Jim Keller: The Future of Computing, AI, Life, and Consciousness | Lex Fridman Podcast #162 | Lex Fridman

In this episode of the Lex Fridman Podcast, Lex interviews Jim Keller, a microprocessor architect, about the value of theory versus engineering in building software and hardware systems. They discuss branch prediction, the importance of craftsmanship in engineering, and the evolution of instruction sets, including RISC-V. Keller shares insights on Intel's historical success, the leadership styles of Steve Jobs and Elon Musk, and the balance between chaos and order in organizations. The conversation shifts to deep learning hardware, with Keller detailing his work at Tenstorrent and the potential of graph-based computing. They explore Moore's Law, the future of computing, and the nature of consciousness, also touching on the GameStop saga, advice for young people, and the role of love and family in a successful life.

Outlines

Part 1: Foundations and Evolution

Part 2: Management, Design, and Cognitive Limits

Part 3: Tenstorrent and AI Hardware

Part 4: Autonomous Driving and Neural Networks

Part 5: Brain, Consciousness, and the Human Mind

Part 6: Society, Advice, and Personal Reflections

Sign in to continue reading, translating and more.

Open full episode in Podwise

Jim Keller: The Future of Computing, AI, Life, and Consciousness | Lex Fridman Podcast #162

Lex Fridman

Part 1: Foundations and Evolution

Theory vs. Engineering in System Building and the Importance of Craftsmanship

The Right Tool at the Right Time: JavaScript, x86, and the Evolution of Processor Design

ARM vs. Intel: Mobile Platforms, Pivots, and the Order vs. Chaos Dynamic

Part 2: Management, Design, and Cognitive Limits

Managing People, Childhood Influences, and the Creative Tension Between Flaws and Successes

Abstraction Layers, Modular Design, and the Limits of Human Cognitive Capacity

Moore's Law, Scaling, and the Inefficient Future of Computing

Part 3: Tenstorrent and AI Hardware

Given vs. Found Parallelism and Tenstorrent's Graph-Centric Hardware

PyTorch, Graph Compilers, and the Goal of Efficient AI Execution

Chris Lattner, MLIR, and the Future of Deep Learning Hardware

Part 4: Autonomous Driving and Neural Networks

Tesla Autopilot, Data Engines, and the Limits of Self-Supervised Learning

Dojo, General-Purpose AI, and the Convergence of Neural Networks

Part 5: Brain, Consciousness, and the Human Mind

The Brain, Neuroplasticity, and the Potential for Brain-Computer Interfaces

Psychedelics, Dreams, and the Source of Creative Ideas

The Nature of Consciousness and the Search for Alien Intelligence

Part 6: Society, Advice, and Personal Reflections

Ideas, Groupthink, and the Roughness of the Human Journey

Pandemics, GameStop, and the Power of Distributed Crowds

Advice for Young People: Finding Your Passion and Avoiding Traps

The Functional Nature of Love and the Importance of Balance

Regrets, Politics, and the Human Layer of the Stack

Legacy and Closing Remarks

Jim Keller: The Future of Computing, AI, Life, and Consciousness | Lex Fridman Podcast #162

Lex Fridman

Part 1: Foundations and Evolution

00:00Theory vs. Engineering in System Building and the Importance of Craftsmanship

Theory vs. Engineering in System Building and the Importance of Craftsmanship

07:34The Right Tool at the Right Time: JavaScript, x86, and the Evolution of Processor Design

The Right Tool at the Right Time: JavaScript, x86, and the Evolution of Processor Design

17:23ARM vs. Intel: Mobile Platforms, Pivots, and the Order vs. Chaos Dynamic

ARM vs. Intel: Mobile Platforms, Pivots, and the Order vs. Chaos Dynamic

Part 2: Management, Design, and Cognitive Limits

25:52Managing People, Childhood Influences, and the Creative Tension Between Flaws and Successes

Managing People, Childhood Influences, and the Creative Tension Between Flaws and Successes

35:37Abstraction Layers, Modular Design, and the Limits of Human Cognitive Capacity

Abstraction Layers, Modular Design, and the Limits of Human Cognitive Capacity

42:52Moore's Law, Scaling, and the Inefficient Future of Computing

Moore's Law, Scaling, and the Inefficient Future of Computing

Part 3: Tenstorrent and AI Hardware

50:05Given vs. Found Parallelism and Tenstorrent's Graph-Centric Hardware

Given vs. Found Parallelism and Tenstorrent's Graph-Centric Hardware

57:12PyTorch, Graph Compilers, and the Goal of Efficient AI Execution

PyTorch, Graph Compilers, and the Goal of Efficient AI Execution

1:04:52Chris Lattner, MLIR, and the Future of Deep Learning Hardware

Chris Lattner, MLIR, and the Future of Deep Learning Hardware

Part 4: Autonomous Driving and Neural Networks

1:15:54Tesla Autopilot, Data Engines, and the Limits of Self-Supervised Learning

Tesla Autopilot, Data Engines, and the Limits of Self-Supervised Learning

1:22:01Dojo, General-Purpose AI, and the Convergence of Neural Networks

Dojo, General-Purpose AI, and the Convergence of Neural Networks

Part 5: Brain, Consciousness, and the Human Mind

1:31:37The Brain, Neuroplasticity, and the Potential for Brain-Computer Interfaces

The Brain, Neuroplasticity, and the Potential for Brain-Computer Interfaces

1:40:11Psychedelics, Dreams, and the Source of Creative Ideas

Psychedelics, Dreams, and the Source of Creative Ideas

1:49:50The Nature of Consciousness and the Search for Alien Intelligence

The Nature of Consciousness and the Search for Alien Intelligence

Part 6: Society, Advice, and Personal Reflections

1:57:22Ideas, Groupthink, and the Roughness of the Human Journey

Ideas, Groupthink, and the Roughness of the Human Journey

2:04:44Pandemics, GameStop, and the Power of Distributed Crowds

Pandemics, GameStop, and the Power of Distributed Crowds

2:15:12Advice for Young People: Finding Your Passion and Avoiding Traps

Advice for Young People: Finding Your Passion and Avoiding Traps

2:25:05The Functional Nature of Love and the Importance of Balance

The Functional Nature of Love and the Importance of Balance

2:31:23Regrets, Politics, and the Human Layer of the Stack

Regrets, Politics, and the Human Layer of the Stack

2:38:35Legacy and Closing Remarks

Legacy and Closing Remarks