Future of Life Institute Podcast Anders Sandberg on ChatGPT and the Future of AI
5 snips
Dec 15, 2022 Chapters
Transcript
Episode notes
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
Introduction
00:00 • 2min
GPT Line Models - How Do They Work?
02:01 • 3min
How Do These Models Actually Work?
05:02 • 2min
Is the Language Model Really Intelligent?
06:34 • 3min
The Future of Autonomous Cars
09:24 • 1min
Using GPT Two to Train a Neural Network
10:48 • 5min
The Failure Rate of Language Models
16:12 • 4min
Automated Fact Checking Could Be Very Powerful
20:21 • 4min
Language Models Are Creating a Spam of Collapsing
24:25 • 3min
Are Auto Generated Philosophy Essays a Threat?
27:30 • 3min
Why Should We Artificially Handicap Ourselves?
30:19 • 5min
Do Language Models Really Understand a Topic?
35:03 • 3min
Do Language Models Generate Lists of Actions?
37:52 • 2min
How to Make a Multi-Modal Language Model?
39:31 • 3min
How Have Your Opinions About AI Safety Changed?
42:06 • 2min
Artificial Intelligence and the Safety Debate
44:18 • 6min
AI Safety - What Values Should We Give to AI?
50:07 • 1min
Is There a Value Alignment Problem?
51:36 • 5min
Is There a Black and White Utility Function in AI?
56:23 • 2min