title
Jim Keller: Moore's Law, Microprocessors, and First Principles | Lex Fridman Podcast #70

description
Jim Keller is a legendary microprocessor engineer, having worked at AMD, Apple, Tesla, and now Intel. He's known for his work on the AMD K7, K8, K12 and Zen microarchitectures, Apple A4, A5 processors, and co-author of the specifications for the x86-64 instruction set and HyperTransport interconnect. This episode is presented by Cash App. Download it & use code "LexPodcast": Cash App (App Store): https://apple.co/2sPrUHe Cash App (Google Play): https://bit.ly/2MlvP5w PODCAST INFO: Podcast website: https://lexfridman.com/podcast Apple Podcasts: https://apple.co/2lwqZIr Spotify: https://spoti.fi/2nEwCF8 RSS: https://lexfridman.com/feed/podcast/ Full episodes playlist: https://www.youtube.com/playlist?list=PLrAXtmErZgOdP_8GztsuKi9nrraNbKKp4 Clips playlist: https://www.youtube.com/playlist?list=PLrAXtmErZgOeciFP3CBCIEElOJeitOr41 OUTLINE: 0:00 - Introduction 2:12 - Difference between a computer and a human brain 3:43 - Computer abstraction layers and parallelism 17:53 - If you run a program multiple times, do you always get the same answer? 20:43 - Building computers and teams of people 22:41 - Start from scratch every 5 years 30:05 - Moore's law is not dead 55:47 - Is superintelligence the next layer of abstraction? 1:00:02 - Is the universe a computer? 1:03:00 - Ray Kurzweil and exponential improvement in technology 1:04:33 - Elon Musk and Tesla Autopilot 1:20:51 - Lessons from working with Elon Musk 1:28:33 - Existential threats from AI 1:32:38 - Happiness and the meaning of life CONNECT: - Subscribe to this YouTube channel - Twitter: https://twitter.com/lexfridman - LinkedIn: https://www.linkedin.com/in/lexfridman - Facebook: https://www.facebook.com/LexFridmanPage - Instagram: https://www.instagram.com/lexfridman - Medium: https://medium.com/@lexfridman - Support on Patreon: https://www.patreon.com/lexfridman

detail
{'title': "Jim Keller: Moore's Law, Microprocessors, and First Principles | Lex Fridman Podcast #70", 'heatmap': [{'end': 1768.48, 'start': 1703.676, 'weight': 1}], 'summary': 'Renowned microprocessor engineer jim keller discusses his career, computer architecture, quantum transistors, computing performance, future technology, exponential growth in technology, autonomous vehicle safety, and lessons from tesla, with insights into ai computations, potential 100x innovation, and vehicle autonomy challenges.', 'chapters': [{'end': 131.188, 'segs': [{'end': 25.155, 'src': 'embed', 'start': 0.049, 'weight': 0, 'content': [{'end': 9.131, 'text': 'The following is a conversation with Jim Keller, legendary microprocessor engineer who has worked at AMD, Apple, Tesla, and now Intel.', 'start': 0.049, 'duration': 9.082}, {'end': 18.213, 'text': "He's known for his work on AMD K7, K8, K12, and Zen microarchitectures, Apple A4 and A5 processors,", 'start': 10.211, 'duration': 8.002}, {'end': 25.155, 'text': 'and co-author of the specification for the x86-64 instruction set and HyperTransport Interconnect.', 'start': 18.213, 'duration': 6.942}], 'summary': 'Jim keller is a renowned microprocessor engineer who has worked at amd, apple, tesla, and intel, known for his work on various microarchitectures and processor specifications.', 'duration': 25.106, 'max_score': 0.049, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/Nb2tebYAaOA/pics/Nb2tebYAaOA49.jpg'}, {'end': 131.188, 'src': 'embed', 'start': 110.774, 'weight': 1, 'content': [{'end': 120.521, 'text': "When you get Cash App from the App Store or Google Play and use code LEXPODCAST, you'll get $10, and Cash App will also donate $10 to FIRST, which,", 'start': 110.774, 'duration': 9.747}, {'end': 127.105, 'text': "again, is an organization that I've personally seen inspire girls and boys to dream of engineering a better world.", 'start': 120.521, 'duration': 6.584}, {'end': 131.188, 'text': "And now, here's my conversation with Jim Keller.", 'start': 128.106, 'duration': 3.082}], 'summary': 'Use code lexpodcast on cash app for $10 and donate $10 to first', 'duration': 20.414, 'max_score': 110.774, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/Nb2tebYAaOA/pics/Nb2tebYAaOA110774.jpg'}], 'start': 0.049, 'title': 'Conversation with jim keller', 'summary': 'Features a conversation with jim keller, a renowned microprocessor engineer who has contributed to various significant microarchitectures and processors. it highlights his impressive career, current endeavors, and sponsorship details for the artificial intelligence podcast, including a partnership with cash app to support the organization first, offering listeners a $10 reward and a $10 donation to first when using the code lexpodcast on cash app.', 'chapters': [{'end': 131.188, 'start': 0.049, 'title': 'Conversation with jim keller', 'summary': 'Features a conversation with jim keller, a legendary microprocessor engineer who has worked at amd, apple, tesla, and now intel, known for his work on amd k7, k8, k12, and zen microarchitectures, apple a4 and a5 processors, and co-author of the specification for the x86-64 instruction set and hypertransport interconnect, with a focus on his impressive career and current endeavors, as well as sponsorship details for the artificial intelligence podcast, including a partnership with cash app to support the organization first, known for their robotics and lego competitions, offering listeners a $10 reward and a $10 donation to first when using the code lexpodcast on cash app.', 'duration': 131.139, 'highlights': ["Jim Keller's background and work at AMD, Apple, Tesla, and Intel, including his contributions to microarchitectures and processors, such as AMD K7, K8, K12, Zen, Apple A4, and A5.", "Details about the Artificial Intelligence Podcast, its promotion on various platforms, and the presenter's partnership with Cash App, showcasing his involvement in supporting the organization FIRST and the $10 reward and donation offer for listeners using the code LEXPODCAST on Cash App."]}], 'duration': 131.139, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/Nb2tebYAaOA/pics/Nb2tebYAaOA49.jpg', 'highlights': ["Jim Keller's significant contributions to microarchitectures and processors at AMD, Apple, Tesla, and Intel.", "Details about the Artificial Intelligence Podcast's promotion and the presenter's partnership with Cash App to support the organization FIRST, offering a $10 reward and donation for listeners using the code LEXPODCAST on Cash App."]}, {'end': 1279.049, 'segs': [{'end': 208.109, 'src': 'embed', 'start': 182.476, 'weight': 0, 'content': [{'end': 189.259, 'text': "Well, what people observe is there's some number of layers of neurons which have local and global connections,", 'start': 182.476, 'duration': 6.783}, {'end': 192.661, 'text': 'and information is stored in some distributed fashion.', 'start': 189.259, 'duration': 3.402}, {'end': 202.667, 'text': 'And people build things called neural networks in computers where the information is distributed in some kind of fashion.', 'start': 193.942, 'duration': 8.725}, {'end': 204.287, 'text': "There's a mathematics behind it.", 'start': 202.687, 'duration': 1.6}, {'end': 208.109, 'text': "I don't know that the understanding of that is super deep.", 'start': 205.568, 'duration': 2.541}], 'summary': 'Neural networks have multiple layers of connected neurons, storing information in a distributed fashion, with a mathematical basis.', 'duration': 25.633, 'max_score': 182.476, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/Nb2tebYAaOA/pics/Nb2tebYAaOA182476.jpg'}, {'end': 440.728, 'src': 'embed', 'start': 413.007, 'weight': 1, 'content': [{'end': 415.789, 'text': 'Do the load, do the add, do the compare.', 'start': 413.007, 'duration': 2.782}, {'end': 419.111, 'text': 'The way a modern computer works is.', 'start': 417.15, 'duration': 1.961}, {'end': 422.713, 'text': 'you fetch large numbers of instructions say 500,', 'start': 419.111, 'duration': 3.602}, {'end': 433.978, 'text': 'and then you find the dependency graph between the instructions and then you execute in independent units those little micrographs.', 'start': 422.713, 'duration': 11.265}, {'end': 440.728, 'text': 'So a modern computer, like people like to say, computers should be simple and clean.', 'start': 435.325, 'duration': 5.403}], 'summary': 'Modern computers fetch and execute 500 instructions with independent micrographs.', 'duration': 27.721, 'max_score': 413.007, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/Nb2tebYAaOA/pics/Nb2tebYAaOA413007.jpg'}, {'end': 670.539, 'src': 'embed', 'start': 642.29, 'weight': 2, 'content': [{'end': 645.552, 'text': 'how much found parallelism is there in them?', 'start': 642.29, 'duration': 3.262}, {'end': 646.673, 'text': 'What does 10x mean?', 'start': 645.833, 'duration': 0.84}, {'end': 654.999, 'text': "If you execute it in order, you would get what's called cycles per instruction and it would be about.", 'start': 647.634, 'duration': 7.365}, {'end': 662.006, 'text': 'you know, three instructions, three cycles per instruction because of the latency of the operations and stuff.', 'start': 656.798, 'duration': 5.208}, {'end': 668.696, 'text': 'And a modern computer executes it, but like 0.2, 0.25 cycles per instruction.', 'start': 662.968, 'duration': 5.728}, {'end': 670.539, 'text': "So it's about, we today find 10X.", 'start': 668.716, 'duration': 1.823}], 'summary': 'Modern computers achieve 10x faster execution than sequential ones.', 'duration': 28.249, 'max_score': 642.29, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/Nb2tebYAaOA/pics/Nb2tebYAaOA642290.jpg'}], 'start': 132.671, 'title': 'Computer architecture and brain comparison', 'summary': "Explores the differences and similarities between the human brain and computer architecture, emphasizing the complexity of the human brain's neural network, and discusses the evolution of computer architecture, including instruction set stability, shift from sequential to parallel narrative, and techniques of predicting branches in modern computers.", 'chapters': [{'end': 356.273, 'start': 132.671, 'title': 'Human brain vs. computer architecture', 'summary': "Explores the differences and similarities between the human brain and computer architecture, emphasizing the complexity of the human brain's neural network and the abstraction layers in computer engineering.", 'duration': 223.602, 'highlights': ["The complexity of the human brain's neural network with local and global connections and distributed information storage is emphasized, contrasting with the decoupled memory and computation architecture of modern computers.", 'The abstraction layers in computer engineering, from atoms to functional units and software, are highlighted, illustrating the hierarchical nature of computer architecture and the diverse disciplines involved in computer design.', "The challenge of comparing the human brain and computers due to the lack of comprehensive understanding of the brain's functioning and the straightforward computations performed by neural networks in computers is discussed."]}, {'end': 642.29, 'start': 356.293, 'title': 'Evolution of computer architecture', 'summary': 'Discusses the stability and evolution of instruction sets, the shift from sequential to parallel narrative in modern computers and gpus, and the determination of parallelism based on transistor count and algorithmic dependencies.', 'duration': 285.997, 'highlights': ['Instruction sets like x86 and ARM are stable and constitute 90% of program execution with only 25 opcodes. Programs run with 90% of execution on 25 opcodes, reflecting the stability of instruction sets like x86 and ARM.', 'Modern computers fetch a large number of instructions, compute the dependency graph, and execute in independent units to optimize performance. Modern computers fetch and execute a large number of instructions, computing dependency graphs to optimize performance.', 'GPUs exhibit found parallelism by running simple programs on pixels with a million of them, which do not depend on sequential order for display. GPUs leverage found parallelism to run simple programs on pixels, independent of sequential order for display.', 'The determination of parallelism in modern programs is linked to the transistor count and algorithmic dependencies. The determination of parallelism in modern programs is related to the transistor count and algorithmic dependencies.']}, {'end': 1279.049, 'start': 642.29, 'title': 'Predicting branches in modern computers', 'summary': 'Discusses the challenges and techniques of predicting branches in modern computers, including the found parallelism and predictability of the narrative, the evolution of branch prediction accuracy, the impact of branch prediction on performance, and the consideration of determinism in program execution.', 'duration': 636.759, 'highlights': ['Modern computers aim to achieve about 10x improvement in cycles per instruction compared to executing instructions in order, primarily through exploiting found parallelism and accurately predicting branches, with modern branch prediction achieving high 90% accuracy.', 'The evolution of branch prediction accuracy from simple methods using history and counters to modern techniques resembling neural networks, with the prediction accuracy increasing from 85% to over 99% and requiring tens of megabits of storage.', 'The significance of branch prediction on window size and performance, where a larger effective window requires higher accuracy in predicting branches to maintain efficiency, and a mispredicted branch results in flushing the pipeline and performance cost.', 'The consideration of determinism in program execution, where the formal definition in programming languages requires the same result every time a program is run, despite the inherent non-deterministic execution flow in modern computers.']}], 'duration': 1146.378, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/Nb2tebYAaOA/pics/Nb2tebYAaOA132671.jpg', 'highlights': ["The complexity of the human brain's neural network with local and global connections and distributed information storage is emphasized, contrasting with the decoupled memory and computation architecture of modern computers.", 'Modern computers fetch and execute a large number of instructions, computing dependency graphs to optimize performance.', 'Modern computers aim to achieve about 10x improvement in cycles per instruction compared to executing instructions in order, primarily through exploiting found parallelism and accurately predicting branches, with modern branch prediction achieving high 90% accuracy.']}, {'end': 2021.526, 'segs': [{'end': 1343.841, 'src': 'embed', 'start': 1319.942, 'weight': 1, 'content': [{'end': 1327.008, 'text': 'Yeah, I said, I realized, after a lot of years of building computers where you sort of build them out of transistors, logic gates, functional units,', 'start': 1319.942, 'duration': 7.066}, {'end': 1330.491, 'text': 'computational elements, that you could think of people the same way.', 'start': 1327.008, 'duration': 3.483}, {'end': 1332.133, 'text': 'So people are functional units.', 'start': 1330.792, 'duration': 1.341}, {'end': 1335.756, 'text': 'And then you could think of organizational design as a computer architectural problem.', 'start': 1332.673, 'duration': 3.083}, {'end': 1343.841, 'text': "And then it was like oh, that's super cool, because the people are all different, just like the computational elements are all different,", 'start': 1336.957, 'duration': 6.884}], 'summary': 'Organizational design can be viewed as a computer architectural problem, treating people as functional units.', 'duration': 23.899, 'max_score': 1319.942, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/Nb2tebYAaOA/pics/Nb2tebYAaOA1319942.jpg'}, {'end': 1473.153, 'src': 'embed', 'start': 1443.885, 'weight': 0, 'content': [{'end': 1446.649, 'text': "There's a different way of viewing everything.", 'start': 1443.885, 'duration': 2.764}, {'end': 1450.115, 'text': 'And most people.', 'start': 1447.871, 'duration': 2.244}, {'end': 1459.663, 'text': "when you get to be an expert at something, You're hoping to achieve deeper understanding, not just a large set of recipes to go execute.", 'start': 1450.115, 'duration': 9.548}, {'end': 1468.91, 'text': "And it's interesting to walk groups of people because executing recipes is unbelievably efficient if it's what you want to do.", 'start': 1461.004, 'duration': 7.906}, {'end': 1473.153, 'text': "If it's not what you want to do, you're really stuck.", 'start': 1470.551, 'duration': 2.602}], 'summary': 'Becoming an expert means deeper understanding, not just following recipes for efficient execution.', 'duration': 29.268, 'max_score': 1443.885, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/Nb2tebYAaOA/pics/Nb2tebYAaOA1443885.jpg'}, {'end': 1698.834, 'src': 'embed', 'start': 1670.122, 'weight': 2, 'content': [{'end': 1673.904, 'text': 'And instruction sets are less interesting than the implementation underneath.', 'start': 1670.122, 'duration': 3.782}, {'end': 1681.627, 'text': "There's been, on x86 architecture, Intel's designed a few, AMD's designed a few very different architectures.", 'start': 1674.764, 'duration': 6.863}, {'end': 1692.531, 'text': "And I don't want to go into too much of the detail about how often, but there's a tendency to rewrite it every 10 years,", 'start': 1682.707, 'duration': 9.824}, {'end': 1693.892, 'text': 'and it really should be every five.', 'start': 1692.531, 'duration': 1.361}, {'end': 1698.834, 'text': "So you're saying you're an outlier in that sense? Rewrite more often.", 'start': 1695.212, 'duration': 3.622}], 'summary': 'X86 architecture undergoes redesign every 10 years, should be every 5.', 'duration': 28.712, 'max_score': 1670.122, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/Nb2tebYAaOA/pics/Nb2tebYAaOA1670122.jpg'}, {'end': 1768.48, 'src': 'heatmap', 'start': 1703.676, 'weight': 1, 'content': [{'end': 1709.824, 'text': 'Well, scary to who? to everybody involved, because like you said, repeating the recipe is efficient.', 'start': 1703.676, 'duration': 6.148}, {'end': 1713.225, 'text': 'Companies want to make money.', 'start': 1710.744, 'duration': 2.481}, {'end': 1720.488, 'text': 'No, individual engineers want to succeed, so you want to incrementally improve, increase the buffer from three to four.', 'start': 1714.226, 'duration': 6.262}, {'end': 1724.61, 'text': 'This is where you get into diminishing return curves.', 'start': 1720.548, 'duration': 4.062}, {'end': 1731.834, 'text': 'I think Steve Jobs said this, right? You have a project and you start here and it goes up and you have the mission return.', 'start': 1725.45, 'duration': 6.384}, {'end': 1734.715, 'text': 'And to get to the next level, you have to do a new one.', 'start': 1732.394, 'duration': 2.321}, {'end': 1741.079, 'text': "And the initial starting point will be lower than the old optimization point, but it'll get higher.", 'start': 1734.835, 'duration': 6.244}, {'end': 1746.523, 'text': 'So now you have two kinds of fear, short-term disaster and long-term disaster.', 'start': 1741.9, 'duration': 4.623}, {'end': 1757.412, 'text': 'People with a quarter by quarter business objective are terrified about changing everything.', 'start': 1752.207, 'duration': 5.205}, {'end': 1768.48, 'text': 'And people who are trying to run a business or build a computer for a long-term objective know that the short-term limitations block them from the long-term success.', 'start': 1757.912, 'duration': 10.568}], 'summary': 'Efficiency in repeating recipe scares all involved, impacting short-term and long-term success in business and engineering.', 'duration': 64.804, 'max_score': 1703.676, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/Nb2tebYAaOA/pics/Nb2tebYAaOA1703676.jpg'}, {'end': 1768.48, 'src': 'embed', 'start': 1741.9, 'weight': 5, 'content': [{'end': 1746.523, 'text': 'So now you have two kinds of fear, short-term disaster and long-term disaster.', 'start': 1741.9, 'duration': 4.623}, {'end': 1757.412, 'text': 'People with a quarter by quarter business objective are terrified about changing everything.', 'start': 1752.207, 'duration': 5.205}, {'end': 1768.48, 'text': 'And people who are trying to run a business or build a computer for a long-term objective know that the short-term limitations block them from the long-term success.', 'start': 1757.912, 'duration': 10.568}], 'summary': 'Two types of fear: short-term disaster and long-term disaster impact business objectives.', 'duration': 26.58, 'max_score': 1741.9, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/Nb2tebYAaOA/pics/Nb2tebYAaOA1741900.jpg'}, {'end': 1854.837, 'src': 'embed', 'start': 1831.075, 'weight': 3, 'content': [{'end': 1838.562, 'text': 'Well, the simple statement was from Gordon Moore was double the number of transistors every two years, something like that.', 'start': 1831.075, 'duration': 7.487}, {'end': 1847.911, 'text': 'And then my operational model is we increase the performance of computers by two X every two or three years.', 'start': 1839.383, 'duration': 8.528}, {'end': 1850.753, 'text': "And it's wiggled around substantially over time.", 'start': 1848.591, 'duration': 2.162}, {'end': 1854.837, 'text': 'And also in how we deliver performance has changed.', 'start': 1851.534, 'duration': 3.303}], 'summary': "Gordon moore's law of doubling transistors has driven computer performance to increase by two x every two or three years, adapting over time.", 'duration': 23.762, 'max_score': 1831.075, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/Nb2tebYAaOA/pics/Nb2tebYAaOA1831075.jpg'}, {'end': 2025.688, 'src': 'embed', 'start': 1997.398, 'weight': 4, 'content': [{'end': 2000.739, 'text': 'So if you graph it, it looks like a cascade of diminishing return curves.', 'start': 1997.398, 'duration': 3.341}, {'end': 2007.481, 'text': "I don't know what to call that, but the result is an exponential curve, at least it has been.", 'start': 2000.759, 'duration': 6.722}, {'end': 2010.842, 'text': 'So, and we keep inventing new things.', 'start': 2007.981, 'duration': 2.861}, {'end': 2019.885, 'text': "So if you're an expert in one of the things on a diminishing return curve right and you can see its plateau, you will probably tell people well,", 'start': 2010.862, 'duration': 9.023}, {'end': 2021.526, 'text': 'this is done.', 'start': 2019.885, 'duration': 1.641}, {'end': 2025.688, 'text': 'Meanwhile, some other pilot people are doing something different.', 'start': 2022.226, 'duration': 3.462}], 'summary': 'Innovation leads to exponential growth despite diminishing returns.', 'duration': 28.29, 'max_score': 1997.398, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/Nb2tebYAaOA/pics/Nb2tebYAaOA1997398.jpg'}], 'start': 1280.748, 'title': "Organizational computers and moore's law", 'summary': "Discusses the parallels between building computers and organizing people, emphasizing the importance of understanding and rewrites, and also examines the challenges of innovation and adaptation in the tech industry, specifically in relation to moore's law.", 'chapters': [{'end': 1693.892, 'start': 1280.748, 'title': 'Reframing organizations as computers', 'summary': 'Discusses the parallels between building computers and organizing people, emphasizing the importance of deeper understanding and periodic rewrites for progress in computer architecture, with insights into the x86-64 standard.', 'duration': 413.144, 'highlights': ['The importance of deeper understanding and periodic rewrites for progress in computer architecture, with insights into the x86-64 standard The chapter emphasizes the significance of achieving deeper understanding rather than executing a large set of recipes, and advocates for periodic rewrites in computer architecture every three to five years, with insights into the x86-64 standard and its extensions.', 'Parallels between building computers and organizing people The chapter draws parallels between building computers and organizing people, highlighting the fascination with how people and computers go together and reframing organizational design as a computer architectural problem.', 'The tendency to rewrite computer architecture every three to five years The speaker suggests rewriting computer architecture every three to five years to make significant progress, with a personal tendency to rewrite more often and emphasizing the importance of periodic rewrites for advancement.']}, {'end': 2021.526, 'start': 1695.212, 'title': "The challenge of moore's law", 'summary': "Discusses the challenges of innovation and adaptation in the tech industry, specifically in relation to moore's law, highlighting the fear of change, the need for incremental improvement, and the complex nature of technological advancements.", 'duration': 326.314, 'highlights': ["Moore's Law: Double the number of transistors every two years Gordon Moore's statement of doubling the number of transistors every two years serves as the foundational idea of Moore's Law.", 'Diminishing Return Curves in Technological Innovation The chapter discusses the cascade of diminishing return curves in technological innovations, highlighting the existence of thousands of innovations with their own diminishing return curves, ultimately contributing to an exponential curve of progress.', 'Fear of Change and Incremental Improvement The fear of change and the need for incremental improvement are emphasized, with individuals and companies facing the challenge of balancing short-term and long-term objectives in technological advancements.']}], 'duration': 740.778, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/Nb2tebYAaOA/pics/Nb2tebYAaOA1280748.jpg', 'highlights': ['The importance of achieving deeper understanding rather than executing a large set of recipes in computer architecture', 'The parallels between building computers and organizing people, reframing organizational design as a computer architectural problem', 'The tendency to rewrite computer architecture every three to five years for significant progress', "Moore's Law: Double the number of transistors every two years", 'The cascade of diminishing return curves in technological innovations, contributing to an exponential curve of progress', 'Emphasis on the fear of change and the need for incremental improvement in technological advancements']}, {'end': 2330.216, 'segs': [{'end': 2073.181, 'src': 'embed', 'start': 2044.719, 'weight': 0, 'content': [{'end': 2052.166, 'text': "so you can imagine a transistor as small as ten by ten by ten, so that's a million times smaller.", 'start': 2044.719, 'duration': 7.447}, {'end': 2057.571, 'text': 'and then the quantum computational people are working away at how to use quantum effects.', 'start': 2052.166, 'duration': 5.405}, {'end': 2066.637, 'text': "so a thousand by a thousand by a thousand atoms, That's a really clean way of putting it.", 'start': 2057.571, 'duration': 9.066}, {'end': 2073.181, 'text': "Well, a fan, like a modern transistor, if you look at the fan, it's like 120 atoms wide, but we can make that thinner.", 'start': 2066.657, 'duration': 6.524}], 'summary': 'Quantum computing aims for a thousandfold reduction in transistor size.', 'duration': 28.462, 'max_score': 2044.719, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/Nb2tebYAaOA/pics/Nb2tebYAaOA2044719.jpg'}, {'end': 2144.698, 'src': 'embed', 'start': 2099.629, 'weight': 3, 'content': [{'end': 2107.795, 'text': 'if placing an atom takes 10 minutes and you need to put 10 to the 23rd atoms together to make a computer, it would take a long time.', 'start': 2099.629, 'duration': 8.166}, {'end': 2117.052, 'text': "The methods are both shrinking things and then coming up with effective ways to control what's happening.", 'start': 2109.156, 'duration': 7.896}, {'end': 2119.578, 'text': 'Manufacture stably and cheaply.', 'start': 2117.994, 'duration': 1.584}, {'end': 2123.487, 'text': "Yeah So the innovation stack's pretty broad.", 'start': 2120.046, 'duration': 3.441}, {'end': 2130.29, 'text': "There's equipment, there's optics, there's chemistry, there's physics, there's material science, there's metallurgy.", 'start': 2123.828, 'duration': 6.462}, {'end': 2133.732, 'text': "There's lots of ideas about when you put different materials together.", 'start': 2131.03, 'duration': 2.702}, {'end': 2134.432, 'text': 'how do they interact?', 'start': 2133.732, 'duration': 0.7}, {'end': 2135.372, 'text': 'Are they stable?', 'start': 2134.512, 'duration': 0.86}, {'end': 2136.853, 'text': 'Are they stable over temperature?', 'start': 2135.572, 'duration': 1.281}, {'end': 2144.698, 'text': 'Are they repeatable? like literally thousands of technologies involved?', 'start': 2136.873, 'duration': 7.825}], 'summary': 'Placing 10^23 atoms to make a computer takes a long time. innovation stack includes optics, chemistry, physics, material science, metallurgy, and thousands of technologies.', 'duration': 45.069, 'max_score': 2099.629, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/Nb2tebYAaOA/pics/Nb2tebYAaOA2099629.jpg'}, {'end': 2197.729, 'src': 'embed', 'start': 2169.686, 'weight': 1, 'content': [{'end': 2175.975, 'text': 'So I believe that the next 10 or 20 years of shrinking is gonna happen, right?', 'start': 2169.686, 'duration': 6.289}, {'end': 2180.301, 'text': 'Now, as a computer designer, you have two stances.', 'start': 2176.395, 'duration': 3.906}, {'end': 2188.245, 'text': "You think it's going to shrink, in which case you're designing and thinking about architecture in a way that you'll use more transistors.", 'start': 2180.982, 'duration': 7.263}, {'end': 2197.729, 'text': 'Or conversely, not be swamped by the complexity of all the transistors you get, right? You have to have a strategy.', 'start': 2189.065, 'duration': 8.664}], 'summary': 'Computer designers must strategize for increasing transistor usage in the next 10-20 years.', 'duration': 28.043, 'max_score': 2169.686, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/Nb2tebYAaOA/pics/Nb2tebYAaOA2169686.jpg'}, {'end': 2270.268, 'src': 'embed', 'start': 2243.215, 'weight': 2, 'content': [{'end': 2249.779, 'text': "because maybe you're getting something out of the smaller bricks more strength, thinner walls, you know, less material efficiency out of that.", 'start': 2243.215, 'duration': 6.564}, {'end': 2255.762, 'text': "So once you have a roadmap with what's going to happen, transistors, they're going to get, we're going to get more of them.", 'start': 2250.379, 'duration': 5.383}, {'end': 2261.384, 'text': 'Then you design all this collateral around it to take advantage of it and also to cope with it.', 'start': 2256.542, 'duration': 4.842}, {'end': 2263.725, 'text': "Like that's the thing people don't understand.", 'start': 2262.424, 'duration': 1.301}, {'end': 2270.268, 'text': "It's like, if I didn't believe in Moore's law and then Moore's law transistors showed up, my design teams were all drowned.", 'start': 2263.765, 'duration': 6.503}], 'summary': 'Leveraging smaller bricks for more strength and efficiency, anticipating more transistors, and adapting designs to cope with technological advancements are crucial for success.', 'duration': 27.053, 'max_score': 2243.215, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/Nb2tebYAaOA/pics/Nb2tebYAaOA2243215.jpg'}, {'end': 2307.388, 'src': 'embed', 'start': 2282.279, 'weight': 5, 'content': [{'end': 2289.767, 'text': 'what fundamentally changes when you add more transistors in the task of designing an architecture?', 'start': 2282.279, 'duration': 7.488}, {'end': 2292.429, 'text': "Well, there's two constants, right?", 'start': 2289.787, 'duration': 2.642}, {'end': 2293.851, 'text': "One is people don't get smarter.", 'start': 2292.59, 'duration': 1.261}, {'end': 2300.015, 'text': "By the way, there's some science showing that we do get smarter because of nutrition or whatever.", 'start': 2296.089, 'duration': 3.926}, {'end': 2301.898, 'text': 'Sorry to bring that up.', 'start': 2301.277, 'duration': 0.621}, {'end': 2302.399, 'text': 'Plant effect.', 'start': 2301.919, 'duration': 0.48}, {'end': 2303.862, 'text': "Yes Yeah, I'm familiar with it.", 'start': 2302.58, 'duration': 1.282}, {'end': 2304.783, 'text': 'Nobody understands it.', 'start': 2303.902, 'duration': 0.881}, {'end': 2307.388, 'text': "Nobody knows if it's still going on.", 'start': 2304.823, 'duration': 2.565}], 'summary': "Increasing transistors doesn't make people smarter in architecture design.", 'duration': 25.109, 'max_score': 2282.279, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/Nb2tebYAaOA/pics/Nb2tebYAaOA2282279.jpg'}], 'start': 2022.226, 'title': "Quantum transistors and moore's law", 'summary': "Explores the potential of quantum transistors to be a million times smaller than modern transistors and the implications of moore's law on transistor shrinking, addressing challenges and strategies for computer designers, and the impact of increasing transistors on design teams.", 'chapters': [{'end': 2144.698, 'start': 2022.226, 'title': 'Quantum transistors and nanotechnology', 'summary': 'Discusses the potential for quantum transistors to be a million times smaller than modern transistors, with quantum computational research aiming to utilize quantum effects, while also addressing the challenges of manufacturing and controlling atomic-scale components for computer production.', 'duration': 122.472, 'highlights': ['Quantum transistors could be a million times smaller than modern transistors, with quantum computational research exploring the use of quantum effects', 'Manufacturing a computer would require placing 10 to the 23rd atoms together, posing challenges in terms of time and efficiency', 'The innovation stack encompasses equipment, optics, chemistry, physics, material science, and metallurgy, with thousands of technologies involved in the process']}, {'end': 2330.216, 'start': 2144.998, 'title': "Moore's law and transistor shrinking", 'summary': "Discusses the implications of moore's law on transistor shrinking, emphasizing the challenges and strategies for computer designers, and the impact of increasing transistors on design teams.", 'duration': 185.218, 'highlights': ['The next 10 or 20 years of shrinking is expected to happen, leading to a significant increase in the number of transistors every two or three years. Expectation of more transistors every two or three years, requiring a change in design and architecture strategies.', 'The challenge of coping with the influx of new transistors and designing collateral to take advantage of it is crucial for computer designers. Importance of designing collateral to take advantage of and cope with the increasing number of transistors.', 'The limitation of human intelligence and team size poses a fundamental constraint on dealing with the increasing number of transistors. Limitation of human intelligence and team size in managing the growth in transistors.']}], 'duration': 307.99, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/Nb2tebYAaOA/pics/Nb2tebYAaOA2022226.jpg', 'highlights': ['Quantum transistors could be a million times smaller than modern transistors, with quantum computational research exploring the use of quantum effects', 'The next 10 or 20 years of shrinking is expected to happen, leading to a significant increase in the number of transistors every two or three years. Expectation of more transistors every two or three years, requiring a change in design and architecture strategies.', 'The challenge of coping with the influx of new transistors and designing collateral to take advantage of it is crucial for computer designers. Importance of designing collateral to take advantage of and cope with the increasing number of transistors.', 'Manufacturing a computer would require placing 10 to the 23rd atoms together, posing challenges in terms of time and efficiency', 'The innovation stack encompasses equipment, optics, chemistry, physics, material science, and metallurgy, with thousands of technologies involved in the process', 'The limitation of human intelligence and team size poses a fundamental constraint on dealing with the increasing number of transistors. Limitation of human intelligence and team size in managing the growth in transistors.']}, {'end': 3010.3, 'segs': [{'end': 2401.542, 'src': 'embed', 'start': 2364.322, 'weight': 2, 'content': [{'end': 2369.465, 'text': 'So some algorithms run twice as fast on new computers, but a lot of algorithms are N squared.', 'start': 2364.322, 'duration': 5.143}, {'end': 2375.888, 'text': 'So, you know, a computer with twice as many transistors and it might take four times as long to run.', 'start': 2370.466, 'duration': 5.422}, {'end': 2378.41, 'text': 'So you have to refactor the software.', 'start': 2376.549, 'duration': 1.861}, {'end': 2382.672, 'text': "Like simply using faster computers to build bigger computers doesn't work.", 'start': 2379.33, 'duration': 3.342}, {'end': 2386.399, 'text': 'You have to think about all these things.', 'start': 2384.958, 'duration': 1.441}, {'end': 2401.542, 'text': "In terms of computing performance and the exciting possibility that more powerful computers bring is shrinking the thing we've just been talking about for you one of the biggest exciting possibilities of advancement in performance.", 'start': 2386.419, 'duration': 15.123}], 'summary': 'New computers can run algorithms twice as fast, but many are n squared, requiring software refactoring for performance improvement.', 'duration': 37.22, 'max_score': 2364.322, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/Nb2tebYAaOA/pics/Nb2tebYAaOA2364322.jpg'}, {'end': 2490.903, 'src': 'embed', 'start': 2455.213, 'weight': 1, 'content': [{'end': 2458.055, 'text': 'to the deeper kind of computation.', 'start': 2455.213, 'duration': 2.842}, {'end': 2463.94, 'text': 'And the data sets are getting so big that people are thinking of data as a topology problem.', 'start': 2458.815, 'duration': 5.125}, {'end': 2467.263, 'text': 'You know, data is organized in some immense shape.', 'start': 2464.36, 'duration': 2.903}, {'end': 2474.57, 'text': 'And then the computation, which sort of wants to be get data from immense shape and do some computation on it.', 'start': 2468.044, 'duration': 6.526}, {'end': 2481.076, 'text': 'So what computers have allowed people to do is have algorithms go much, much further.', 'start': 2475.37, 'duration': 5.706}, {'end': 2490.903, 'text': 'So that paper you referenced, the Sutton paper, they talked about, you know, like when AI started, it was apply rulesets to something.', 'start': 2482.517, 'duration': 8.386}], 'summary': 'Computers enable algorithms to process big data sets for deeper computation.', 'duration': 35.69, 'max_score': 2455.213, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/Nb2tebYAaOA/pics/Nb2tebYAaOA2455213.jpg'}, {'end': 2565.201, 'src': 'embed', 'start': 2533.931, 'weight': 6, 'content': [{'end': 2535.732, 'text': 'Yeah, the kind of computation that might I mean?', 'start': 2533.931, 'duration': 1.801}, {'end': 2539.333, 'text': 'I would argue that all of it is still a search, right?', 'start': 2535.732, 'duration': 3.601}, {'end': 2542.773, 'text': 'Just like you said, a topology problem of data sets.', 'start': 2540.013, 'duration': 2.76}, {'end': 2553.035, 'text': "you're searching the data sets for valuable data and also the actual optimization of neural networks is a kind of search for the.", 'start': 2542.773, 'duration': 10.262}, {'end': 2557.979, 'text': "I don't know, if you had looked at the interlayers of finding a cat, It's not a search.", 'start': 2553.035, 'duration': 4.944}, {'end': 2561.06, 'text': "It's a set of endless projections.", 'start': 2559.18, 'duration': 1.88}, {'end': 2565.201, 'text': "So you know a projection here's a shadow of this phone right?", 'start': 2561.1, 'duration': 4.101}], 'summary': 'Computation involves search for valuable data and optimizing neural networks, akin to topology problem of data sets.', 'duration': 31.27, 'max_score': 2533.931, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/Nb2tebYAaOA/pics/Nb2tebYAaOA2533931.jpg'}, {'end': 2709.039, 'src': 'embed', 'start': 2678.57, 'weight': 8, 'content': [{'end': 2684.133, 'text': 'Well, the operations continue to be add, subtract, load, store, compare, and branch.', 'start': 2678.57, 'duration': 5.563}, {'end': 2685.614, 'text': "It's remarkable.", 'start': 2684.714, 'duration': 0.9}, {'end': 2692.338, 'text': "So it's interesting that the building blocks of computers are transistors, and under that, atoms.", 'start': 2686.174, 'duration': 6.164}, {'end': 2697.601, 'text': "So you've got atoms, transistors, logic gates, computers, functional units of computers.", 'start': 2692.358, 'duration': 5.243}, {'end': 2709.039, 'text': 'The building blocks of mathematics at some level are things like adds and subtracts and multiplies, but the space mathematics can describe is,', 'start': 2698.393, 'duration': 10.646}], 'summary': 'Computer operations include add, subtract, load, store, compare, and branch, based on transistors and atoms.', 'duration': 30.469, 'max_score': 2678.57, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/Nb2tebYAaOA/pics/Nb2tebYAaOA2678570.jpg'}, {'end': 2828.343, 'src': 'embed', 'start': 2750.022, 'weight': 3, 'content': [{'end': 2754.623, 'text': "And then there's people who think about analog computing because you look in the brain and it seems to be more analog-ish.", 'start': 2750.022, 'duration': 4.601}, {'end': 2758.564, 'text': "You know, that maybe there's a way to do that more efficiently.", 'start': 2755.863, 'duration': 2.701}, {'end': 2762.605, 'text': 'But we have a million X on computation.', 'start': 2759.144, 'duration': 3.461}, {'end': 2765.866, 'text': "And I don't know the representation.", 'start': 2763.625, 'duration': 2.241}, {'end': 2775.572, 'text': "The relationship between computational, let's say, intensity, and ability to hit mathematical abstractions.", 'start': 2766.926, 'duration': 8.646}, {'end': 2786.659, 'text': "I don't know any way to describe that, but just like you saw in AI, you went from rule sets to simple search to complex search, to say, found search.", 'start': 2775.572, 'duration': 11.087}, {'end': 2790.321, 'text': 'Those are orders of magnitude more computation to do.', 'start': 2786.819, 'duration': 3.502}, {'end': 2798.864, 'text': 'And as we get the next two orders of magnitude, like a friend, Roger Godori said, like every order of magnitude changes the computation.', 'start': 2791.562, 'duration': 7.302}, {'end': 2802.665, 'text': 'Fundamentally changes what the computation is doing.', 'start': 2800.284, 'duration': 2.381}, {'end': 2807.946, 'text': 'Yeah Oh, you know, the expression, the difference in quantity is a difference in kind.', 'start': 2802.685, 'duration': 5.261}, {'end': 2814.268, 'text': 'You know, the difference between ant and anthill, right? Or neuron and brain.', 'start': 2809.487, 'duration': 4.781}, {'end': 2824.919, 'text': "You know, there's this indefinable place where the quantity changed the quality, right? And we've seen that happen in mathematics multiple times.", 'start': 2815.97, 'duration': 8.949}, {'end': 2828.343, 'text': "And, you know, my guess is it's going to keep happening.", 'start': 2824.999, 'duration': 3.344}], 'summary': 'Analog computing may be more efficient with a million x computation, with each order of magnitude fundamentally changing computation, impacting mathematical abstractions and ai advancements.', 'duration': 78.321, 'max_score': 2750.022, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/Nb2tebYAaOA/pics/Nb2tebYAaOA2750022.jpg'}, {'end': 2938.861, 'src': 'embed', 'start': 2918.189, 'weight': 0, 'content': [{'end': 2928.115, 'text': "If Moore's Law is going to continue, then your AI research should expect that to show up, and then you make a slightly different set of choices then.", 'start': 2918.189, 'duration': 9.926}, {'end': 2933.998, 'text': "We've hit the wall, nothing's going to happen, and from here it's just us rewriting algorithms.", 'start': 2928.715, 'duration': 5.283}, {'end': 2938.861, 'text': "That seems like a failed strategy for the last 30 years of Moore's Law's death.", 'start': 2935.198, 'duration': 3.663}], 'summary': "Moore's law's death has led to a failed ai research strategy for the last 30 years.", 'duration': 20.672, 'max_score': 2918.189, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/Nb2tebYAaOA/pics/Nb2tebYAaOA2918189.jpg'}, {'end': 2994.772, 'src': 'embed', 'start': 2965.528, 'weight': 4, 'content': [{'end': 2969.553, 'text': 'Well, the shrinking of the transistor is literally thousands of innovations.', 'start': 2965.528, 'duration': 4.025}, {'end': 2973.216, 'text': "So there's stacks of S-curves in there.", 'start': 2970.394, 'duration': 2.822}, {'end': 2980.281, 'text': "There's a whole bunch of S-curves just kind of running their course and being reinvented and new things.", 'start': 2973.276, 'duration': 7.005}, {'end': 2987.386, 'text': "The semiconductor fabricators and technologists have all announced what's called nanowires.", 'start': 2981.762, 'duration': 5.624}, {'end': 2994.772, 'text': "So they took a fan which had a gate around it and turned that into little wires so you have better control of that and they're smaller.", 'start': 2987.446, 'duration': 7.326}], 'summary': 'Semiconductor innovations have led to the development of nanowires for better transistor control and smaller size.', 'duration': 29.244, 'max_score': 2965.528, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/Nb2tebYAaOA/pics/Nb2tebYAaOA2965528.jpg'}], 'start': 2330.837, 'title': 'Computing performance and future directions', 'summary': 'Discusses the challenges of computing performance, evolution of ai computations, debate about ai as a form of search, fundamental operations of computer hardware, potential impact of quantum and analog computing, shifts in computation with each order of magnitude increase, and innovations in shrinking transistors and semiconductor technology.', 'chapters': [{'end': 2657.136, 'start': 2330.837, 'title': 'Computing performance and future directions', 'summary': 'Discusses the challenges of computing performance and the exciting potential of more powerful computers, as well as the evolution of ai computations from simple equations to data topology problems, and the ongoing debate about whether ai is fundamentally a form of search.', 'duration': 326.299, 'highlights': ['The evolution of AI computations from simple equations to data topology problems is discussed, highlighting the increasing complexity of AI algorithms and the shift towards viewing data as a topology problem. The chapter explains the progression of AI computations from simple equations to data topology problems, demonstrating the increasing complexity of AI algorithms and the changing perspective on data as a topology problem.', 'The challenges of computing performance and the exciting potential of more powerful computers are detailed, emphasizing the need to refactor software and the impact of faster computers on algorithm run times. The transcript outlines the challenges of computing performance and the potential of more powerful computers, stressing the necessity to refactor software and the implications of faster computers on algorithm run times.', 'The ongoing debate about whether AI is fundamentally a form of search is explored, with differing perspectives on the nature of AI computations and the concept of optimization as search. The chapter delves into the ongoing debate regarding the fundamental nature of AI as a form of search, presenting differing viewpoints on the nature of AI computations and the characterization of optimization as a form of search.']}, {'end': 2828.343, 'start': 2657.256, 'title': 'Evolution of computers and mathematical operations', 'summary': 'Discusses the fundamental operations of computer hardware and mathematical algorithms, highlighting the enduring nature of basic operations like adds and multiplies, the potential impact of quantum and analog computing, and the significant shifts in computation with each order of magnitude increase, impacting the computational intensity and ability to handle mathematical abstractions.', 'duration': 171.087, 'highlights': ['The enduring nature of basic operations like adds and multiplies in computer hardware, despite variations in algorithm and data types, is highlighted, indicating the fundamental nature of these operations.', 'The potential impact of quantum and analog computing on changing the fundamental nature of computation is mentioned, suggesting a potential radical shift in computational processes.', 'The significant impact of each order of magnitude increase in computation, as indicated by the statement that every order of magnitude fundamentally changes what the computation is doing, is emphasized, showcasing the transformative effect on computational intensity and the ability to handle mathematical abstractions.', 'The comparison of the impact of quantity on quality in mathematics, paralleled with the potential continual evolution in the field, is discussed, indicating the potential for ongoing significant shifts in mathematical operations and computational processes.']}, {'end': 3010.3, 'start': 2828.603, 'title': "Future of moore's law", 'summary': "Discusses the future of moore's law, emphasizing the ongoing innovations in shrinking transistors and the multiple s-curves in semiconductor technology, suggesting that moore's law is likely to continue influencing ai research and computational advancements.", 'duration': 181.697, 'highlights': ['The semiconductor fabricators have announced nanowires, which provide better control and smaller size, contributing to the ongoing innovations in shrinking transistors. The development of nanowires by semiconductor fabricators presents a significant innovation in shrinking transistors, indicating the continuous advancements in semiconductor technology.', "The ongoing innovations in shrinking transistors and the presence of multiple S-curves in semiconductor technology suggest that Moore's Law is likely to continue influencing AI research and computational advancements. The presence of multiple S-curves in semiconductor technology, along with the ongoing innovations in shrinking transistors, indicates the potential continuation of Moore's Law's influence on AI research and computational advancements.", "The discussion emphasizes that the semiconductor technologists have been reinventing and innovating numerous S-curves, including the development of nanowires and metallurgical advancements, contributing to the potential continuation of Moore's Law. The emphasis on the reinvention and innovation of numerous S-curves in semiconductor technology, including the development of nanowires and metallurgical advancements, signifies the potential for Moore's Law to continue influencing computational advancements."]}], 'duration': 679.463, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/Nb2tebYAaOA/pics/Nb2tebYAaOA2330837.jpg', 'highlights': ["The ongoing innovations in shrinking transistors and the presence of multiple S-curves in semiconductor technology suggest that Moore's Law is likely to continue influencing AI research and computational advancements.", 'The evolution of AI computations from simple equations to data topology problems is discussed, highlighting the increasing complexity of AI algorithms and the shift towards viewing data as a topology problem.', 'The challenges of computing performance and the exciting potential of more powerful computers are detailed, emphasizing the need to refactor software and the impact of faster computers on algorithm run times.', 'The comparison of the impact of quantity on quality in mathematics, paralleled with the potential continual evolution in the field, is discussed, indicating the potential for ongoing significant shifts in mathematical operations and computational processes.', 'The semiconductor fabricators have announced nanowires, which provide better control and smaller size, contributing to the ongoing innovations in shrinking transistors.', 'The significant impact of each order of magnitude increase in computation, as indicated by the statement that every order of magnitude fundamentally changes what the computation is doing, is emphasized, showcasing the transformative effect on computational intensity and the ability to handle mathematical abstractions.', 'The ongoing debate about whether AI is fundamentally a form of search is explored, with differing perspectives on the nature of AI computations and the concept of optimization as search.', 'The potential impact of quantum and analog computing on changing the fundamental nature of computation is mentioned, suggesting a potential radical shift in computational processes.', 'The enduring nature of basic operations like adds and multiplies in computer hardware, despite variations in algorithm and data types, is highlighted, indicating the fundamental nature of these operations.']}, {'end': 3768.948, 'segs': [{'end': 3086.713, 'src': 'embed', 'start': 3011.02, 'weight': 0, 'content': [{'end': 3015.503, 'text': "Your sense is that we're going to get a lot if this innovation from just that shrinking.", 'start': 3011.02, 'duration': 4.483}, {'end': 3017.684, 'text': "Yeah, like a factor of 100, it's a lot.", 'start': 3016.663, 'duration': 1.021}, {'end': 3021.612, 'text': "Yeah, I would say that's incredible.", 'start': 3019.471, 'duration': 2.141}, {'end': 3023.632, 'text': "And it's totally unknown.", 'start': 3021.992, 'duration': 1.64}, {'end': 3025.033, 'text': "It's only 10 or 15 years.", 'start': 3023.732, 'duration': 1.301}, {'end': 3026.433, 'text': "Now you're smart and you might know,", 'start': 3025.173, 'duration': 1.26}, {'end': 3034.075, 'text': "but to me it's totally unpredictable of what that 100x would bring in terms of the nature of the computation that people would be.", 'start': 3026.433, 'duration': 7.642}, {'end': 3041.617, 'text': "Yeah, are you familiar with Bell's Law? So for a long time it was mainframes, minis, workstation, PC, mobile.", 'start': 3034.415, 'duration': 7.202}, {'end': 3045.058, 'text': "Moore's Law drove faster, smaller computers.", 'start': 3042.597, 'duration': 2.461}, {'end': 3053.28, 'text': "Right And then when we were thinking about Moore's law, uh, Roger Godori said every 10 X generates a new computation.", 'start': 3045.936, 'duration': 7.344}, {'end': 3059.704, 'text': 'So scalar vector matrix, topological computation, right?', 'start': 3053.32, 'duration': 6.384}, {'end': 3068.949, 'text': 'And if you go look at the industry trends, there was mainframes and then mini computers and then PCs, and then the internet took off,', 'start': 3061.187, 'duration': 7.762}, {'end': 3074.15, 'text': "and then we got mobile devices and now we're building 5G wireless with one millisecond latency.", 'start': 3068.949, 'duration': 5.201}, {'end': 3080.931, 'text': 'And people are starting to think about the smart world where everything knows you, recognizes you.', 'start': 3075.05, 'duration': 5.881}, {'end': 3086.713, 'text': 'Like the transformations are gonna be, like unpredictable.', 'start': 3081.851, 'duration': 4.862}], 'summary': "Anticipating 100x innovation, unpredictable transformations in computation, influenced by moore's and bell's law.", 'duration': 75.693, 'max_score': 3011.02, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/Nb2tebYAaOA/pics/Nb2tebYAaOA3011020.jpg'}, {'end': 3250.908, 'src': 'embed', 'start': 3197.077, 'weight': 6, 'content': [{'end': 3203.879, 'text': "You can't deny the fact that these tools are changing our world.", 'start': 3197.077, 'duration': 6.802}, {'end': 3205.36, 'text': "That's right.", 'start': 3204.44, 'duration': 0.92}, {'end': 3208.161, 'text': "Do you think it's changing for the better?", 'start': 3205.38, 'duration': 2.781}, {'end': 3211.322, 'text': 'I read this thing recently.', 'start': 3208.181, 'duration': 3.141}, {'end': 3218.184, 'text': 'it said the two disciplines with the highest GRE scores in college are physics and philosophy.', 'start': 3211.322, 'duration': 6.862}, {'end': 3222.645, 'text': "They're both trying to answer the question why is there anything?", 'start': 3219.844, 'duration': 2.801}, {'end': 3231.527, 'text': 'And the philosophers are on the kind of theological side and the physicists are obviously on the material side.', 'start': 3224.065, 'duration': 7.462}, {'end': 3236.869, 'text': "And there's 100 billion galaxies with 100 billion stars.", 'start': 3232.668, 'duration': 4.201}, {'end': 3239.769, 'text': 'It seems, well, repetitive at best.', 'start': 3236.989, 'duration': 2.78}, {'end': 3245.491, 'text': "So there's on our way to 10 billion people.", 'start': 3241.05, 'duration': 4.441}, {'end': 3249.152, 'text': "I mean, it's hard to say what it's all for, if that's what you're asking.", 'start': 3245.511, 'duration': 3.641}, {'end': 3250.908, 'text': 'Yeah, I guess I am.', 'start': 3249.628, 'duration': 1.28}], 'summary': 'Tools are changing the world, questioning existence and purpose, with 100 billion galaxies and 10 billion people on the way.', 'duration': 53.831, 'max_score': 3197.077, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/Nb2tebYAaOA/pics/Nb2tebYAaOA3197077.jpg'}, {'end': 3321.817, 'src': 'embed', 'start': 3298.158, 'weight': 7, 'content': [{'end': 3308.606, 'text': 'Now, recently, the computation lets you do mathematical computations that are sophisticated enough that nobody understands how the answers came out.', 'start': 3298.158, 'duration': 10.448}, {'end': 3310.627, 'text': 'Machine learning.', 'start': 3310.047, 'duration': 0.58}, {'end': 3311.388, 'text': 'Machine learning.', 'start': 3310.788, 'duration': 0.6}, {'end': 3316.092, 'text': 'It used to be you get data set, you guess at a function.', 'start': 3312.029, 'duration': 4.063}, {'end': 3321.817, 'text': "The function is considered physics if it's predictive of new functions, new data sets.", 'start': 3316.813, 'duration': 5.004}], 'summary': 'Machine learning allows sophisticated mathematical computations for predicting new data sets.', 'duration': 23.659, 'max_score': 3298.158, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/Nb2tebYAaOA/pics/Nb2tebYAaOA3298158.jpg'}, {'end': 3530.079, 'src': 'embed', 'start': 3500.627, 'weight': 4, 'content': [{'end': 3505.53, 'text': 'Yeah, like what consciousness love?', 'start': 3500.627, 'duration': 4.903}, {'end': 3509.612, 'text': 'things that are these deeply human things that seems to emerge from our brain.', 'start': 3505.53, 'duration': 4.082}, {'end': 3518.237, 'text': "is that something that we'll be able to make encode in chips that get faster and faster and faster and faster?", 'start': 3509.612, 'duration': 8.625}, {'end': 3519.698, 'text': "That's like a 10-hour conversation.", 'start': 3518.257, 'duration': 1.441}, {'end': 3520.973, 'text': 'Nobody really knows.', 'start': 3520.212, 'duration': 0.761}, {'end': 3524.695, 'text': 'Can you summarize it in a couple of sentences? A couple of words.', 'start': 3521.013, 'duration': 3.682}, {'end': 3530.079, 'text': 'Many people have observed that organisms run at lots of different levels.', 'start': 3525.396, 'duration': 4.683}], 'summary': 'Exploring the potential of encoding consciousness in chips for faster processing remains uncertain.', 'duration': 29.452, 'max_score': 3500.627, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/Nb2tebYAaOA/pics/Nb2tebYAaOA3500627.jpg'}], 'start': 3011.02, 'title': 'Future technology and philosophical reflections', 'summary': "Discusses potential 100x innovation in computation, evolution of computing trends, and a hardware architect's philosophical musings on technology's impact on humanity, artificial intelligence, and the universe as a computer.", 'chapters': [{'end': 3140.734, 'start': 3011.02, 'title': 'Future of computation and technology', 'summary': 'Discusses the unpredictability of future technological advancements, with a potential 100x innovation in computation and the evolution of computing trends from mainframes to 5g wireless and smart devices.', 'duration': 129.714, 'highlights': ['The potential for a 100x innovation in computation is highlighted, indicating a significant advancement in technology.', 'The evolution of computing trends from mainframes to 5G wireless and smart devices is discussed, emphasizing the unpredictable nature of technological transformations.', "The concept of Bell's Law and its impact on driving new computations is mentioned, providing insight into the evolution of technology and computing.", 'The unpredictability of future technological advancements and the changing nature of communication and technology are discussed, highlighting the fast-paced evolution of the digital landscape.']}, {'end': 3768.948, 'start': 3140.734, 'title': "Future hardware architect's philosophical reflections", 'summary': 'Delves into the philosophical musings of a hardware architect regarding the impact of technology on humanity, the nature of computation, and the potential for creating artificial intelligence, while questioning the universe as a computer. it also touches on the limitations of human understanding in the context of physics and the universe.', 'duration': 628.214, 'highlights': ['The chapter discusses the philosophical reflections of a hardware architect on the impact of technology on humanity, the nature of computation, and the potential for creating artificial intelligence, while questioning the universe as a computer.', 'The conversation encompasses the unpredictability of actions and the parallel occurrences in the world of technology, with millions of people working in this sphere.', 'The discussion delves into the changes brought about by tools and the uncertainty of whether these changes are for the better or not, touching on the interconnectedness of physics, philosophy, and the complexity of the universe.', 'The conversation also explores the advancements in computation, particularly in the realm of machine learning, and its ability to generate results that may not be completely mathematically describable.', 'The chapter questions the potential emergence of consciousness and love in artificial intelligence and ponders on whether human understanding of the universe is limited by computational constraints.', 'The conversation ends on a contemplative note, discussing the limitations of human understanding in the context of physics and the universe and the ongoing struggles of both physicists and philosophers in comprehending the fundamental nature of the universe.']}], 'duration': 757.928, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/Nb2tebYAaOA/pics/Nb2tebYAaOA3011020.jpg', 'highlights': ['The potential for a 100x innovation in computation is highlighted, indicating a significant advancement in technology.', 'The evolution of computing trends from mainframes to 5G wireless and smart devices is discussed, emphasizing the unpredictable nature of technological transformations.', "The concept of Bell's Law and its impact on driving new computations is mentioned, providing insight into the evolution of technology and computing.", 'The unpredictability of future technological advancements and the changing nature of communication and technology are discussed, highlighting the fast-paced evolution of the digital landscape.', 'The chapter discusses the philosophical reflections of a hardware architect on the impact of technology on humanity, the nature of computation, and the potential for creating artificial intelligence, while questioning the universe as a computer.', 'The conversation encompasses the unpredictability of actions and the parallel occurrences in the world of technology, with millions of people working in this sphere.', 'The discussion delves into the changes brought about by tools and the uncertainty of whether these changes are for the better or not, touching on the interconnectedness of physics, philosophy, and the complexity of the universe.', 'The conversation also explores the advancements in computation, particularly in the realm of machine learning, and its ability to generate results that may not be completely mathematically describable.', 'The chapter questions the potential emergence of consciousness and love in artificial intelligence and ponders on whether human understanding of the universe is limited by computational constraints.', 'The conversation ends on a contemplative note, discussing the limitations of human understanding in the context of physics and the universe and the ongoing struggles of both physicists and philosophers in comprehending the fundamental nature of the universe.']}, {'end': 4266.976, 'segs': [{'end': 3838.104, 'src': 'embed', 'start': 3790.882, 'weight': 0, 'content': [{'end': 3801.308, 'text': "Do you see Moore's Law more broadly in the sense that technology of all kinds has a way of stacking S-curves on top of each other,", 'start': 3790.882, 'duration': 10.426}, {'end': 3804.53, 'text': "where it'll be exponential, and then we'll see all kinds of.", 'start': 3801.308, 'duration': 3.222}, {'end': 3809.373, 'text': "What does an exponential of a million mean? That's a pretty amazing number.", 'start': 3804.53, 'duration': 4.843}, {'end': 3811.534, 'text': "And that's just for a local little piece of silicon.", 'start': 3809.473, 'duration': 2.061}, {'end': 3822.9, 'text': "Now let's imagine you, say, decided to get a thousand tons of silicon to collaborate in one computer at a million times the density.", 'start': 3812.234, 'duration': 10.666}, {'end': 3832.558, 'text': "Like now you're talking, I don't know, 10 to the 20th more computation power than our current already unbelievably fast computers.", 'start': 3824.389, 'duration': 8.169}, {'end': 3835.482, 'text': "Like nobody knows what that's going to mean.", 'start': 3834.06, 'duration': 1.422}, {'end': 3838.104, 'text': 'You know, the sci-fi guys call it, you know, computronium.', 'start': 3835.582, 'duration': 2.522}], 'summary': 'Technology stacking s-curves can lead to exponential growth, such as a potential 10^20 increase in computation power with dense silicon collaboration.', 'duration': 47.222, 'max_score': 3790.882, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/Nb2tebYAaOA/pics/Nb2tebYAaOA3790882.jpg'}, {'end': 3911.1, 'src': 'embed', 'start': 3857.666, 'weight': 1, 'content': [{'end': 3867.113, 'text': 'so computers are made out of silicon and minor amounts of metals, and this and that none of those things cost any money.', 'start': 3857.666, 'duration': 9.447}, {'end': 3868.513, 'text': "There's plenty of sand.", 'start': 3867.513, 'duration': 1}, {'end': 3873.616, 'text': 'You could just turn the beach and a little bit ocean water into computers.', 'start': 3868.834, 'duration': 4.782}, {'end': 3876.177, 'text': 'All the cost is in the equipment to do it.', 'start': 3874.116, 'duration': 2.061}, {'end': 3881.78, 'text': 'The trend on equipment is, once you figure out how to build the equipment, the trend of cost is zero.', 'start': 3876.858, 'duration': 4.922}, {'end': 3888.684, 'text': 'Elon said, first you figure out what configuration you want the atoms in, and then how to put them there.', 'start': 3881.86, 'duration': 6.824}, {'end': 3896.189, 'text': 'His great insight is people are how constrained.', 'start': 3892.746, 'duration': 3.443}, {'end': 3897.21, 'text': 'I have this thing.', 'start': 3896.549, 'duration': 0.661}, {'end': 3898.15, 'text': 'I know how it works.', 'start': 3897.25, 'duration': 0.9}, {'end': 3906.637, 'text': 'And then little tweaks to that will generate something as opposed to what do I actually want and then figure out how to build it.', 'start': 3898.791, 'duration': 7.846}, {'end': 3908.418, 'text': "It's a very different mindset.", 'start': 3907.077, 'duration': 1.341}, {'end': 3911.1, 'text': 'And almost nobody has it, obviously.', 'start': 3909.319, 'duration': 1.781}], 'summary': 'Computers can be made from sand and ocean water, with cost in equipment. elon musk emphasizes on innovative mindset.', 'duration': 53.434, 'max_score': 3857.666, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/Nb2tebYAaOA/pics/Nb2tebYAaOA3857666.jpg'}, {'end': 4040.791, 'src': 'embed', 'start': 4016.696, 'weight': 4, 'content': [{'end': 4024.341, 'text': "So the beautiful thing about the human vision system and our entire brain around the whole thing is we're able to fill in the gaps.", 'start': 4016.696, 'duration': 7.645}, {'end': 4027.703, 'text': "It's not just about perfectly detecting cars.", 'start': 4025.582, 'duration': 2.121}, {'end': 4029.904, 'text': "It's inferring the occluded cars.", 'start': 4028.283, 'duration': 1.621}, {'end': 4032.406, 'text': "It's trying to, it's understanding the physics.", 'start': 4029.965, 'duration': 2.441}, {'end': 4033.827, 'text': "I think that's mostly a data problem.", 'start': 4032.426, 'duration': 1.401}, {'end': 4040.791, 'text': 'So you think what data would compute with improvement in computation, with improvement in collection?', 'start': 4034.608, 'duration': 6.183}], 'summary': 'Human vision system fills in gaps, infers occluded cars, and understands physics, mostly a data problem.', 'duration': 24.095, 'max_score': 4016.696, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/Nb2tebYAaOA/pics/Nb2tebYAaOA4016696.jpg'}, {'end': 4106.694, 'src': 'embed', 'start': 4076.384, 'weight': 5, 'content': [{'end': 4079.325, 'text': 'You can place every object really thoroughly.', 'start': 4076.384, 'duration': 2.941}, {'end': 4084.521, 'text': 'You can calculate trajectories of things really thoroughly.', 'start': 4081.578, 'duration': 2.943}, {'end': 4091.966, 'text': 'But everything you said about really thoroughly has a different degree of difficulty.', 'start': 4086.562, 'duration': 5.404}, {'end': 4099.81, 'text': 'You could say at some point, computer autonomous systems will be way better at things that humans are lousy at.', 'start': 4093.227, 'duration': 6.583}, {'end': 4101.953, 'text': "They'll be better at attention.", 'start': 4100.692, 'duration': 1.261}, {'end': 4106.694, 'text': "They'll always remember there was a pothole in the road that humans keep forgetting about.", 'start': 4102.473, 'duration': 4.221}], 'summary': 'Computer autonomous systems will excel in attention and memory, surpassing human abilities.', 'duration': 30.31, 'max_score': 4076.384, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/Nb2tebYAaOA/pics/Nb2tebYAaOA4076384.jpg'}], 'start': 3768.948, 'title': 'The future of technology', 'summary': "Discusses the potential implications of moore's law and exponential growth in technology, including the possibility of a million times increase in computation power, zero cost trend of equipment, and vehicle autonomy challenges, with a prediction of exponential improvement in the long run.", 'chapters': [{'end': 3881.78, 'start': 3768.948, 'title': "Future of moore's law and exponential growth", 'summary': "Discusses the potential implications of moore's law and exponential growth in technology, including the possibility of a million times increase in computation power and the cost trend of equipment being zero.", 'duration': 112.832, 'highlights': ["Moore's Law suggests the exponential growth of technology will continue indefinitely, leading to a million times increase in computation power.", 'The cost trend of equipment for building computers is zero once the manufacturing process is established, as it primarily involves readily available materials like silicon and minor metals.', 'The potential implications of a thousand tons of silicon collaborating in one computer at a million times the density, leading to 10 to the 20th more computation power than current computers.']}, {'end': 4266.976, 'start': 3881.86, 'title': 'Autonomy in vehicles', 'summary': 'Discusses the challenges and potential of vehicle autonomy, including the different mindset required for innovation, the simplicity of driving for humans, the complexities of computer vision, and the potential for autonomous systems to surpass human capabilities, with a prediction of exponential improvement in the long run.', 'duration': 385.116, 'highlights': ["Elon Musk's insight on innovation: focusing on what one wants and then figuring out how to build it. Elon's approach is to focus on what one wants and then figure out how to build it, rather than being constrained by existing knowledge, highlighting the need for a different mindset in innovation.", 'Challenges of computer vision in vehicle autonomy: understanding scenes and inferring occluded objects. Computer vision in vehicle autonomy involves understanding scenes and inferring occluded objects, representing a complex data problem that requires advanced capabilities to avoid errors and inaccuracies.', 'Potential for autonomous systems to surpass human capabilities: excelling in attention and memory, and maximizing the given information. Autonomous systems have the potential to surpass human capabilities in attention, memory, and maximizing the given information, indicating the possibility of exponential improvement in the long run.', 'Complexities of driving and human behavior: the intricate nature of driving and the challenges involving human behavior. Driving is highlighted as a complex activity, requiring the understanding of human behavior and the intricate nature of driving, challenging the common perception of its simplicity.']}], 'duration': 498.028, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/Nb2tebYAaOA/pics/Nb2tebYAaOA3768948.jpg', 'highlights': ["Moore's Law predicts a million times increase in computation power.", 'The cost trend of equipment for building computers is zero once the manufacturing process is established.', 'A thousand tons of silicon collaborating in one computer at a million times the density leads to 10 to the 20th more computation power than current computers.', "Elon Musk's approach to innovation focuses on what one wants and then figuring out how to build it.", 'Challenges of computer vision in vehicle autonomy involve understanding scenes and inferring occluded objects.', 'Autonomous systems have the potential to surpass human capabilities in attention, memory, and maximizing the given information.']}, {'end': 4564.136, 'segs': [{'end': 4357.179, 'src': 'embed', 'start': 4268.838, 'weight': 0, 'content': [{'end': 4275.464, 'text': "I'm hoping that there's a lot of things that humans aren't good at that machines are definitely good at like you said, attention and things like that.", 'start': 4268.838, 'duration': 6.626}, {'end': 4284.192, 'text': "Well, they'll be so much better that the overall picture of safety and autonomy will be obviously cars will be safer, even if they're not as good.", 'start': 4275.644, 'duration': 8.548}, {'end': 4285.833, 'text': "I'm a big believer in safety.", 'start': 4284.332, 'duration': 1.501}, {'end': 4292.72, 'text': "I mean, there are already the current safety systems like cruise control that doesn't let you run into people and lane keeping.", 'start': 4285.853, 'duration': 6.867}, {'end': 4301.503, 'text': 'Right There are so many features that you just look at the Pareto of accidents and knocking off like 80% of them is super doable.', 'start': 4292.74, 'duration': 8.763}, {'end': 4308.026, 'text': 'Just to linger on the autopilot team and the efforts there.', 'start': 4302.444, 'duration': 5.582}, {'end': 4314.309, 'text': "it seems to be that there's a very intense scrutiny by the media and the public in terms of safety,", 'start': 4308.026, 'duration': 6.283}, {'end': 4317.27, 'text': 'the pressure the bar put before autonomous vehicles.', 'start': 4314.309, 'duration': 2.961}, {'end': 4327.194, 'text': 'As a person there, working on the hardware and trying to build a system that builds a safe vehicle and so on.', 'start': 4318.05, 'duration': 9.144}, {'end': 4328.775, 'text': 'what was your sense about that pressure??', 'start': 4327.194, 'duration': 1.581}, {'end': 4329.835, 'text': 'Is it unfair??', 'start': 4328.995, 'duration': 0.84}, {'end': 4331.616, 'text': 'Is it expected of new technology??', 'start': 4329.895, 'duration': 1.721}, {'end': 4333.137, 'text': 'Yeah, it seems reasonable.', 'start': 4331.636, 'duration': 1.501}, {'end': 4334.137, 'text': 'I was interested.', 'start': 4333.577, 'duration': 0.56}, {'end': 4342.975, 'text': 'I talked to both American and European regulators and I was worried that the regulations would write into the rules.', 'start': 4334.177, 'duration': 8.798}, {'end': 4348.937, 'text': 'technology solutions like modern brake systems imply hydraulic brakes.', 'start': 4342.975, 'duration': 5.962}, {'end': 4357.179, 'text': 'So if you read the regulations, to meet the letter of the law for brakes it sort of has to be hydraulic right?', 'start': 4350.037, 'duration': 7.142}], 'summary': 'Machines can improve safety, making cars safer by addressing 80% of accidents. intense scrutiny on safety for autonomous vehicles is expected and reasonable.', 'duration': 88.341, 'max_score': 4268.838, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/Nb2tebYAaOA/pics/Nb2tebYAaOA4268838.jpg'}, {'end': 4450.039, 'src': 'embed', 'start': 4404.355, 'weight': 1, 'content': [{'end': 4410.18, 'text': "So they're safe and safer than people seemed, since the goal is to be 10X safer than people.", 'start': 4404.355, 'duration': 5.825}, {'end': 4418.107, 'text': 'having the bar to be safer than people and scrutinizing accidents seems, philosophically, you know, correct.', 'start': 4410.18, 'duration': 7.927}, {'end': 4420.229, 'text': "So I think that's a good thing.", 'start': 4419.269, 'duration': 0.96}, {'end': 4421.751, 'text': 'What are.', 'start': 4421.31, 'duration': 0.441}, {'end': 4433.001, 'text': "It's different than the things that you worked at Intel, AMD, Apple with autopilot chip design and hardware design.", 'start': 4423.533, 'duration': 9.468}, {'end': 4438.966, 'text': 'What are interesting or challenging aspects of building this specialized kind of computing system in the automotive space?', 'start': 4433.441, 'duration': 5.525}, {'end': 4442.717, 'text': "I mean there's two tricks to building an automotive computer.", 'start': 4440.276, 'duration': 2.441}, {'end': 4450.039, 'text': 'One is the software team, the machine learning team, is developing algorithms that are changing fast.', 'start': 4442.817, 'duration': 7.222}], 'summary': 'The goal is to make automotive systems 10x safer than people, with a focus on specialized computing and fast-changing algorithms.', 'duration': 45.684, 'max_score': 4404.355, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/Nb2tebYAaOA/pics/Nb2tebYAaOA4404355.jpg'}, {'end': 4546.544, 'src': 'embed', 'start': 4522.139, 'weight': 2, 'content': [{'end': 4527.946, 'text': 'so that the new algorithm is so much more effective that you would have been better off on a GPU.', 'start': 4522.139, 'duration': 5.807}, {'end': 4528.947, 'text': 'So there is a tension there.', 'start': 4527.966, 'duration': 0.981}, {'end': 4538.456, 'text': "To build a good computer for an application like automotive, there's all kinds of sensor inputs and safety processors and a bunch of stuff.", 'start': 4529.989, 'duration': 8.467}, {'end': 4541.619, 'text': "One of Elon's goals is to make it super affordable.", 'start': 4539.217, 'duration': 2.402}, {'end': 4544.322, 'text': 'Every car gets an autopilot computer.', 'start': 4542.64, 'duration': 1.682}, {'end': 4546.544, 'text': 'Some of the recent startups you look at.', 'start': 4545.042, 'duration': 1.502}], 'summary': 'New algorithm is more effective on gpu. elon aims for affordable autopilot computer for every car.', 'duration': 24.405, 'max_score': 4522.139, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/Nb2tebYAaOA/pics/Nb2tebYAaOA4522139.jpg'}], 'start': 4268.838, 'title': 'Autonomous vehicle safety', 'summary': 'Discusses the potential for machines to enhance safety in vehicles, particularly autonomous cars, with a focus on reducing accidents, making autonomous systems 10x safer than human drivers, and the challenges of building specialized computing systems for automotive applications.', 'chapters': [{'end': 4357.179, 'start': 4268.838, 'title': 'Autonomy and safety in vehicles', 'summary': 'Discusses the potential for machines to enhance safety in vehicles, particularly autonomous cars, with a focus on the scrutiny and pressure surrounding the development of these technologies, and the belief in the ability to significantly reduce accidents by implementing modern safety features.', 'duration': 88.341, 'highlights': ["The overall picture of safety and autonomy will be obviously cars will be safer, even if they're not as good.", 'Implementing modern safety features like cruise control and lane-keeping can potentially reduce around 80% of accidents.', 'There is intense scrutiny by the media and the public in terms of safety, putting pressure on the development of autonomous vehicles.', 'The regulations may pose challenges, as they tend to favor traditional technology solutions, such as hydraulic brakes, potentially hindering the adoption of modern safety systems.']}, {'end': 4564.136, 'start': 4357.92, 'title': 'Autonomous driving and automotive computing', 'summary': "Discusses the regulator's interest in autonomous driving use cases and the goal of making autonomous systems 10x safer than human drivers. it also explores the challenges of building specialized computing systems for automotive applications, emphasizing the need for affordable and adaptable solutions.", 'duration': 206.216, 'highlights': ["The regulator's interest in autonomous driving use cases, with a focus on safety and injury data, reflects the goal of making autonomous systems 10X safer than human drivers.", 'Challenges in building specialized computing systems for automotive applications, including the rapid evolution of algorithms and the need to balance specialization with adaptability for cost-effective solutions.', "Elon Musk's goal of making autonomous driving systems super affordable, aiming to include an autopilot computer in every car regardless of the purchase of autonomous driving features."]}], 'duration': 295.298, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/Nb2tebYAaOA/pics/Nb2tebYAaOA4268838.jpg', 'highlights': ['Implementing modern safety features like cruise control and lane-keeping can potentially reduce around 80% of accidents.', "The regulator's interest in autonomous driving use cases reflects the goal of making autonomous systems 10X safer than human drivers.", "Elon Musk's goal of making autonomous driving systems super affordable, aiming to include an autopilot computer in every car regardless of the purchase of autonomous driving features.", 'There is intense scrutiny by the media and the public in terms of safety, putting pressure on the development of autonomous vehicles.', 'Challenges in building specialized computing systems for automotive applications, including the rapid evolution of algorithms and the need to balance specialization with adaptability for cost-effective solutions.', 'The regulations may pose challenges, as they tend to favor traditional technology solutions, potentially hindering the adoption of modern safety systems.', "The overall picture of safety and autonomy will be obviously cars will be safer, even if they're not as good."]}, {'end': 5674.505, 'segs': [{'end': 4666.062, 'src': 'embed', 'start': 4643.777, 'weight': 4, 'content': [{'end': 4652.359, 'text': "In engineers, engineering is complicated enough that you have to learn a lot of skills, and then a lot of what you do is then craftsman's work,", 'start': 4643.777, 'duration': 8.582}, {'end': 4652.839, 'text': 'which is fun.', 'start': 4652.359, 'duration': 0.48}, {'end': 4657.84, 'text': 'Autonomous driving building a very resource-constrained computer.', 'start': 4653.479, 'duration': 4.361}, {'end': 4661.121, 'text': 'so a computer has to be cheap enough to put in every single car.', 'start': 4657.84, 'duration': 3.281}, {'end': 4664.802, 'text': "that essentially boils down to craftsman's work.", 'start': 4661.121, 'duration': 3.681}, {'end': 4666.062, 'text': "It's engineering.", 'start': 4665.142, 'duration': 0.92}], 'summary': "Engineering requires diverse skills and involves craftsman's work, like building resource-constrained computers for autonomous driving.", 'duration': 22.285, 'max_score': 4643.777, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/Nb2tebYAaOA/pics/Nb2tebYAaOA4643777.jpg'}, {'end': 4744.078, 'src': 'embed', 'start': 4717.082, 'weight': 7, 'content': [{'end': 4721.684, 'text': 'By the way, like factory work, like lots of people think factory work is road assembly stuff.', 'start': 4717.082, 'duration': 4.602}, {'end': 4723.685, 'text': "I've been on the assembly line.", 'start': 4722.485, 'duration': 1.2}, {'end': 4726.187, 'text': 'Like the people who work there really like it.', 'start': 4724.206, 'duration': 1.981}, {'end': 4727.287, 'text': "It's a really great job.", 'start': 4726.247, 'duration': 1.04}, {'end': 4728.708, 'text': "It's really complicated.", 'start': 4727.808, 'duration': 0.9}, {'end': 4730.849, 'text': 'Putting cars together is hard right?', 'start': 4728.728, 'duration': 2.121}, {'end': 4736.272, 'text': 'And the car is moving and the parts are moving and sometimes the parts are damaged and you have to coordinate,', 'start': 4730.929, 'duration': 5.343}, {'end': 4738.513, 'text': 'putting all the stuff together and people are good at it.', 'start': 4736.272, 'duration': 2.241}, {'end': 4739.734, 'text': "They're good at it.", 'start': 4739.033, 'duration': 0.701}, {'end': 4744.078, 'text': 'And I remember one day I went to work and the line was shut down for some reason.', 'start': 4740.394, 'duration': 3.684}], 'summary': 'Factory work involves complicated tasks like putting cars together, where workers excel, but face challenges like line shutdowns.', 'duration': 26.996, 'max_score': 4717.082, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/Nb2tebYAaOA/pics/Nb2tebYAaOA4717082.jpg'}, {'end': 4893.211, 'src': 'embed', 'start': 4866.035, 'weight': 1, 'content': [{'end': 4870.017, 'text': 'You think you have an understanding about what first principles of something is,', 'start': 4866.035, 'duration': 3.982}, {'end': 4875.038, 'text': "and then you talk to Elon about it and you didn't scratch the surface.", 'start': 4870.017, 'duration': 5.021}, {'end': 4879.22, 'text': "He has a deep belief that no matter what you do, it's a local maximum.", 'start': 4875.519, 'duration': 3.701}, {'end': 4883.447, 'text': 'Right And I had a friend, he invented a better electric motor.', 'start': 4880.686, 'duration': 2.761}, {'end': 4886.588, 'text': 'And it was like a lot better than what we were using.', 'start': 4884.367, 'duration': 2.221}, {'end': 4891.83, 'text': "And one day he came by, he said, you know, I'm a little disappointed because, you know, this is really great.", 'start': 4887.008, 'duration': 4.822}, {'end': 4893.211, 'text': "And you didn't seem that impressed.", 'start': 4891.85, 'duration': 1.361}], 'summary': 'Elon musk emphasizes reaching beyond local maximums in innovation and improvement.', 'duration': 27.176, 'max_score': 4866.035, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/Nb2tebYAaOA/pics/Nb2tebYAaOA4866035.jpg'}, {'end': 4938.575, 'src': 'embed', 'start': 4908.318, 'weight': 2, 'content': [{'end': 4911.901, 'text': "that's both innovative and, let's say, craftsman's work on the current thing.", 'start': 4908.318, 'duration': 3.583}, {'end': 4915.203, 'text': "it's really satisfying, it's good and that's cool.", 'start': 4911.901, 'duration': 3.302}, {'end': 4919.125, 'text': 'and then elon was good, taking everything apart.', 'start': 4915.203, 'duration': 3.922}, {'end': 4921.767, 'text': "like what's the deep first principle?", 'start': 4919.125, 'duration': 2.642}, {'end': 4923.008, 'text': "oh no, what's really.", 'start': 4921.767, 'duration': 1.241}, {'end': 4923.988, 'text': "no, what's really?", 'start': 4923.008, 'duration': 0.98}, {'end': 4932.873, 'text': 'you know, you know That ability to look at it without assumptions and how constraints is super wild.', 'start': 4923.988, 'duration': 8.885}, {'end': 4938.575, 'text': 'He built a rocket ship and electric car and everything.', 'start': 4934.473, 'duration': 4.102}], 'summary': "Elon musk's ability to innovate and think from first principles led to the creation of a rocket ship, electric car, and more.", 'duration': 30.257, 'max_score': 4908.318, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/Nb2tebYAaOA/pics/Nb2tebYAaOA4908318.jpg'}, {'end': 5138.203, 'src': 'embed', 'start': 5108.596, 'weight': 3, 'content': [{'end': 5111.077, 'text': 'And half the stuff worked like first time.', 'start': 5108.596, 'duration': 2.481}, {'end': 5112.278, 'text': "It wasn't even rocket science.", 'start': 5111.097, 'duration': 1.181}, {'end': 5121.332, 'text': 'But at the core of that is questioning the assumptions or sort of entering the thinking first principles.', 'start': 5113.547, 'duration': 7.785}, {'end': 5128.857, 'text': 'thinking sort of looking at the reality of the situation and using that knowledge, applying that knowledge.', 'start': 5121.332, 'duration': 7.525}, {'end': 5138.203, 'text': 'So I would say my brain has this idea that you can question first assumptions, but I can go days at a time and forget that.', 'start': 5129.057, 'duration': 9.146}], 'summary': 'Half the tested approaches succeeded; questioning assumptions is key.', 'duration': 29.607, 'max_score': 5108.596, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/Nb2tebYAaOA/pics/Nb2tebYAaOA5108596.jpg'}, {'end': 5200.045, 'src': 'embed', 'start': 5153.646, 'weight': 0, 'content': [{'end': 5159.568, 'text': 'or you know being happy takes priority, or you know screwing around takes priority.', 'start': 5153.646, 'duration': 5.922}, {'end': 5162.188, 'text': 'Like how you go through life is complicated.', 'start': 5160.428, 'duration': 1.76}, {'end': 5166.35, 'text': 'Yeah And then you remember, oh, yeah, I could really think first principles.', 'start': 5162.208, 'duration': 4.142}, {'end': 5169.09, 'text': "Oh, shit, that's tiring, you know.", 'start': 5166.55, 'duration': 2.54}, {'end': 5171.111, 'text': "But you do for a while and that's kind of cool.", 'start': 5169.591, 'duration': 1.52}, {'end': 5181.513, 'text': 'So, just as a last question in your sense, from the big picture, from the first principles, do you think you kind of answered it already?', 'start': 5172.768, 'duration': 8.745}, {'end': 5192.239, 'text': 'but do you think autonomous driving is something we can solve on a timeline of years? So one, two, three, five, 10 years,', 'start': 5181.513, 'duration': 10.726}, {'end': 5194.44, 'text': 'as opposed to a century? Yeah, definitely.', 'start': 5192.239, 'duration': 2.201}, {'end': 5197.522, 'text': 'Just to linger on it a little longer.', 'start': 5195.42, 'duration': 2.102}, {'end': 5200.045, 'text': "where's the confidence coming from??", 'start': 5197.522, 'duration': 2.523}], 'summary': 'Navigating life is complex, prioritizing happiness, screwing around, and considering first principles, while also expressing confidence in solving autonomous driving within the next 10 years.', 'duration': 46.399, 'max_score': 5153.646, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/Nb2tebYAaOA/pics/Nb2tebYAaOA5153646.jpg'}], 'start': 4565.257, 'title': "Craftsman's work and lessons from tesla", 'summary': "Discusses the satisfaction and complexity of craftsman's work in engineering, emphasizing intricate skills in autonomous driving and factory assembly, lessons learned from working with elon musk at tesla, and the solvability of autonomous driving within a few years.", 'chapters': [{'end': 4830.545, 'start': 4565.257, 'title': "Craftsman's work in engineering", 'summary': "Discusses the satisfaction and complexity of craftsman's work in engineering, highlighting the need for intricate skills in autonomous driving and factory assembly, and contrasting it with the invention in new technological developments.", 'duration': 265.288, 'highlights': ["Craftsman's work in engineering involves satisfaction in mastering complicated skills and making thoughtful decisions, such as in autonomous driving and factory assembly.", "Engineering often consists of craftsman's work, requiring intricate skills and thoughtful decisions in areas like autonomous driving and factory assembly.", 'Factory assembly work is complicated and satisfying, involving coordination and meticulous tasks, such as attaching trim on a moving assembly line.']}, {'end': 5171.111, 'start': 4831.186, 'title': 'Lessons from working with elon musk', 'summary': 'Discusses the lessons learned from working with elon musk at tesla, highlighting the importance of questioning assumptions, applying first principles thinking, and the emotional and intellectual challenges involved, with examples of innovation and craftsmanship at tesla and spacex.', 'duration': 339.925, 'highlights': ["Elon Musk's deep belief in challenging local maximum and the example of a friend inventing a better electric motor Elon Musk's belief in challenging local maximum is exemplified by a friend's invention of a better electric motor.", "The satisfaction of doing innovative and craftsman's work, as well as the importance of taking things apart to understand deep first principles The satisfaction of innovative and craftsman's work is emphasized, along with the importance of understanding deep first principles by taking things apart.", 'Emotional and intellectual challenges of stripping away assumptions and the ability to think better and more clearly The emotional and intellectual challenges of stripping away assumptions are discussed, along with the ability to think better and more clearly.', 'The value of reading and applying knowledge from books, as well as the core principle of questioning assumptions and using first principles thinking The value of reading and applying knowledge from books is highlighted, along with the core principle of questioning assumptions and using first principles thinking.', 'The complexity of life and the challenge of maintaining focus on first principles thinking The complexity of life and the challenge of maintaining focus on first principles thinking is acknowledged.']}, {'end': 5674.505, 'start': 5172.768, 'title': 'Autonomous driving: challenges and surprises', 'summary': 'Discusses the solvability of autonomous driving within a timeline of years, the potential of superhuman level ai, and the philosophical perspectives on human nature and the future of society.', 'duration': 501.737, 'highlights': ['The solvability of autonomous driving is discussed, indicating the potential for solving the problem within a timeline of years, leveraging the increasing data gathering and computation, and the ongoing improvements in hardware and software.', 'The potential existential threats from superhuman level AI are debated, addressing the concerns of societal impact and the likelihood of coexistence with superintelligent entities, while highlighting the diversity of human experiences and the potential for meaningful pursuits.', 'Philosophical perspectives on human nature and the future of society are explored, reflecting on the competitive and creative tensions within society, the pursuit of meaningful experiences, and the potential for unexplored avenues in life.', 'The interviewee expresses a jaded optimism and a love for the unknown, emphasizing the satisfaction derived from the mysteries and uncertainties of life, while rejecting the idea of reliving past moments and embracing the anxiety of the unknown.', 'The conversation concludes with reflections on the purpose of life, highlighting the inherent nature of human exploration and discovery as a fundamental aspect of existence, with a closing quote urging continuous effort and experimentation in life.']}], 'duration': 1109.248, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/Nb2tebYAaOA/pics/Nb2tebYAaOA4565257.jpg', 'highlights': ['The solvability of autonomous driving is discussed, indicating the potential for solving the problem within a timeline of years, leveraging the increasing data gathering and computation, and the ongoing improvements in hardware and software.', "Elon Musk's deep belief in challenging local maximum and the example of a friend inventing a better electric motor Elon Musk's belief in challenging local maximum is exemplified by a friend's invention of a better electric motor.", "The satisfaction of doing innovative and craftsman's work, as well as the importance of taking things apart to understand deep first principles The satisfaction of innovative and craftsman's work is emphasized, along with the importance of understanding deep first principles by taking things apart.", 'The value of reading and applying knowledge from books, as well as the core principle of questioning assumptions and using first principles thinking The value of reading and applying knowledge from books is highlighted, along with the core principle of questioning assumptions and using first principles thinking.', "Craftsman's work in engineering involves satisfaction in mastering complicated skills and making thoughtful decisions, such as in autonomous driving and factory assembly.", "Engineering often consists of craftsman's work, requiring intricate skills and thoughtful decisions in areas like autonomous driving and factory assembly.", 'The complexity of life and the challenge of maintaining focus on first principles thinking The complexity of life and the challenge of maintaining focus on first principles thinking is acknowledged.', 'Factory assembly work is complicated and satisfying, involving coordination and meticulous tasks, such as attaching trim on a moving assembly line.']}], 'highlights': ["Jim Keller's significant contributions to microarchitectures and processors at AMD, Apple, Tesla, and Intel.", 'The importance of achieving deeper understanding rather than executing a large set of recipes in computer architecture', 'The parallels between building computers and organizing people, reframing organizational design as a computer architectural problem', "Moore's Law: Double the number of transistors every two years", 'The cascade of diminishing return curves in technological innovations, contributing to an exponential curve of progress', 'The potential for a 100x innovation in computation is highlighted, indicating a significant advancement in technology.', "The complexity of the human brain's neural network with local and global connections and distributed information storage is emphasized, contrasting with the decoupled memory and computation architecture of modern computers.", "The ongoing innovations in shrinking transistors and the presence of multiple S-curves in semiconductor technology suggest that Moore's Law is likely to continue influencing AI research and computational advancements.", 'The evolution of AI computations from simple equations to data topology problems is discussed, highlighting the increasing complexity of AI algorithms and the shift towards viewing data as a topology problem.', 'The challenges of computing performance and the exciting potential of more powerful computers are detailed, emphasizing the need to refactor software and the impact of faster computers on algorithm run times.', 'The potential impact of quantum and analog computing on changing the fundamental nature of computation is mentioned, suggesting a potential radical shift in computational processes.', 'The ongoing debate about whether AI is fundamentally a form of search is explored, with differing perspectives on the nature of AI computations and the concept of optimization as search.', 'The complexity of life and the challenge of maintaining focus on first principles thinking The complexity of life and the challenge of maintaining focus on first principles thinking is acknowledged.']}