

Dr. Keith Duggar has a high P(doom)?! Debate with MLST Co-host
Oct 8, 2024
Dr. Keith Duggar, Co-host of Machine Learning Street Talk, brings his expertise in AI and computation theory to the table. He engages in a compelling debate on the distinction between Turing Machines and LLMs, exploring their implications for AI limitations. The conversation dives into the P(doom) concept, human misuse of superintelligence, and urgent calls for policy action. Additionally, they unravel the boundaries of AI problem-solving capabilities and navigate the complexities of aligning AI with human values, making for a thought-provoking discussion.
Chapters
Transcript
Episode notes
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
Intro
00:00 • 3min
Exploring the Concept of P(doom) and Human Misuse of Superintelligence
02:51 • 2min
Navigating AI Risks: A Call to Action
04:46 • 9min
Exploring the Limits of Large Language Models
13:31 • 6min
Embracing Mistakes: The Value of Acknowledgment in Debates
19:09 • 2min
The Memory Argument: Context Sizes in Human and Machine Tasks
21:21 • 3min
Limits of Language Models and Human Intuition
24:50 • 14min
Exploring AI's Problem-Solving Limitations
39:11 • 16min
Cognitive Parallels: Humans vs. LLMs
54:59 • 14min
Consciousness and AI's Future
01:08:45 • 9min
Intelligence, Optimization, and Existential Risks
01:17:59 • 18min
Navigating the Future of AI
01:36:17 • 12min
Exploring the Orthogonality Thesis
01:48:41 • 14min
Navigating the Challenges of AI Guidance and Alignment
02:02:57 • 2min
Exploring Disagreements and Common Ground in Computational Theory
02:05:21 • 2min
Promoting Civil Discourse and Community Engagement
02:06:55 • 4min