Bonus: Preventing an AI-Related Catastrophe

1

Introduction

00:00 • 3min

2

Working on Reducing Risks From AI

02:57 • 3min

3

Artificial Intelligence (AI) Risks

05:58 • 4min

4

Is There a Risk of an Existential Catastrophe in AI?

10:09 • 2min

5

The Future of Artificial Intelligence

11:58 • 4min

6

How Much Compute Is Needed to Train the Largest Machine Learning Models?

16:25 • 2min

7

Artificial Intelligence Scaling Hypothesis

18:08 • 2min

8

The Scaling Hypothesis of Transformative AI

19:44 • 4min

9

Advanced AI Systems Could Pose a Threat to Humanity

24:04 • 4min

10

Planning in the Real World

27:51 • 2min

11

Planning Systems - A Short Footnote

29:59 • 3min

12

Advanced Planning AI Systems Could Easily Be Misaligned

32:36 • 3min

13

Artificial Intelligence and Misalignment

35:13 • 3min

14

Advanced AI Systems Could Be Dangerously Misaligned

38:08 • 3min

15

A Planned AI System That Doesn't Seek Power

41:03 • 3min

16

How to Control the Objectives of an AI System

43:57 • 3min

17

The Second Requirement for Any Strategy to Work

46:35 • 4min

18

Are We Concerned About Existential Catastrophes?

50:13 • 1min

19

Why People Might Deploy Misaligned AI

51:42 • 4min

20

What Could an Existential Catastrophe Caused by AI Really Look Like?

55:17 • 2min

21

A Power-Seeking AI Could Take Power

57:14 • 3min

22

How Advanced Planning AI Systems Could Change Our Lives

01:00:20 • 4min

23

Technique Number 6: Developing Advanced Technology

01:03:54 • 3min

24

How Could an AI Takeover Cause an Existential Catastrophe?

01:06:54 • 3min

25

Artificial Intelligence - What Failure Looks Like?

01:10:23 • 3min

26

The Second Story, Existential Catastrophe Through AI Development

01:13:30 • 3min

27

Automate Automation - Numbers 5 and 6

01:16:48 • 2min

28

AI-Caused Existential Catastrophe

01:18:21 • 4min

29

Heading AI Could Worse War

01:22:29 • 4min

30

Dot Point, AI-Related Catastrophes

01:26:03 • 3min

31

Are There Any Estimates of Existential Risks From AI?

01:29:27 • 4min

32

Working on Technical AI Safety

01:33:31 • 2min

33

How to Reduce Existential Risks From Artificial Intelligence

01:35:25 • 3min

34

Why We Shouldn't Build Transformative AI

01:38:06 • 3min

35

AI Development

01:40:45 • 4min

36

AI Alignment - The End of the Dot Points

01:44:19 • 2min

37

AI Safety

01:46:17 • 3min

38

Arguments Against Working on AI Risk

01:49:41 • 3min

39

Artificial General Intelligence

01:52:31 • 4min

40

How Can We Sandbox a Dangerous AI System?

01:57:00 • 4min

41

Why We Should Not Give AI Systems Bad Goals

02:00:31 • 3min

42

Can AI Really Do a Lot of Good?

02:03:14 • 4min

43

The Importance of Monetary Incentives for AI Systems

02:06:57 • 5min

44

The End of the Dot Points

02:11:43 • 4min

45

Working on Reducing Risks From AI Is a Good Idea

02:15:58 • 4min

46

How to Make Future AI Systems Safe

02:19:58 • 3min

47

List of Conceptual AI Safety Labs

02:23:04 • 3min

48

Technical AI Safety Research

02:25:54 • 3min

49

AI Impacts - Research and Policy

02:28:41 • 3min

50

Career Review of AI Strategy and Policy Careers

02:31:44 • 3min

51

Risks From AI - A New Career Path?

02:34:50 • 5min