min
17
The Problem With AI Brainwashing
58:11 • 2min
18
The Importance of Consensus on Values
01:00:25 • 3min
19
The Ethics of Deep Learning
01:03:53 • 2min
20
The Problems With AI
01:06:15 • 2min
21
The Future of AI
01:08:39 • 4min
22
The Business Model of the Internet Has Something to Do With This
01:12:27 • 4min
23
The Future of Artificial Intelligence
01:16:48 • 5min
24
The AGI Alignment and the Control Problem
01:21:20 • 6min
25
The Importance of Constraints in AI
01:27:07 • 4min
26
The Problem With Locking Down Values in Systems
01:30:45 • 2min
27
How to Lock Down Core Values in Machine Learning
01:32:35 • 4min
28
The Importance of Deep Reinforcement Learning
01:36:29 • 3min
29
The Importance of Reasoning in AI
01:39:15 • 4min
30
Door Four: Designing AI Systems in a Different Way
01:42:48 • 3min
31
The Importance of Reinforcement Learning
01:45:59 • 3min
32
The Problem With Generalized Intelligence
01:49:22 • 2min
33
The Universality of Computing
01:51:04 • 2min
34
The Internalization of Objectives
01:53:03 • 2min
35
The Incentive Constitution and the Future of Intelligent Machines
01:55:02 • 5min
36
The Importance of Human Compatibility in AI
01:59:48 • 6min
37
The Problem With Too Much Drift and Too Much Exploration of the Space
02:05:31 • 6min
38
The Failure Mode of the YouTube Algorithm
02:11:03 • 3min
39
The Promise of Real Superintelligence
02:13:53 • 6min
40
The Importance of Solving Research Questions
02:19:56 • 3min
41
Should We Stop Progress on AI Until We've Figured Out How to Do It Safely?
02:22:36 • 5min