Future of Life Institute Podcast

Connor Leahy on AI Safety and Why the World is Fragile

21 snips
Jan 26, 2023
Guest
Connor Leahy
Ask episode
Chapters
Transcript
Episode notes
1
Introduction
00:00 • 2min
2
How Could AI Go Wrong?
01:40 • 3min
3
The Minimum Viable Catastrophe
04:12 • 6min
4
Is It Just Because People Are Too Nice or Too Good?
09:59 • 2min
5
Why AIs Would Be Different Than Humans?
12:04 • 3min
6
The Existence Proof That the World Is Unstable
15:20 • 4min
7
Encoding Human Preferences and the Ontology of Weight Shifts
18:59 • 2min
8
Open AI
20:36 • 2min
9
The Fragment of the Fragile World
22:47 • 4min
10
Don't Be Stupid
26:49 • 2min
11
What Would We Do if We Were Not Stupid?
29:12 • 4min
12
Is There a Future for AI?
33:25 • 4min
13
How to Help Governments Not Be Stupid
37:16 • 4min
14
Is DARPA a Smart World?
41:10 • 2min
15
How Could Government Intervention Go Wrong?
43:14 • 2min
16
I Safety Funding for AI Alignment Research
45:13 • 6min
17
The Best Funding Model for a Smart Society
51:03 • 2min
18
Impact Markets
52:43 • 3min
19
Is There a for Profit Model?
55:28 • 5min
20
What Do You Think About Win Four Clauses?
01:00:42 • 2min
21
What Do Signals Matter?
01:02:59 • 2min