
LessWrong (Curated & Popular) The Plan - 2023 Version
Jan 4, 2024
The hosts discuss their plans for AI alignment, focusing on interpretability and finding alignment targets. They also highlight the importance of robust bottlenecks. The podcast explores the role of abstraction in AI systems and the challenges in choosing ontologies. It delves into good heart problems, approximation, and optimizing for true names. The concept of designing for zero information leak and the role of chaos is discussed. The challenges of abstraction and reward-based approaches in AI training are explored. The podcast also looks at the iterative process in engineering and software/AI development.
Chapters
Transcript
Episode notes
1 2 3 4 5 6 7
Introduction
00:00 • 2min
Understanding Abstraction and its Significance in AI Systems
02:23 • 10min
Exploring Good Heart Problems, Approximation, and True Names
11:58 • 2min
Designing for Zero Information Leak and the Role of Chaos
14:28 • 3min
Exploring the Challenges of Abstraction
17:41 • 14min
Challenges of Reward-Based Approaches in AI Training
31:23 • 20min
Iterative Process in Engineering and Software/AI Development
51:19 • 7min
