Dwarkesh Podcast Shane Legg (DeepMind Founder) - 2028 AGI, New Architectures, Aligning Superhuman Models
131 snips
Oct 26, 2023 
Shane Legg, Founder and Chief AGI Scientist at Google DeepMind, shares his insights on the future of artificial general intelligence. He predicts we might see AGI by 2028 and discusses the importance of new architectures in achieving this goal. The dialogue dives deep into aligning superhuman models with human values and the balance between AI capabilities and safety. Legg also highlights multimodality as the next frontier for AI, emphasizing its potential to interpret diverse inputs beyond text.
AI Snips
Chapters
Books
Transcript
Episode notes
Measuring AGI Progress
- AGI is defined by its generality, not specific skills.
- Measuring progress requires diverse tests spanning human cognitive abilities.
LLM Limitations
- Current language models lack episodic memory, crucial for rapid learning.
- They also struggle with understanding streaming video, a key human cognitive ability.
Sample Efficiency and Episodic Memory
- Episodic memory is related to sample efficiency, enabling rapid learning.
- LLMs possess a form of sample efficiency within their context window.
