Roman Yampolskiy on Objections to AI Safety

1

Introduction

00:00 • 2min

2

The Objection to AI Safety Research

01:38 • 2min

3

The Limits of Intelligence

03:52 • 2min

4

The Importance of Consciousness in AI

05:38 • 2min

5

AI's Will Be Tools for Us

07:48 • 2min

6

The Pros and Cons of Shutting Off AI

09:26 • 2min

7

The Limits of AI Monitoring

11:16 • 3min

8

The Pros and Cons of Artificial Intelligence

13:48 • 2min

9

The Double Standard of AI

16:06 • 2min

10

The Drive Towards Self-Preservation in AIs

17:45 • 2min

11

The Evolution of Artificial Intelligence

19:42 • 2min

12

How AI Is Planning to Do More of What We Want

21:44 • 2min

13

The Future of Language Models

23:26 • 2min

14

The Future of AI

25:09 • 3min

15

The History of AI Safety

27:52 • 2min

16

The Future of Artificial Intelligence

29:27 • 2min

17

The Impossibility of Safely Aligning AI

31:13 • 5min

18

The Impossibility of 100% Accuracy in Cyber Security

36:07 • 2min

19

The Importance of AGI in Life

38:26 • 2min

20

The Impossibility of AI Alignment

40:44 • 2min

21

The Impossibility of Accepting Existential Risk

43:11 • 3min

22

The Immorality of Consent in Scientific Experiments

45:50 • 2min

23

The Importance of Long-Term Planning

47:26 • 2min

24

The Consequences of Increased Intelligence

49:12 • 2min

25

The Effective Acceleration Is

51:07 • 2min

26

The Pro-Human Bias

52:59 • 2min

27

AI and the Dangers It Poses for Humans

55:19 • 2min

28

The Importance of Proof in AI Safety

56:49 • 2min

29

The Cognitive Bias in AI

59:12 • 2min

30

The Problem With Extrapolating Trends

01:01:22 • 2min

31

The Evolution of AI Safety

01:03:19 • 2min

32

Ray Kurzweil's Predictions of the Future

01:05:31 • 2min

33

The Importance of Deep Learning in AI Safety

01:07:06 • 2min

34

The Importance of Monitoring AI

01:09:01 • 2min

35

How to Be Optimistic About AI Development

01:11:08 • 3min

36

The Exponent Curve of AI Accidents

01:13:45 • 2min

37

The Proportion of AI Accidents

01:15:40 • 4min

38

The Lack of a Journal for AI Safety

01:19:24 • 2min

39

AI Safety in the Machine Learning Community

01:21:26 • 3min

40

The Future of AI Safety

01:23:57 • 3min

41

The Importance of Regulation in AI

01:26:33 • 3min

42

The Upsides of Open Source Language Models

01:29:36 • 2min

43

How to Improve Mechanistic Interpretability

01:31:19 • 2min

44

The Difference Between What Comes Out of a Language Model and What's Happening at the Level of the Actual Weights

01:33:35 • 2min

45

The Preference of the Model for Predicting the Next Token

01:35:07 • 2min

46

The Importance of Concepts in Language Models

01:36:53 • 2min

47

The Implications of Specialization in AI Safety

01:39:13 • 3min