title
George Hotz: Comma.ai, OpenPilot, and Autonomous Vehicles | Lex Fridman Podcast #31

description

detail
{'title': 'George Hotz: Comma.ai, OpenPilot, and Autonomous Vehicles | Lex Fridman Podcast #31', 'heatmap': [{'end': 4023.512, 'start': 3948.277, 'weight': 1}], 'summary': "Features a conversation with george hotz, founder of kama ai, discussing ai, technology, machine learning-based vehicle automation, and unlocking the first iphone, covering topics such as hacking, ethical considerations, challenges in autonomous driving, tesla's autonomous driving challenges, ai improvements, waymo's impact, innovations in autonomous vehicles, and discussions on ai, tech, revenue, and ethical practices.", 'chapters': [{'end': 61.359, 'segs': [{'end': 38.289, 'src': 'embed', 'start': 0.129, 'weight': 0, 'content': [{'end': 2.05, 'text': 'The following is a conversation with George Hotz.', 'start': 0.129, 'duration': 1.921}, {'end': 6.894, 'text': "He's the founder of Kama AI, a machine learning based vehicle automation company.", 'start': 2.631, 'duration': 4.263}, {'end': 12.677, 'text': 'He is most certainly an outspoken personality in the field of AI and technology in general.', 'start': 7.534, 'duration': 5.143}, {'end': 17.921, 'text': 'He first gained recognition for being the first person to carrier unlock an iPhone.', 'start': 13.298, 'duration': 4.623}, {'end': 23.765, 'text': "And since then, he's done quite a few interesting things at the intersection of hardware and software.", 'start': 18.501, 'duration': 5.264}, {'end': 27.126, 'text': 'This is the Artificial Intelligence Podcast.', 'start': 24.505, 'duration': 2.621}, {'end': 35.008, 'text': 'If you enjoy it, subscribe on YouTube, give it five stars on iTunes, support it on Patreon or simply connect with me on Twitter.', 'start': 27.526, 'duration': 7.482}, {'end': 38.289, 'text': 'at Lex Friedman spelled F-R-I-D-M-A-N.', 'start': 35.008, 'duration': 3.281}], 'summary': 'George hotz, founder of kama ai, is a prominent figure in ai and technology, known for unlocking the first iphone and working at the intersection of hardware and software.', 'duration': 38.16, 'max_score': 0.129, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/iwcYp-XT7UI/pics/iwcYp-XT7UI129.jpg'}], 'start': 0.129, 'title': 'A conversation with george hotz', 'summary': 'Features a conversation with george hotz, founder of kama ai, discussing his impactful work in ai and technology, particularly in machine learning based vehicle automation and his notable achievement of being the first person to carrier unlock an iphone.', 'chapters': [{'end': 61.359, 'start': 0.129, 'title': 'Conversation with george hotz', 'summary': 'Features a conversation with george hotz, founder of kama ai, discussing his impactful work in ai and technology, particularly in machine learning based vehicle automation and his notable achievement of being the first person to carrier unlock an iphone.', 'duration': 61.23, 'highlights': ['George Hotz is the founder of Kama AI, a machine learning based vehicle automation company.', 'He gained recognition for being the first person to carrier unlock an iPhone.', 'He has made significant contributions at the intersection of hardware and software.', 'He is known for being an outspoken personality in the field of AI and technology.', 'The podcast host expresses gratitude to a supporter of the podcast on Patreon, Jennifer from Canada.']}], 'duration': 61.23, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/iwcYp-XT7UI/pics/iwcYp-XT7UI129.jpg', 'highlights': ['George Hotz is the founder of Kama AI, a machine learning based vehicle automation company.', 'He gained recognition for being the first person to carrier unlock an iPhone.', 'He has made significant contributions at the intersection of hardware and software.', 'He is known for being an outspoken personality in the field of AI and technology.']}, {'end': 1189.226, 'segs': [{'end': 116.542, 'src': 'embed', 'start': 87.915, 'weight': 1, 'content': [{'end': 90.977, 'text': "it may be impossible to prove whether we're in a simulation or not.", 'start': 87.915, 'duration': 3.062}, {'end': 95.618, 'text': "So they've designed it such that it's a closed system.", 'start': 92.597, 'duration': 3.021}, {'end': 96.919, 'text': "You can't get outside the system.", 'start': 95.678, 'duration': 1.241}, {'end': 98.704, 'text': "Well, maybe it's one of three worlds.", 'start': 97.182, 'duration': 1.522}, {'end': 100.966, 'text': "We're either in a simulation which can be exploited.", 'start': 98.744, 'duration': 2.222}, {'end': 105.731, 'text': "We're in a simulation which not only can't be exploited, but the same thing is true about VMs.", 'start': 101.427, 'duration': 4.304}, {'end': 110.196, 'text': "A really well-designed VM, you can't even detect if you're in a VM or not.", 'start': 106.492, 'duration': 3.704}, {'end': 112.098, 'text': "That's brilliant.", 'start': 111.377, 'duration': 0.721}, {'end': 116.542, 'text': 'Yeah, so the simulation is running on a virtual machine.', 'start': 112.698, 'duration': 3.844}], 'summary': 'The discussion explores the possibility of being in a simulation with closed systems and virtual machines.', 'duration': 28.627, 'max_score': 87.915, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/iwcYp-XT7UI/pics/iwcYp-XT7UI87915.jpg'}, {'end': 317.949, 'src': 'embed', 'start': 294.119, 'weight': 4, 'content': [{'end': 302.303, 'text': "The space narrative that held for the modernist generation doesn't hold as well for the postmodern generation.", 'start': 294.119, 'duration': 8.184}, {'end': 305.404, 'text': "What's the space narrative??", 'start': 304.564, 'duration': 0.84}, {'end': 307.325, 'text': 'Are we talking about the same space, the three-dimensional space??', 'start': 305.444, 'duration': 1.881}, {'end': 311.987, 'text': "No, no, space like going up in space, like building like Elon Musk, like we're going to build rockets, we're going to go to Mars,", 'start': 307.345, 'duration': 4.642}, {'end': 313.307, 'text': "we're going to colonize the universe.", 'start': 311.987, 'duration': 1.32}, {'end': 317.949, 'text': "And the narrative you're referring, I was born in the Soviet Union, you're referring to the race to space.", 'start': 313.527, 'duration': 4.422}], 'summary': 'The space narrative of colonizing the universe differs between modernist and postmodern generations.', 'duration': 23.83, 'max_score': 294.119, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/iwcYp-XT7UI/pics/iwcYp-XT7UI294119.jpg'}, {'end': 426.489, 'src': 'embed', 'start': 395.06, 'weight': 0, 'content': [{'end': 406.829, 'text': 'The world first learned of George Hotz in August 2007, maybe before then, but certainly in August 2007, when you were the first person to unlock.', 'start': 395.06, 'duration': 11.769}, {'end': 407.97, 'text': 'carry unlock an iPhone.', 'start': 406.829, 'duration': 1.141}, {'end': 414.855, 'text': 'How did you get into hacking? What was the first system you discovered vulnerabilities for and broke into? So..', 'start': 408.931, 'duration': 5.924}, {'end': 420.905, 'text': 'That was really kind of the first thing.', 'start': 418.903, 'duration': 2.002}, {'end': 426.489, 'text': 'I had a book in 2006 called Gray Hat Hacking.', 'start': 421.685, 'duration': 4.804}], 'summary': 'George hotz gained attention in 2007 for being the first to unlock an iphone and had a book in 2006 called gray hat hacking.', 'duration': 31.429, 'max_score': 395.06, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/iwcYp-XT7UI/pics/iwcYp-XT7UI395060.jpg'}, {'end': 595.845, 'src': 'embed', 'start': 565.892, 'weight': 2, 'content': [{'end': 567.553, 'text': 'How do you pronounce this? Kira? Kira, yeah.', 'start': 565.892, 'duration': 1.661}, {'end': 576.821, 'text': "So it's essentially the most efficient way to visualize the change of state of the computer as the program is running.", 'start': 568.414, 'duration': 8.407}, {'end': 578.182, 'text': "That's what you mean by debugger.", 'start': 576.841, 'duration': 1.341}, {'end': 584.478, 'text': "Yeah, it's a timeless debugger, so you can rewind just as easily as going forward.", 'start': 578.915, 'duration': 5.563}, {'end': 589.341, 'text': "Think about if you're using GDB, you have to put a watch on a variable if you want to see if that variable changes.", 'start': 585.099, 'duration': 4.242}, {'end': 595.845, 'text': 'In Cura, you can just click on that variable and then it shows every single time when that variable was changed or accessed.', 'start': 589.701, 'duration': 6.144}], 'summary': 'Discussion on efficient debugging using cura, with easy variable tracking.', 'duration': 29.953, 'max_score': 565.892, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/iwcYp-XT7UI/pics/iwcYp-XT7UI565892.jpg'}, {'end': 960.244, 'src': 'embed', 'start': 934.324, 'weight': 3, 'content': [{'end': 939.868, 'text': "But if you're good and you have these skills, you probably have a nice life too, right? Right, you can use it for other things.", 'start': 934.324, 'duration': 5.544}, {'end': 941.129, 'text': 'But is there an ethical?', 'start': 940.128, 'duration': 1.001}, {'end': 947.194, 'text': 'is there a little voice in your head that says well, yeah,', 'start': 941.129, 'duration': 6.065}, {'end': 956.321, 'text': 'if you could hack something to where you could hurt people and you could earn a lot of money doing it, though not hurt physically, perhaps,', 'start': 947.194, 'duration': 9.127}, {'end': 960.244, 'text': 'but disrupt their life in some kind of way.', 'start': 956.321, 'duration': 3.923}], 'summary': 'Having hacking skills can lead to ethical dilemmas and potential harm to others.', 'duration': 25.92, 'max_score': 934.324, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/iwcYp-XT7UI/pics/iwcYp-XT7UI934324.jpg'}], 'start': 62.8, 'title': 'Hacking and ethics', 'summary': "Discusses living in a simulation, hackability, george hotz's hacking journey including unlocking the first iphone, and ethical considerations in hacking and security.", 'chapters': [{'end': 394.179, 'start': 62.8, 'title': 'Simulation and hacking: exploring virtual realities', 'summary': 'Delves into the concept of living in a simulation, discussing its unfalsifiability and potential hackability, while also exploring the implications on human progress and narratives like space exploration and virtual reality.', 'duration': 331.379, 'highlights': ["Living in a simulation may be unfalsifiable due to a closed system design, making it impossible to prove whether we're in a simulation or not. The concept of living in a simulation is discussed, highlighting the challenge of proving its existence due to a closed system design.", 'The possibility of being in a simulation that is not exploitable and indistinguishable from reality is raised, emphasizing the complexity of virtual environments. The discussion explores the potential existence of non-exploitable simulations and their indistinguishability from reality, highlighting the complexity of virtual environments.', 'Exploring the idea of hacking a simulation and its potential impact on the universe, emphasizing the need for a shift in rhetoric and thinking upwards to address technological challenges. The concept of hacking a simulation is considered, emphasizing the need for a shift in rhetoric and thinking upwards to address technological challenges and potential impacts on the universe.', 'The shift in space exploration narratives from modernist to postmodern perspectives is discussed, highlighting the evolving cultural attitudes toward space exploration. The shift in space exploration narratives from modernist to postmodern perspectives is explored, highlighting the evolving cultural attitudes toward space exploration.', 'The interest in virtual reality and the desire to physically move to a virtual world is mentioned, emphasizing the potential allure of virtual environments. The interest in virtual reality and the desire to physically move to a virtual world is highlighted, emphasizing the potential allure of virtual environments.']}, {'end': 882.409, 'start': 395.06, 'title': 'George hotz: the journey of a hacker', 'summary': "Highlights george hotz's journey into hacking, including the unlock of the first iphone in august 2007, his learning process, development of a debugger tool, and insights into the world of competitive hacking.", 'duration': 487.349, 'highlights': ['George Hotz unlocked the first iPhone in August 2007, marking his entry into hacking. George Hotz gained recognition for being the first to unlock an iPhone, signifying the start of his hacking journey.', "George Hotz's learning process involved starting with electronics, progressing to software exploitation, and acquiring programming skills over years of practice. Hotz initially focused on electronics and physical hacks, then transitioned to software exploitation and programming through years of practice and learning.", "Hotz developed a debugger tool called Kira, offering efficient visualization of computer state changes while running programs. Hotz's creation of Kira provided an innovative debugger tool for visualizing computer state changes during program execution.", 'Competitive hacking through CTFs involved finding vulnerabilities, creating exploits, and earning points by running arbitrary code on systems. Hotz participated in Capture The Flag (CTF) competitions, where the goal was to find vulnerabilities, create exploits, and run arbitrary code on systems to earn points.']}, {'end': 1189.226, 'start': 882.83, 'title': 'Ethics in hacking and security', 'summary': "Explores the ethical considerations and insights from a talented programmer and hacker, the evolution as a programmer, and the purpose of google's project zero in enhancing security efforts.", 'duration': 306.396, 'highlights': ['The chapter explores the ethical considerations and insights from a talented programmer and hacker. The discussion delves into the ethical dilemma of hacking, the distribution of good and bad people, and the influence of ethical principles on the behavior of individuals.', "The evolution as a programmer and insights from challenging CS courses. The individual's programming evolution, including learning from challenging CS courses, such as operating systems, compilers, and AI, and the transition from C to Python.", "Insights into Google's Project Zero and its purpose in enhancing security efforts. An overview of Project Zero as Google's offensive security team aimed at identifying and addressing vulnerabilities to enhance overall security, particularly in bug bounties."]}], 'duration': 1126.426, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/iwcYp-XT7UI/pics/iwcYp-XT7UI62800.jpg', 'highlights': ['George Hotz unlocked the first iPhone in August 2007, marking his entry into hacking.', 'The concept of living in a simulation is discussed, highlighting the challenge of proving its existence due to a closed system design.', "Hotz's creation of Kira provided an innovative debugger tool for visualizing computer state changes during program execution.", 'The discussion delves into the ethical dilemma of hacking, the distribution of good and bad people, and the influence of ethical principles on the behavior of individuals.', 'The shift in space exploration narratives from modernist to postmodern perspectives is explored, highlighting the evolving cultural attitudes toward space exploration.']}, {'end': 2342.39, 'segs': [{'end': 1223.718, 'src': 'embed', 'start': 1189.466, 'weight': 0, 'content': [{'end': 1190.187, 'text': "They don't even fix the bug.", 'start': 1189.466, 'duration': 0.721}, {'end': 1193.83, 'text': "It doesn't hurt for there to be a vulnerability.", 'start': 1190.607, 'duration': 3.223}, {'end': 1195.371, 'text': "So Project Zero is like, we're going to do it different.", 'start': 1193.85, 'duration': 1.521}, {'end': 1199.374, 'text': "We're going to announce a vulnerability and we're going to give them 90 days to fix it.", 'start': 1195.991, 'duration': 3.383}, {'end': 1202.017, 'text': "And then whether they fix it or not, we're going to drop the zero day.", 'start': 1199.715, 'duration': 2.302}, {'end': 1203.775, 'text': 'Oh, wow.', 'start': 1203.374, 'duration': 0.401}, {'end': 1204.997, 'text': "We're going to drop the weapon.", 'start': 1204.116, 'duration': 0.881}, {'end': 1205.718, 'text': "That's so cool.", 'start': 1205.017, 'duration': 0.701}, {'end': 1206.7, 'text': 'That is so cool.', 'start': 1205.959, 'duration': 0.741}, {'end': 1208.844, 'text': 'I love that deadlines.', 'start': 1207.582, 'duration': 1.262}, {'end': 1209.986, 'text': "Oh, that's so cool.", 'start': 1209.385, 'duration': 0.601}, {'end': 1210.687, 'text': 'Give them real deadlines.', 'start': 1210.046, 'duration': 0.641}, {'end': 1214.835, 'text': "Yeah And I think it's done a lot for moving the industry forward.", 'start': 1210.948, 'duration': 3.887}, {'end': 1218.676, 'text': 'I watched your coding sessions that you streamed online.', 'start': 1215.874, 'duration': 2.802}, {'end': 1223.718, 'text': 'You code things up, basic projects, usually from scratch.', 'start': 1220.457, 'duration': 3.261}], 'summary': 'Project zero announces vulnerabilities, giving 90-day fix deadlines, advancing industry.', 'duration': 34.252, 'max_score': 1189.466, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/iwcYp-XT7UI/pics/iwcYp-XT7UI1189466.jpg'}, {'end': 1284.814, 'src': 'embed', 'start': 1234.765, 'weight': 1, 'content': [{'end': 1237.046, 'text': "I don't know if that's always true, but certainly for the live streams.", 'start': 1234.765, 'duration': 2.281}, {'end': 1244.489, 'text': "So it's interesting to me, because I'm more, I'm much slower and systematic and careful, and you just move.", 'start': 1237.686, 'duration': 6.803}, {'end': 1246.67, 'text': 'I mean probably an order of magnitude faster.', 'start': 1244.489, 'duration': 2.181}, {'end': 1254.133, 'text': "So I'm curious, is there a method to your madness? Is it just who you are? There's pros and cons.", 'start': 1246.69, 'duration': 7.443}, {'end': 1259.056, 'text': "There's pros and cons to my programming style, and I'm aware of them.", 'start': 1254.753, 'duration': 4.303}, {'end': 1266.601, 'text': "If you ask me to get something up and working quickly with an API that's kind of undocumented,", 'start': 1259.716, 'duration': 6.885}, {'end': 1269.643, 'text': 'I will do this super fast because I will throw things at it until it works.', 'start': 1266.601, 'duration': 3.042}, {'end': 1281.49, 'text': "If you ask me to take a vector and rotate it 90 degrees and then flip it over the XY plane, I'll spam program for two hours and won't get it.", 'start': 1270.343, 'duration': 11.147}, {'end': 1284.814, 'text': "Oh, because it's something that you could do with a sheet of paper.", 'start': 1282.392, 'duration': 2.422}], 'summary': 'Different programming styles: fast and experimental vs. slow and systematic, with examples of their strengths and weaknesses.', 'duration': 50.049, 'max_score': 1234.765, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/iwcYp-XT7UI/pics/iwcYp-XT7UI1234765.jpg'}, {'end': 1400.524, 'src': 'embed', 'start': 1370.777, 'weight': 3, 'content': [{'end': 1375.299, 'text': 'If you want to install something like Ceph.', 'start': 1370.777, 'duration': 4.522}, {'end': 1386.824, 'text': "Ceph is, I think, the open infrastructure distributed file system, or there's newer ones like SeaweedFS, but these are all 10,000 plus line projects.", 'start': 1375.299, 'duration': 11.525}, {'end': 1387.704, 'text': 'I think some of them are even 100,000 line.', 'start': 1386.864, 'duration': 0.84}, {'end': 1390.768, 'text': 'and just configuring them was a nightmare.', 'start': 1389.506, 'duration': 1.262}, {'end': 1393.933, 'text': 'So I wrote one.', 'start': 1391.189, 'duration': 2.744}, {'end': 1400.524, 'text': "It's 200 lines, and it uses like Nginx and the Lime servers and has this little master server that I wrote in Go.", 'start': 1394.814, 'duration': 5.71}], 'summary': 'Developed a 200-line distributed file system using nginx and lime servers, as an alternative to complex 10,000+ line projects like ceph.', 'duration': 29.747, 'max_score': 1370.777, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/iwcYp-XT7UI/pics/iwcYp-XT7UI1370777.jpg'}, {'end': 1478.688, 'src': 'embed', 'start': 1439.723, 'weight': 4, 'content': [{'end': 1444.244, 'text': 'When I first looked at it, I was like, this is like Python, but it takes twice as long to do anything.', 'start': 1439.723, 'duration': 4.521}, {'end': 1451.025, 'text': 'Now that OpenPilot is migrating to C, but it still has large Python components,', 'start': 1445.704, 'duration': 5.321}, {'end': 1454.906, 'text': "I now understand why Python doesn't work for large code bases and why you want something like Go.", 'start': 1451.025, 'duration': 3.881}, {'end': 1460.361, 'text': "Interesting. So why why doesn't Python work for so even most?", 'start': 1455.859, 'duration': 4.502}, {'end': 1468.184, 'text': 'uh, speaking for myself at least, like we do a lot of stuff basically demo level work with autonomous vehicles, and most of the work is Python.', 'start': 1460.361, 'duration': 7.823}, {'end': 1477.868, 'text': "Yeah Why doesn't Python work for large code bases? Because, well, lack of type checking is, is, is a big.", 'start': 1468.204, 'duration': 9.664}, {'end': 1478.688, 'text': 'So errors creep in.', 'start': 1477.888, 'duration': 0.8}], 'summary': "Openpilot is migrating to c due to python's inefficiency for large code bases, lack of type checking leading to errors.", 'duration': 38.965, 'max_score': 1439.723, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/iwcYp-XT7UI/pics/iwcYp-XT7UI1439723.jpg'}, {'end': 1538.012, 'src': 'embed', 'start': 1502.641, 'weight': 6, 'content': [{'end': 1504.382, 'text': "Well, maybe it's kind of related to it, so it's lack of.", 'start': 1502.641, 'duration': 1.741}, {'end': 1509.593, 'text': "What's in your toolbox these days? Is it Python? What else? Go? I need to move to something else.", 'start': 1504.967, 'duration': 4.626}, {'end': 1514.159, 'text': 'My adventure into dependently typed languages, I love these languages.', 'start': 1510.334, 'duration': 3.825}, {'end': 1516.882, 'text': 'They just have syntax from the 80s.', 'start': 1514.199, 'duration': 2.683}, {'end': 1528.741, 'text': 'What do you think about JavaScript, ES6, TypeScript? JavaScript is, the whole ecosystem is unbelievably confusing.', 'start': 1516.902, 'duration': 11.839}, {'end': 1538.012, 'text': "NPM updates a package from 0.2.2 to 0.2.5, and that breaks your Babel linter, which translates your ES5 into ES6, which doesn't run on.", 'start': 1529.342, 'duration': 8.67}], 'summary': 'Discussion on programming languages: python, go, javascript, es6, typescript. frustration with npm updates and babel linter.', 'duration': 35.371, 'max_score': 1502.641, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/iwcYp-XT7UI/pics/iwcYp-XT7UI1502641.jpg'}, {'end': 1660.661, 'src': 'embed', 'start': 1617.898, 'weight': 7, 'content': [{'end': 1621.181, 'text': "Can you also, just for people who don't know, tell the story of Karma AI?", 'start': 1617.898, 'duration': 3.283}, {'end': 1621.321, 'text': 'Sure.', 'start': 1621.201, 'duration': 0.12}, {'end': 1632.088, 'text': "Sure. So I was working at this AI startup and a friend approached me and he's like dude, I don't know where this is going,", 'start': 1621.401, 'duration': 10.687}, {'end': 1636.049, 'text': 'but the coolest applied AI problem today is self-driving cars.', 'start': 1632.088, 'duration': 3.961}, {'end': 1637.449, 'text': "I'm like, well, absolutely.", 'start': 1636.069, 'duration': 1.38}, {'end': 1646.752, 'text': "Do you want to meet with Elon Musk? And he's looking for somebody to build a vision system for autopilot.", 'start': 1637.469, 'duration': 9.283}, {'end': 1648.992, 'text': 'This is when they were still on AP1.', 'start': 1647.592, 'duration': 1.4}, {'end': 1650.192, 'text': 'They were still using Mobileye.', 'start': 1649.292, 'duration': 0.9}, {'end': 1652.733, 'text': 'Elon back then was looking for a replacement.', 'start': 1650.892, 'duration': 1.841}, {'end': 1660.661, 'text': 'And he brought me in and we talked about a contract where I would deliver something that meets Mobileye level performance.', 'start': 1653.934, 'duration': 6.727}], 'summary': 'Karma ai founder was approached to build vision system for self-driving cars, aiming for mobileye level performance.', 'duration': 42.763, 'max_score': 1617.898, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/iwcYp-XT7UI/pics/iwcYp-XT7UI1617898.jpg'}, {'end': 1723.357, 'src': 'embed', 'start': 1692.26, 'weight': 9, 'content': [{'end': 1696.662, 'text': "So that's the best place to be, because you basically, full steam, take on a problem.", 'start': 1692.26, 'duration': 4.402}, {'end': 1701.349, 'text': 'How confident how from that time, because you know a lot more now.', 'start': 1697.222, 'duration': 4.127}, {'end': 1705.315, 'text': 'at that time, how hard do you think it is to solve all of autonomous driving?', 'start': 1701.349, 'duration': 3.966}, {'end': 1714.195, 'text': 'I remember I suggested to Elon in the meeting putting a GPU behind each camera to keep the compute local.', 'start': 1705.914, 'duration': 8.281}, {'end': 1717.776, 'text': 'This is an incredibly stupid idea.', 'start': 1715.116, 'duration': 2.66}, {'end': 1723.357, 'text': "I leave the meeting 10 minutes later and I'm like, I could have spent a little bit of time thinking about this problem before I went in.", 'start': 1718.056, 'duration': 5.301}], 'summary': 'Discussing autonomous driving challenges and past suggestions, regretting not spending more time on problem-solving.', 'duration': 31.097, 'max_score': 1692.26, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/iwcYp-XT7UI/pics/iwcYp-XT7UI1692260.jpg'}, {'end': 2051.911, 'src': 'embed', 'start': 2000.098, 'weight': 8, 'content': [{'end': 2001.259, 'text': 'It feels, it feels right.', 'start': 2000.098, 'duration': 1.161}, {'end': 2008.922, 'text': "That's interesting because every time you touch a lane, that's a source of a little bit of stress and kind of lane keeping is removing that stress.", 'start': 2001.739, 'duration': 7.183}, {'end': 2015.224, 'text': "That's ultimately the biggest value add, honestly, is just removing the stress of having to stay in lane.", 'start': 2009.262, 'duration': 5.962}, {'end': 2024.008, 'text': "And I don't think people fully realize, first of all, that that's a big value add, but also that that's all it is.", 'start': 2015.665, 'duration': 8.343}, {'end': 2028.31, 'text': 'And that not only, I find it a huge value add.', 'start': 2025.008, 'duration': 3.302}, {'end': 2033.253, 'text': 'I drove down, when we moved to San Diego, I drove down in an Enterprise Rent-A-Car and I missed it.', 'start': 2028.71, 'duration': 4.543}, {'end': 2034.674, 'text': 'So I missed having the system so much.', 'start': 2033.333, 'duration': 1.341}, {'end': 2039.717, 'text': "It's so much more tiring to drive without it.", 'start': 2035.895, 'duration': 3.822}, {'end': 2044.36, 'text': "It is that lane centering that's the key feature.", 'start': 2040.377, 'duration': 3.983}, {'end': 2051.911, 'text': "And in a way, it's the only feature that actually adds value to people's lives in autonomous vehicles today.", 'start': 2046.647, 'duration': 5.264}], 'summary': 'Lane keeping reduces stress, adds value, and is the key feature in autonomous vehicles.', 'duration': 51.813, 'max_score': 2000.098, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/iwcYp-XT7UI/pics/iwcYp-XT7UI2000098.jpg'}], 'start': 1189.466, 'title': 'Impact of project zero and challenges in autonomous driving', 'summary': 'Discusses the impact of project zero in setting 90-day deadlines for fixing vulnerabilities, leading to industry advancement and improved efficiency, contrasts coding styles resulting in industry advancement and improved efficiency. it also highlights the challenges faced in developing autonomous driving systems, including financial incentives and penalties, technological challenges, and the importance of good lane keeping.', 'chapters': [{'end': 1346.933, 'start': 1189.466, 'title': 'Impact of project zero on coding and programming style', 'summary': 'Discusses the impact of project zero in setting 90-day deadlines for fixing vulnerabilities, leading to industry advancement, and contrasts coding styles, with one being methodical and careful while the other being fast and chaotic, resulting in improved efficiency in different scenarios.', 'duration': 157.467, 'highlights': ["Project Zero's approach of setting 90-day deadlines for fixing vulnerabilities has advanced the industry Project Zero announces vulnerabilities and gives 90 days to fix them, regardless of which it drops the zero day.", 'Contrast in coding styles: methodical and careful vs. fast and chaotic The discussion contrasts a slower, systematic coding style with a faster, chaotic approach, emphasizing the pros and cons of each.', 'Efficiency in coding approach: fast problem-solving with undocumented APIs The speaker excels at quickly getting things to work with undocumented APIs by throwing things at it until it works, highlighting the efficiency of this approach.']}, {'end': 1660.661, 'start': 1347.053, 'title': 'Learning new tools and programming languages', 'summary': 'Discusses the development of a 200-line distributed file system, the challenges of using python for large code bases, the advantages of go over python, and the complexities of javascript, es6, and typescript development. the chapter also touches on the journey into autonomous vehicle automation and the founding of karma ai.', 'duration': 313.608, 'highlights': ['The development of a 200-line distributed file system using Nginx and Lime servers, and a master server written in Go after discarding three previous versions, showcases the iterative and efficient approach to software development. The speaker developed a 200-line distributed file system using Nginx, Lime servers, and a master server written in Go after discarding three previous versions, highlighting an iterative and efficient approach to software development.', "The limitations of Python for large code bases, attributed to lack of type checking leading to errors and the language's slower performance, are discussed, offering insights into the challenges with Python in certain contexts. The limitations of Python for large code bases, attributed to lack of type checking leading to errors and the language's slower performance, are discussed, offering insights into the challenges with Python in certain contexts.", 'The advantages of Go over Python for large code bases, including its strong typing and ecosystem, are highlighted, providing valuable insights into the considerations for selecting programming languages for different projects. The advantages of Go over Python for large code bases, including its strong typing and ecosystem, are highlighted, providing valuable insights into the considerations for selecting programming languages for different projects.', 'The complexities and challenges of JavaScript, ES6, and TypeScript development, as well as the uncertainties surrounding the role of JavaScript in the future of distributed computation, are explored, shedding light on the intricacies of JavaScript-based development. The complexities and challenges of JavaScript, ES6, and TypeScript development, as well as the uncertainties surrounding the role of JavaScript in the future of distributed computation, are explored, shedding light on the intricacies of JavaScript-based development.', "The speaker's involvement in autonomous vehicle automation, including the founding of Karma AI and the opportunity to work on a vision system for autopilot, provides a glimpse into the journey and experiences in the realm of vehicle automation. The speaker's involvement in autonomous vehicle automation, including the founding of Karma AI and the opportunity to work on a vision system for autopilot, provides a glimpse into the journey and experiences in the realm of vehicle automation."]}, {'end': 2342.39, 'start': 1661.662, 'title': 'Challenges in autonomous driving', 'summary': 'Highlights the challenges and experiences faced in developing autonomous driving systems, including the financial incentives and penalties, technological challenges, comparison with existing systems, and the importance of good lane keeping. notably, the speaker discusses the financial incentives and penalties, technological challenges faced in developing autonomous driving systems, and the importance of good lane keeping. the speaker also highlights the comparison with existing systems.', 'duration': 680.728, 'highlights': ['The speaker emphasizes the financial incentives and penalties associated with delivering autonomous driving systems, citing the potential $12 million reward for delivering within a short timeline and the loss of $1 million for every month of delay.', "The discussion on the technological challenges faced in developing autonomous driving systems includes the speaker's experience in proposing a computational solution for camera processing, reflecting on the mistake and learning from the encounter with Elon Musk.", "The comparison with existing systems involves the speaker's experience in testing and evaluating the performance of Mobileye's system, highlighting the surprise at its robustness and stability, as well as the comparison with Tesla's Autopilot system.", 'The importance of good lane keeping is underscored, with the speaker expressing the value of this feature in reducing driver stress, highlighting the potential of lane keeping as a reason for consumers to choose a particular car model.']}], 'duration': 1152.924, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/iwcYp-XT7UI/pics/iwcYp-XT7UI1189466.jpg', 'highlights': ["Project Zero's 90-day deadlines for fixing vulnerabilities advanced the industry.", 'Contrast in coding styles: methodical vs. fast and chaotic approach.', 'Efficiency in coding approach: fast problem-solving with undocumented APIs.', 'Development of a 200-line distributed file system showcases an iterative and efficient approach.', 'Limitations of Python for large code bases due to lack of type checking and slower performance.', 'Advantages of Go over Python for large code bases, including strong typing and ecosystem.', 'Complexities and challenges of JavaScript, ES6, and TypeScript development explored.', 'Involvement in autonomous vehicle automation, including founding of Karma AI and work on a vision system for autopilot.', 'Emphasis on financial incentives and penalties for delivering autonomous driving systems.', 'Technological challenges faced in developing autonomous driving systems, including proposing a computational solution for camera processing.', "Comparison with existing systems, including testing and evaluating the performance of Mobileye's system and comparison with Tesla's Autopilot system.", 'Importance of good lane keeping in reducing driver stress and potential consumer appeal.']}, {'end': 3597.418, 'segs': [{'end': 2366.14, 'src': 'embed', 'start': 2342.39, 'weight': 0, 'content': [{'end': 2349.851, 'text': "i don't want a yacht, And I think Tesla actually has a small inkling of that as well.", 'start': 2342.39, 'duration': 7.461}, {'end': 2353.673, 'text': 'With Autonomy Day, they did reveal more than I mean.', 'start': 2349.891, 'duration': 3.782}, {'end': 2358.096, 'text': "of course there's marketing communications, you can tell, but it's more than most companies would reveal, which is,", 'start': 2353.673, 'duration': 4.423}, {'end': 2362.838, 'text': 'I hope they go towards that direction more other companies, GM Ford.', 'start': 2358.096, 'duration': 4.742}, {'end': 2365.34, 'text': "Oh, Tesla's going to win level five.", 'start': 2362.858, 'duration': 2.482}, {'end': 2366.14, 'text': 'They really are.', 'start': 2365.48, 'duration': 0.66}], 'summary': "Tesla's autonomy day revealed more than most companies, likely to win level five autonomy.", 'duration': 23.75, 'max_score': 2342.39, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/iwcYp-XT7UI/pics/iwcYp-XT7UI2342390.jpg'}, {'end': 2446.331, 'src': 'embed', 'start': 2415.635, 'weight': 2, 'content': [{'end': 2416.796, 'text': 'What do you think about that approach?', 'start': 2415.635, 'duration': 1.161}, {'end': 2417.976, 'text': 'How successful would it be?', 'start': 2416.856, 'duration': 1.12}, {'end': 2425.019, 'text': "It's a ton better approach because Tesla is gathering data on a scale that none of them are.", 'start': 2417.996, 'duration': 7.023}, {'end': 2428.46, 'text': "They're putting real users behind the wheel of the cars.", 'start': 2425.219, 'duration': 3.241}, {'end': 2433.736, 'text': "It's, I think, the only strategy that works, the incremental.", 'start': 2429.891, 'duration': 3.845}, {'end': 2438.882, 'text': "Well, so there's a few components to the Tesla approach that's more than just the incrementalist.", 'start': 2434.517, 'duration': 4.365}, {'end': 2443.628, 'text': 'What you spoke with is the ones, the software, so over-the-air software updates.', 'start': 2438.902, 'duration': 4.726}, {'end': 2446.331, 'text': 'Necessity I mean, Waymo crews have those too.', 'start': 2443.968, 'duration': 2.363}], 'summary': "Tesla's approach gathers data on a large scale, putting real users behind the wheel and using over-the-air software updates.", 'duration': 30.696, 'max_score': 2415.635, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/iwcYp-XT7UI/pics/iwcYp-XT7UI2415635.jpg'}, {'end': 2536.471, 'src': 'embed', 'start': 2507.325, 'weight': 1, 'content': [{'end': 2510.568, 'text': "One we don't consider OpenPilot to be 1.0 until we have 100% driver monitoring.", 'start': 2507.325, 'duration': 3.243}, {'end': 2515.009, 'text': 'You can cheat right now, our driver monitoring system.', 'start': 2512.987, 'duration': 2.022}, {'end': 2515.87, 'text': "There's a few ways to cheat it.", 'start': 2515.029, 'duration': 0.841}, {'end': 2516.871, 'text': "They're pretty obvious.", 'start': 2516.09, 'duration': 0.781}, {'end': 2519.494, 'text': "We're working on making that better.", 'start': 2518.233, 'duration': 1.261}, {'end': 2524.86, 'text': "Before we ship a consumer product that can drive cars, I want to make sure that I have driver monitoring that you can't cheat.", 'start': 2519.754, 'duration': 5.106}, {'end': 2532.525, 'text': "What's a successful driver monitoring system look like? Is it all about just keeping your eyes on the road? Well, a few things.", 'start': 2525.521, 'duration': 7.004}, {'end': 2536.471, 'text': "So that's what we went with at first for driver monitoring.", 'start': 2532.946, 'duration': 3.525}], 'summary': 'Openpilot aims for 100% driver monitoring to prevent cheating and ensure safe autonomous driving.', 'duration': 29.146, 'max_score': 2507.325, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/iwcYp-XT7UI/pics/iwcYp-XT7UI2507325.jpg'}, {'end': 2790.463, 'src': 'embed', 'start': 2761.39, 'weight': 3, 'content': [{'end': 2764.853, 'text': "Hopefully, he's like, okay, I was wrong.", 'start': 2761.39, 'duration': 3.463}, {'end': 2765.894, 'text': "I'm going to add driver monitoring.", 'start': 2764.913, 'duration': 0.981}, {'end': 2771.918, 'text': "Because when these systems get to the point that they're only messing up once every thousand miles, you absolutely need driver monitoring.", 'start': 2766.294, 'duration': 5.624}, {'end': 2777.202, 'text': "So let me play, because I agree with you, but let me play devil's advocate.", 'start': 2774.16, 'duration': 3.042}, {'end': 2787.401, 'text': 'One possibility is that without driver monitoring, people are able to self-regulate, monitor themselves.', 'start': 2777.582, 'duration': 9.819}, {'end': 2790.463, 'text': 'So your idea is..', 'start': 2787.421, 'duration': 3.042}], 'summary': 'Driver monitoring needed for safety, 1 error per 1000 miles', 'duration': 29.073, 'max_score': 2761.39, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/iwcYp-XT7UI/pics/iwcYp-XT7UI2761390.jpg'}, {'end': 2899.87, 'src': 'embed', 'start': 2854.097, 'weight': 4, 'content': [{'end': 2860.521, 'text': "But all those elements, even if you don't have driver monitoring, that whole experience needs to be better.", 'start': 2854.097, 'duration': 6.424}, {'end': 2868.086, 'text': "Driver monitoring, I think would make, I mean, I think Super Cruise is a better experience once it's engaged over autopilot.", 'start': 2861.121, 'duration': 6.965}, {'end': 2873.769, 'text': "I think Super Cruise's transition to engagement and disengagement are significantly worse.", 'start': 2868.406, 'duration': 5.363}, {'end': 2880.734, 'text': "There's a tricky thing, because if I were to criticize Super Cruise, it's a little too crude.", 'start': 2875.169, 'duration': 5.565}, {'end': 2883.576, 'text': "And I think it's like six seconds or something.", 'start': 2880.794, 'duration': 2.782}, {'end': 2885.518, 'text': "If you look off road, it'll start warning you.", 'start': 2883.656, 'duration': 1.862}, {'end': 2888.18, 'text': "It's some ridiculously long period of time.", 'start': 2886.079, 'duration': 2.101}, {'end': 2895.587, 'text': "And just the way, I think it's basically, it's a binary.", 'start': 2889.201, 'duration': 6.386}, {'end': 2896.708, 'text': 'It should be adapted.', 'start': 2895.827, 'duration': 0.881}, {'end': 2898.029, 'text': 'Yeah, it, yeah.', 'start': 2896.748, 'duration': 1.281}, {'end': 2899.87, 'text': 'it needs to learn more about you.', 'start': 2898.029, 'duration': 1.841}], 'summary': 'Enhancing driver monitoring can improve super cruise experience, with a need for better transition and adaptation, as well as a longer warning period.', 'duration': 45.773, 'max_score': 2854.097, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/iwcYp-XT7UI/pics/iwcYp-XT7UI2854097.jpg'}, {'end': 2948.451, 'src': 'embed', 'start': 2921.465, 'weight': 7, 'content': [{'end': 2924.489, 'text': "We've renamed the driver monitoring packet to driver state.", 'start': 2921.465, 'duration': 3.024}, {'end': 2925.89, 'text': 'Driver state.', 'start': 2925.33, 'duration': 0.56}, {'end': 2932.118, 'text': 'We have car state packet, which has the state of the car, and we have driver state packet, which has the state of the driver.', 'start': 2926.271, 'duration': 5.847}, {'end': 2933.099, 'text': 'Estimate their BAC.', 'start': 2932.138, 'duration': 0.961}, {'end': 2935.582, 'text': "What's BAC? Blood alcohol conduct.", 'start': 2934.08, 'duration': 1.502}, {'end': 2939.627, 'text': "You think that's possible with computer vision? Absolutely.", 'start': 2937.324, 'duration': 2.303}, {'end': 2944.648, 'text': "To me, it's an open question.", 'start': 2943.387, 'duration': 1.261}, {'end': 2946.029, 'text': "I haven't looked into it too much.", 'start': 2944.668, 'duration': 1.361}, {'end': 2948.451, 'text': 'Actually, I quite seriously looked at the literature.', 'start': 2946.049, 'duration': 2.402}], 'summary': 'Renamed driver monitoring packet to driver state, with focus on estimating driver bac using computer vision.', 'duration': 26.986, 'max_score': 2921.465, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/iwcYp-XT7UI/pics/iwcYp-XT7UI2921465.jpg'}, {'end': 3007.489, 'src': 'embed', 'start': 2978.346, 'weight': 8, 'content': [{'end': 2980.23, 'text': 'What are the main problems that still need to be solved?', 'start': 2978.346, 'duration': 1.884}, {'end': 2985.46, 'text': "We're hopefully fixing a few of them in 06..", 'start': 2981.879, 'duration': 3.581}, {'end': 2988.201, 'text': "We're not as good as autopilot at stop cars.", 'start': 2985.46, 'duration': 2.741}, {'end': 2999.305, 'text': "So if you're coming up to a red light at 55, so it's the radar stopped car problem which is responsible for two autopilot accidents.", 'start': 2989.522, 'duration': 9.783}, {'end': 3003.367, 'text': "it's hard to differentiate a stopped car from a signpost.", 'start': 2999.305, 'duration': 4.062}, {'end': 3004.708, 'text': 'Yeah, a static object.', 'start': 3003.687, 'duration': 1.021}, {'end': 3006.088, 'text': 'So you have to fuse.', 'start': 3005.388, 'duration': 0.7}, {'end': 3007.489, 'text': 'You have to do this visually.', 'start': 3006.428, 'duration': 1.061}], 'summary': 'Improving autopilot to better detect stopped cars at high speeds', 'duration': 29.143, 'max_score': 2978.346, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/iwcYp-XT7UI/pics/iwcYp-XT7UI2978346.jpg'}, {'end': 3092.926, 'src': 'embed', 'start': 3064.397, 'weight': 9, 'content': [{'end': 3066.238, 'text': 'And our simulator is really cool.', 'start': 3064.397, 'duration': 1.841}, {'end': 3069.52, 'text': 'Our simulator is not like a Unity-based simulator.', 'start': 3066.518, 'duration': 3.002}, {'end': 3071.621, 'text': 'Our simulator lets us load in real states.', 'start': 3069.76, 'duration': 1.861}, {'end': 3079.922, 'text': 'What do you mean? We can load in a drive and simulate what the system would have done on the historical data.', 'start': 3072.918, 'duration': 7.004}, {'end': 3081.163, 'text': 'Ooh, nice.', 'start': 3080.342, 'duration': 0.821}, {'end': 3084.221, 'text': 'Interesting So what, yeah.', 'start': 3082.6, 'duration': 1.621}, {'end': 3089.024, 'text': "Right now we're only using it for testing, but as soon as we start using it for training, that's it.", 'start': 3084.361, 'duration': 4.663}, {'end': 3090.244, 'text': "That's all for testing.", 'start': 3089.184, 'duration': 1.06}, {'end': 3092.926, 'text': "What's your feeling about the real world versus simulation?", 'start': 3090.825, 'duration': 2.101}], 'summary': 'Simulator allows loading real states, historical data testing, ready for training.', 'duration': 28.529, 'max_score': 3064.397, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/iwcYp-XT7UI/pics/iwcYp-XT7UI3064397.jpg'}, {'end': 3410.094, 'src': 'embed', 'start': 3383.123, 'weight': 10, 'content': [{'end': 3389.908, 'text': "Yeah, I mean, certainly it's hard to argue with that because in the end that's going to be true.", 'start': 3383.123, 'duration': 6.785}, {'end': 3392.71, 'text': "Yeah, and I'll tell you what the output of our perception system is.", 'start': 3390.148, 'duration': 2.562}, {'end': 3394.511, 'text': "What's that? It's a 1024-dimensional vector.", 'start': 3392.75, 'duration': 1.761}, {'end': 3395.032, 'text': 'Oh, you know that.', 'start': 3394.572, 'duration': 0.46}, {'end': 3397.193, 'text': "No, it's the 1024 dimensions of who knows what.", 'start': 3395.052, 'duration': 2.141}, {'end': 3407.15, 'text': "because it's operating on real data.", 'start': 3403.547, 'duration': 3.603}, {'end': 3408.232, 'text': "That's the perception.", 'start': 3407.17, 'duration': 1.062}, {'end': 3410.094, 'text': "That's the perception state.", 'start': 3408.472, 'duration': 1.622}], 'summary': 'The output of the perception system is a 1024-dimensional vector.', 'duration': 26.971, 'max_score': 3383.123, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/iwcYp-XT7UI/pics/iwcYp-XT7UI3383123.jpg'}, {'end': 3480.197, 'src': 'embed', 'start': 3451.643, 'weight': 11, 'content': [{'end': 3470.162, 'text': "I think you're just being very poetic about expressing a fundamental problem of simulators that they're going to be missing so much that the feature vector will just look fundamentally different in the simulated world than the real world.", 'start': 3451.643, 'duration': 18.519}, {'end': 3473.475, 'text': "I'm not making a claim about simulators.", 'start': 3471.314, 'duration': 2.161}, {'end': 3478.716, 'text': "I'm making a claim about the spec division between perception and planning.", 'start': 3473.895, 'duration': 4.821}, {'end': 3479.236, 'text': 'And planning.', 'start': 3478.816, 'duration': 0.42}, {'end': 3480.197, 'text': 'Even in your system.', 'start': 3479.377, 'duration': 0.82}], 'summary': 'Simulators may miss crucial data, altering feature vector fundamentally.', 'duration': 28.554, 'max_score': 3451.643, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/iwcYp-XT7UI/pics/iwcYp-XT7UI3451643.jpg'}, {'end': 3552.527, 'src': 'embed', 'start': 3520.162, 'weight': 12, 'content': [{'end': 3522.264, 'text': "Because isn't this a full autonomy??", 'start': 3520.162, 'duration': 2.102}, {'end': 3524.106, 'text': 'Is this something you think about?', 'start': 3522.885, 'duration': 1.221}, {'end': 3526.889, 'text': 'Oh, I want to win self-driving cars.', 'start': 3524.126, 'duration': 2.763}, {'end': 3533.115, 'text': 'So your definition of win includes the full five.', 'start': 3527.629, 'duration': 5.486}, {'end': 3533.655, 'text': 'Level five.', 'start': 3533.155, 'duration': 0.5}, {'end': 3535.517, 'text': "I don't think level four is a real thing.", 'start': 3533.676, 'duration': 1.841}, {'end': 3537.98, 'text': 'I want to build the AlphaGo of driving.', 'start': 3535.898, 'duration': 2.082}, {'end': 3545.784, 'text': 'So AlphaGo is really end to end.', 'start': 3541.182, 'duration': 4.602}, {'end': 3549.746, 'text': "Yeah, it's end to end.", 'start': 3545.804, 'duration': 3.942}, {'end': 3552.527, 'text': 'And do you think this whole problem?', 'start': 3549.906, 'duration': 2.621}], 'summary': 'Aiming for level five autonomy in self-driving cars to achieve full end-to-end capability like alphago.', 'duration': 32.365, 'max_score': 3520.162, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/iwcYp-XT7UI/pics/iwcYp-XT7UI3520162.jpg'}], 'start': 2342.39, 'title': 'Challenges in autonomous driving', 'summary': "Covers tesla's lead in data collection and driver monitoring, the need for improvement in driver monitoring systems, limitations of openpilot, and challenges in perception and planning for autonomous driving, highlighting the need for end-to-end learning approaches.", 'chapters': [{'end': 2592.391, 'start': 2342.39, 'title': "Tesla's autonomy strategy", 'summary': "Discusses tesla's approach to autonomy, highlighting their lead in data collection and driver monitoring, which sets them ahead of other companies such as gm and ford in reaching level 5 autonomy.", 'duration': 250.001, 'highlights': ["Tesla's lead in data collection through real user experience puts them ahead in the race for level 5 autonomy, outpacing other companies like GM and Ford.", "The significance of driver monitoring to ensure safety and attention, with Tesla's focus on achieving 100% driver monitoring before releasing a consumer product for driving cars.", "Comparison of Tesla's approach to autonomy with other companies, emphasizing their gathering of data on a scale unmatched by competitors and the unique over-the-air software updates."]}, {'end': 2920.122, 'start': 2593.13, 'title': 'Driver monitoring innovations', 'summary': "Discusses the need for driver monitoring systems in autonomous vehicles, highlighting the inefficiencies of current systems and the potential for improvement, with a focus on super cruise and tesla's autopilot experience.", 'duration': 326.992, 'highlights': ["The need for driver monitoring systems in autonomous vehicles is emphasized, with a focus on the inefficiencies of current systems and the potential for improvement. The discussion highlights the need for driver monitoring systems in autonomous vehicles, pointing out the inefficiencies of current systems and the potential for improvement in the context of Super Cruise and Tesla's autopilot experience.", "The inefficiencies of Super Cruise's transition to engagement and disengagement are pointed out, indicating the need for smoother and more intuitive operation. The inefficiencies of Super Cruise's transition to engagement and disengagement are highlighted, emphasizing the need for smoother and more intuitive operation to enhance the user experience.", "Criticism of the crude nature of Super Cruise's warning system and the call for improved adaptability and communication with the driver are emphasized. The crude nature of Super Cruise's warning system is criticized, with a call for improved adaptability and communication with the driver to enhance safety and user experience.", "The shortcomings of Tesla's autopilot experience, particularly the annoyance of the wheel engagement requirement and the potential for improvement through driver monitoring, are discussed. Shortcomings of Tesla's autopilot experience, such as the annoyance of the wheel engagement requirement, are discussed, highlighting the potential for improvement through the implementation of driver monitoring systems."]}, {'end': 3300.344, 'start': 2921.465, 'title': 'Driver state monitoring and openpilot limitations', 'summary': "Discusses the possibility of estimating the driver's blood alcohol content (bac) using computer vision, the current limitations of openpilot such as difficulty in differentiating a stopped car from a signpost, and the use of a simulator for testing and training.", 'duration': 378.879, 'highlights': ["The chapter discusses the possibility of estimating the driver's Blood Alcohol Content (BAC) using computer vision. The speaker explores the potential of using computer vision to estimate the driver's BAC, highlighting the importance of detecting and addressing issues related to drunk, distracted, and asleep drivers.", 'The current limitations of OpenPilot are addressed, including the difficulty in differentiating a stopped car from a signpost. The limitations of OpenPilot are highlighted, particularly the challenge of distinguishing a stopped car from a signpost, which poses a safety concern at high speeds.', "The use of a simulator for testing and training is discussed, with an emphasis on its capability to simulate historical driving data. The speaker explains the use of a simulator for testing and training, highlighting its ability to simulate the system's behavior based on historical driving data, with potential for future training purposes."]}, {'end': 3597.418, 'start': 3300.364, 'title': 'Challenges in perception and planning', 'summary': 'Discusses the challenges in simulating real-world perception and planning for autonomous driving, emphasizing the limitations of hand-coded perception systems and advocating for an end-to-end learning approach to exceed human performance.', 'duration': 297.054, 'highlights': ['The output of our perception system is a 1024-dimensional vector, posing the challenge of hand-labeling all dimensions and indicating the complexity of real-world data. The complexity of the 1024-dimensional vector output of the perception system highlights the difficulty in interpreting and hand-labeling all dimensions, emphasizing the complexity of the real-world data that the system processes.', 'The discussion emphasizes the fundamental problem of simulators in missing essential real-world features, leading to a fundamentally different feature vector in the simulated world compared to the real world. The fundamental problem of simulators missing crucial real-world features and resulting in a fundamentally different feature vector in the simulated world compared to reality is highlighted, pointing out the challenges in accurately simulating real-world scenarios for autonomous driving.', 'The conversation advocates for an end-to-end learning approach, stating that it is not only the right way but also the only way to exceed human performance in autonomous driving. The emphasis on an end-to-end learning approach as the only way to surpass human performance in autonomous driving highlights the advocacy for a comprehensive learning methodology to address the complexity and challenges of real-world driving scenarios.']}], 'duration': 1255.028, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/iwcYp-XT7UI/pics/iwcYp-XT7UI2342390.jpg', 'highlights': ["Tesla's lead in data collection through real user experience puts them ahead in the race for level 5 autonomy, outpacing other companies like GM and Ford.", "The significance of driver monitoring to ensure safety and attention, with Tesla's focus on achieving 100% driver monitoring before releasing a consumer product for driving cars.", "Comparison of Tesla's approach to autonomy with other companies, emphasizing their gathering of data on a scale unmatched by competitors and the unique over-the-air software updates.", 'The need for driver monitoring systems in autonomous vehicles is emphasized, with a focus on the inefficiencies of current systems and the potential for improvement.', "The inefficiencies of Super Cruise's transition to engagement and disengagement are pointed out, indicating the need for smoother and more intuitive operation.", "Criticism of the crude nature of Super Cruise's warning system and the call for improved adaptability and communication with the driver are emphasized.", "The shortcomings of Tesla's autopilot experience, particularly the annoyance of the wheel engagement requirement and the potential for improvement through driver monitoring, are discussed.", "The chapter discusses the possibility of estimating the driver's Blood Alcohol Content (BAC) using computer vision.", 'The current limitations of OpenPilot are addressed, including the difficulty in differentiating a stopped car from a signpost.', 'The use of a simulator for testing and training is discussed, with an emphasis on its capability to simulate historical driving data.', 'The output of our perception system is a 1024-dimensional vector, posing the challenge of hand-labeling all dimensions and indicating the complexity of real-world data.', 'The discussion emphasizes the fundamental problem of simulators in missing essential real-world features, leading to a fundamentally different feature vector in the simulated world compared to the real world.', 'The conversation advocates for an end-to-end learning approach, stating that it is not only the right way but also the only way to exceed human performance in autonomous driving.']}, {'end': 4160.478, 'segs': [{'end': 3682.146, 'src': 'embed', 'start': 3653.029, 'weight': 4, 'content': [{'end': 3655.334, 'text': "You mentioned that you didn't like Navigate and Autopilot.", 'start': 3653.029, 'duration': 2.305}, {'end': 3656.616, 'text': 'What advice?', 'start': 3655.815, 'duration': 0.801}, {'end': 3657.418, 'text': 'how would you make it better?', 'start': 3656.616, 'duration': 0.802}, {'end': 3661.165, 'text': "Do you think, as a feature, that if it's done really well, it's a good feature?", 'start': 3657.758, 'duration': 3.407}, {'end': 3667.814, 'text': "I think that it's too reliant on hand-coded hacks.", 'start': 3662.429, 'duration': 5.385}, {'end': 3674.299, 'text': 'How does Navigate and Autopilot do a lane change? It actually does the same lane change every time, and it feels mechanical.', 'start': 3668.034, 'duration': 6.265}, {'end': 3675.32, 'text': 'Humans do different lane changes.', 'start': 3674.339, 'duration': 0.981}, {'end': 3678.363, 'text': 'Humans sometimes will do a slow one, sometimes do a fast one.', 'start': 3676.021, 'duration': 2.342}, {'end': 3682.146, 'text': 'Navigate and Autopilot, at least every time I used it, it did the identical lane change.', 'start': 3678.983, 'duration': 3.163}], 'summary': 'Navigate and autopilot needs improvement in lane change variety and less reliance on hand-coded hacks.', 'duration': 29.117, 'max_score': 3653.029, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/iwcYp-XT7UI/pics/iwcYp-XT7UI3653029.jpg'}, {'end': 3739.063, 'src': 'embed', 'start': 3713.027, 'weight': 0, 'content': [{'end': 3720.71, 'text': "So I feel that we have the solution for lateral, but we don't yet have the solution for longitudinal.", 'start': 3713.027, 'duration': 7.683}, {'end': 3723.151, 'text': "There's a few reasons longitudinal is harder than lateral.", 'start': 3720.75, 'duration': 2.401}, {'end': 3733.418, 'text': "The lane change component, The way that we train on it very simply is like our model has an input for whether it's doing a lane change or not.", 'start': 3723.491, 'duration': 9.927}, {'end': 3739.063, 'text': 'And then when we train the end to end model, we hand label all the lane changes because you have to.', 'start': 3734.159, 'duration': 4.904}], 'summary': 'Addressed lateral solution, working on longitudinal. lane change training component in progress.', 'duration': 26.036, 'max_score': 3713.027, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/iwcYp-XT7UI/pics/iwcYp-XT7UI3713027.jpg'}, {'end': 3813.227, 'src': 'embed', 'start': 3787.046, 'weight': 3, 'content': [{'end': 3793.15, 'text': "How do we label the ones that are natural and feel good? You know, that's the, cause that's your ultimate criticism.", 'start': 3787.046, 'duration': 6.104}, {'end': 3796.692, 'text': "The current, uh, Navigate and Autopilot just doesn't feel good.", 'start': 3793.55, 'duration': 3.142}, {'end': 3805.123, 'text': 'Well, the current Navigator on autopilot is a hand-coded policy written by an engineer in a room who probably went out and tested it a few times on the 280..', 'start': 3797.008, 'duration': 8.115}, {'end': 3808.965, 'text': 'probably a more, a better version of that, but yes.', 'start': 3805.123, 'duration': 3.842}, {'end': 3810.465, 'text': "That's how we would have written it.", 'start': 3809.465, 'duration': 1}, {'end': 3813.227, 'text': 'Yeah Maybe Tesla, they tested it.', 'start': 3811.106, 'duration': 2.121}], 'summary': 'The current navigator on autopilot is hand-coded, but feels unsatisfactory.', 'duration': 26.181, 'max_score': 3787.046, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/iwcYp-XT7UI/pics/iwcYp-XT7UI3787046.jpg'}, {'end': 4026.934, 'src': 'heatmap', 'start': 3934.247, 'weight': 1, 'content': [{'end': 3936.808, 'text': "The hope is that there's that the distribution is not bimodal.", 'start': 3934.247, 'duration': 2.561}, {'end': 3938.188, 'text': "The hope is that it's a nice galaxy.", 'start': 3936.968, 'duration': 1.22}, {'end': 3944.649, 'text': 'So what advice would you give to Tesla? How to fix, how to improve navigating autopilot?', 'start': 3939.128, 'duration': 5.521}, {'end': 3946.99, 'text': "The lessons that you've learned from calm AI?", 'start': 3945.05, 'duration': 1.94}, {'end': 3955.102, 'text': 'The only real advice I would give to Tesla is please put driver monitoring in your cars with respect to improving it.', 'start': 3948.277, 'duration': 6.825}, {'end': 3955.963, 'text': "You can't do that anymore.", 'start': 3955.122, 'duration': 0.841}, {'end': 3956.823, 'text': "I said I'd interrupt.", 'start': 3956.003, 'duration': 0.82}, {'end': 3965.77, 'text': "But, you know, there's a practical nature of many of hundreds of thousands of cars being produced that don't have a good driver facing camera.", 'start': 3957.324, 'duration': 8.446}, {'end': 3967.331, 'text': 'The Model 3 has a selfie cam.', 'start': 3965.79, 'duration': 1.541}, {'end': 3971.594, 'text': "Is it not good enough? Did they not put IR LEDs for night? That's a good question.", 'start': 3967.571, 'duration': 4.023}, {'end': 3975.336, 'text': "But I do know that it's fisheye and it's relatively low resolution.", 'start': 3971.654, 'duration': 3.682}, {'end': 3976.797, 'text': "So it's really not designed..", 'start': 3975.376, 'duration': 1.421}, {'end': 3977.557, 'text': "It wasn't..", 'start': 3976.797, 'duration': 0.76}, {'end': 3978.798, 'text': "It wasn't designed for drone models.", 'start': 3977.557, 'duration': 1.241}, {'end': 3983.9, 'text': 'You can hope that you can kind of scrape up and have something from it.', 'start': 3978.818, 'duration': 5.082}, {'end': 3985.58, 'text': 'Yeah But..', 'start': 3984.24, 'duration': 1.34}, {'end': 3988.242, 'text': "Why didn't they put it in today? Put it in today.", 'start': 3985.58, 'duration': 2.662}, {'end': 3988.882, 'text': 'Put it in today.', 'start': 3988.262, 'duration': 0.62}, {'end': 3995.299, 'text': "Every time I've heard Karpathy talk about the problem and talking about like software 2.0 and how the machine learning is gobbling up everything.", 'start': 3989.577, 'duration': 5.722}, {'end': 3997.24, 'text': 'I think this is absolutely the right strategy.', 'start': 3995.299, 'duration': 1.941}, {'end': 3999.781, 'text': "I think that he didn't write Navigate on autopilot.", 'start': 3997.5, 'duration': 2.281}, {'end': 4003.242, 'text': 'I think somebody else did and kind of hacked it on top of that stuff.', 'start': 4000.221, 'duration': 3.021}, {'end': 4005.323, 'text': 'I think when Karpathy says wait a second.', 'start': 4003.282, 'duration': 2.041}, {'end': 4008.304, 'text': 'Why did we hand code this lane change policy with all these magic numbers?', 'start': 4005.783, 'duration': 2.521}, {'end': 4009.205, 'text': "We're going to learn it from data.", 'start': 4008.324, 'duration': 0.881}, {'end': 4009.825, 'text': "They'll fix it.", 'start': 4009.365, 'duration': 0.46}, {'end': 4010.686, 'text': 'They already know what to do there.', 'start': 4009.845, 'duration': 0.841}, {'end': 4017.169, 'text': "Well, that's Andre's job is to turn everything into a learning problem and collect a huge amount of data.", 'start': 4011.106, 'duration': 6.063}, {'end': 4023.512, 'text': 'The reality is, though, not every problem can be turned into a learning problem in the short term.', 'start': 4017.609, 'duration': 5.903}, {'end': 4026.934, 'text': 'In the end, everything will be a learning problem.', 'start': 4024.232, 'duration': 2.702}], 'summary': 'Advise tesla to improve autopilot with driver monitoring for safety.', 'duration': 92.687, 'max_score': 3934.247, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/iwcYp-XT7UI/pics/iwcYp-XT7UI3934247.jpg'}, {'end': 4026.934, 'src': 'embed', 'start': 3997.5, 'weight': 2, 'content': [{'end': 3999.781, 'text': "I think that he didn't write Navigate on autopilot.", 'start': 3997.5, 'duration': 2.281}, {'end': 4003.242, 'text': 'I think somebody else did and kind of hacked it on top of that stuff.', 'start': 4000.221, 'duration': 3.021}, {'end': 4005.323, 'text': 'I think when Karpathy says wait a second.', 'start': 4003.282, 'duration': 2.041}, {'end': 4008.304, 'text': 'Why did we hand code this lane change policy with all these magic numbers?', 'start': 4005.783, 'duration': 2.521}, {'end': 4009.205, 'text': "We're going to learn it from data.", 'start': 4008.324, 'duration': 0.881}, {'end': 4009.825, 'text': "They'll fix it.", 'start': 4009.365, 'duration': 0.46}, {'end': 4010.686, 'text': 'They already know what to do there.', 'start': 4009.845, 'duration': 0.841}, {'end': 4017.169, 'text': "Well, that's Andre's job is to turn everything into a learning problem and collect a huge amount of data.", 'start': 4011.106, 'duration': 6.063}, {'end': 4023.512, 'text': 'The reality is, though, not every problem can be turned into a learning problem in the short term.', 'start': 4017.609, 'duration': 5.903}, {'end': 4026.934, 'text': 'In the end, everything will be a learning problem.', 'start': 4024.232, 'duration': 2.702}], 'summary': 'Discussion on turning problems into learning tasks, collecting data, and challenges in doing so.', 'duration': 29.434, 'max_score': 3997.5, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/iwcYp-XT7UI/pics/iwcYp-XT7UI3997500.jpg'}], 'start': 3597.418, 'title': "Tesla's autonomous driving challenges and ai improvements", 'summary': "Discusses challenges in developing tesla's autonomous driving technology, including the need to solve lane change learning problems and natural longitudinal driving difficulties. it also covers ai-based improvements for navigating autopilot, emphasizing data-driven lane change policy enhancement and the necessity of driver monitoring in tesla cars.", 'chapters': [{'end': 3771.328, 'start': 3597.418, 'title': "Tesla's autonomous driving challenges", 'summary': "Discusses the challenges in developing tesla's autonomous driving technology, highlighting the need to solve the learning problem of lane changes and the difficulty in achieving natural longitudinal driving, with a focus on the limitations of navigate and autopilot's lane change behavior and the approach to training the end-to-end model.", 'duration': 173.91, 'highlights': ["The difficulty in achieving natural longitudinal driving and the need to solve the learning problem of lane changes The chapter emphasizes the challenge of achieving natural longitudinal driving and the need to solve the learning problem of lane changes in developing Tesla's autonomous driving technology.", "Limitations of Navigate and Autopilot's lane change behavior The discussion highlights the mechanical nature of Navigate and Autopilot's lane change behavior, which fails to replicate the natural variability found in human driving.", 'Approach to training the end-to-end model for lane changes The chapter outlines the approach of training the end-to-end model for lane changes, including the hand labeling of training data and the use of an automatic ground truther to label lane changes.']}, {'end': 4160.478, 'start': 3772.438, 'title': 'Improving navigating autopilot with ai', 'summary': 'Discusses using ai to improve navigating autopilot, focusing on learning from data to enhance lane change policies and the need for driver monitoring in tesla cars.', 'duration': 388.04, 'highlights': ['The current Navigator on autopilot is a hand-coded policy written by an engineer, but the aim is to learn lane change policies from data to improve it. The current Navigator on autopilot is a hand-coded policy written by an engineer, aiming to learn lane change policies from data to improve it.', 'The need for driver monitoring in Tesla cars is emphasized as a crucial aspect for improvement. The need for driver monitoring in Tesla cars is emphasized as a crucial aspect for improvement.', 'The discussion challenges the traditional method of hand-coding policies with magic numbers and emphasizes the shift towards learning from data to improve performance. The discussion challenges the traditional method of hand-coding policies with magic numbers and emphasizes the shift towards learning from data to improve performance.', 'The speaker expresses skepticism about the timeline for achieving fully autonomous vehicles and highlights the need for a realistic approach. The speaker expresses skepticism about the timeline for achieving fully autonomous vehicles and highlights the need for a realistic approach.']}], 'duration': 563.06, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/iwcYp-XT7UI/pics/iwcYp-XT7UI3597418.jpg', 'highlights': ['The difficulty in achieving natural longitudinal driving and the need to solve the learning problem of lane changes', 'The need for driver monitoring in Tesla cars is emphasized as a crucial aspect for improvement', 'The discussion challenges the traditional method of hand-coding policies with magic numbers and emphasizes the shift towards learning from data to improve performance', 'The current Navigator on autopilot is a hand-coded policy written by an engineer, aiming to learn lane change policies from data to improve it', "Limitations of Navigate and Autopilot's lane change behavior"]}, {'end': 4581.387, 'segs': [{'end': 4186.029, 'src': 'embed', 'start': 4160.819, 'weight': 0, 'content': [{'end': 4167.184, 'text': 'If most of the streets are covered in Waymos, Waymo can kind of change the definition of what driving is.', 'start': 4160.819, 'duration': 6.365}, {'end': 4175.77, 'text': 'If your self-driving network is the majority of cars in an area, they only need to be safe with respect to each other,', 'start': 4168.524, 'duration': 7.246}, {'end': 4177.773, 'text': 'and all the humans will need to learn to adapt to them.', 'start': 4175.77, 'duration': 2.003}, {'end': 4181.167, 'text': 'now go drive in downtown new york.', 'start': 4178.765, 'duration': 2.402}, {'end': 4186.029, 'text': "oh yeah, that's i mean already you can talk about autonomy and like like on farms.", 'start': 4181.167, 'duration': 4.862}], 'summary': 'Waymo aims to redefine driving with majority self-driving cars, requiring humans to adapt.', 'duration': 25.21, 'max_score': 4160.819, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/iwcYp-XT7UI/pics/iwcYp-XT7UI4160819.jpg'}, {'end': 4269.238, 'src': 'embed', 'start': 4206.394, 'weight': 1, 'content': [{'end': 4210.417, 'text': 'Yeah, but you must drink champagne every year to celebrate.', 'start': 4206.394, 'duration': 4.023}, {'end': 4220.623, 'text': "So what is a good, what are some wins? A big milestone that we're hoping for by mid next year is profitability of the company.", 'start': 4210.457, 'duration': 10.166}, {'end': 4233.822, 'text': "Um, and we're going to have to revisit the idea of selling a consumer product, but it's not going to be like the comma one when we do it.", 'start': 4222.45, 'duration': 11.372}, {'end': 4234.943, 'text': "it's going to be perfect.", 'start': 4233.822, 'duration': 1.121}, {'end': 4238.881, 'text': 'OpenPilot has gotten so much better in the last two years.', 'start': 4236.3, 'duration': 2.581}, {'end': 4241.663, 'text': "We're going to have a few features.", 'start': 4239.622, 'duration': 2.041}, {'end': 4243.183, 'text': "We're going to have 100% driver monitoring.", 'start': 4241.683, 'duration': 1.5}, {'end': 4245.745, 'text': "We're going to disable no safety features in the car.", 'start': 4243.784, 'duration': 1.961}, {'end': 4248.946, 'text': "Actually, I think it'd be really cool what we're doing right now.", 'start': 4247.145, 'duration': 1.801}, {'end': 4254.189, 'text': "Our project this week is we're analyzing the data set and looking for all the AEB triggers from the manufacturer systems.", 'start': 4249.186, 'duration': 5.003}, {'end': 4258.671, 'text': 'We have a better data set on that than the manufacturers.', 'start': 4255.629, 'duration': 3.042}, {'end': 4261.455, 'text': 'How, how much does how many does Toyota have?', 'start': 4259.454, 'duration': 2.001}, {'end': 4264.936, 'text': '10 million miles of real world driving to know how many times their AEB triggered?', 'start': 4261.455, 'duration': 3.481}, {'end': 4267.437, 'text': 'So let me give you, cuz you asked right?', 'start': 4265.316, 'duration': 2.121}, {'end': 4269.238, 'text': 'Um, financial advice.', 'start': 4267.777, 'duration': 1.461}], 'summary': 'Targeting profitability by mid next year, enhancing openpilot features, and analyzing aeb triggers from a superior data set.', 'duration': 62.844, 'max_score': 4206.394, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/iwcYp-XT7UI/pics/iwcYp-XT7UI4206394.jpg'}, {'end': 4316.793, 'src': 'embed', 'start': 4295.486, 'weight': 5, 'content': [{'end': 4306.569, 'text': 'but you could do it for significant money purposes and make the world a better place by creating a consortia where automakers would pay in and then they get to have free access to the data.', 'start': 4295.486, 'duration': 11.083}, {'end': 4314.052, 'text': 'And I think a lot of people are really hungry for that and would pay significant amount of money for it.', 'start': 4306.99, 'duration': 7.062}, {'end': 4315.132, 'text': "Here's the problem with that.", 'start': 4314.212, 'duration': 0.92}, {'end': 4316.793, 'text': 'I like this idea all in theory.', 'start': 4315.412, 'duration': 1.381}], 'summary': 'Proposal to create a consortia for automakers to pay and access data for significant money purposes and betterment of the world.', 'duration': 21.307, 'max_score': 4295.486, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/iwcYp-XT7UI/pics/iwcYp-XT7UI4295486.jpg'}, {'end': 4447.067, 'src': 'embed', 'start': 4418.587, 'weight': 6, 'content': [{'end': 4420.789, 'text': "If they reached into me, I'm not going to ignore the email.", 'start': 4418.587, 'duration': 2.202}, {'end': 4426.075, 'text': "I'll come back with something like, yeah, if you're willing to pay 100K a month for access to the data, I'm happy to set that up.", 'start': 4420.829, 'duration': 5.246}, {'end': 4427.216, 'text': "That's worth my engineering time.", 'start': 4426.115, 'duration': 1.101}, {'end': 4429.538, 'text': "That's actually quite insightful of you.", 'start': 4428.217, 'duration': 1.321}, {'end': 4429.938, 'text': "You're right.", 'start': 4429.578, 'duration': 0.36}, {'end': 4435.921, 'text': 'Probably because many of the automakers are quite a bit old school, there will be a need to reach out.', 'start': 4430.518, 'duration': 5.403}, {'end': 4439.783, 'text': 'And they want it, but there will need to be some communication.', 'start': 4436.382, 'duration': 3.401}, {'end': 4440.064, 'text': "You're right.", 'start': 4439.803, 'duration': 0.261}, {'end': 4447.067, 'text': 'Mobileye circa 2015 had the lowest R&D spend of any chip maker.', 'start': 4440.304, 'duration': 6.763}], 'summary': 'Offering data access for 100k/month, valuable engineering time, old school automakers need to be approached, lowest r&d spend by mobileye in 2015.', 'duration': 28.48, 'max_score': 4418.587, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/iwcYp-XT7UI/pics/iwcYp-XT7UI4418587.jpg'}, {'end': 4568.112, 'src': 'embed', 'start': 4479.704, 'weight': 8, 'content': [{'end': 4483.027, 'text': "It's years and years ahead of everybody else's.", 'start': 4479.704, 'duration': 3.323}, {'end': 4483.627, 'text': "Maybe not Tesla's.", 'start': 4483.047, 'duration': 0.58}, {'end': 4485.329, 'text': 'I think Tesla has similar stuff to us, actually.', 'start': 4483.727, 'duration': 1.602}, {'end': 4492.555, 'text': "I think Tesla has similar stuff, but when you compare it to what the Toyota Research Institute has, you're not even close to what we have.", 'start': 4485.769, 'duration': 6.786}, {'end': 4498.058, 'text': "No comment, but I also can't, I have to take your comments.", 'start': 4493.576, 'duration': 4.482}, {'end': 4506.223, 'text': 'I intuitively believe you, but I have to take it with a grain of salt, because I mean you are an inspiration,', 'start': 4498.519, 'duration': 7.704}, {'end': 4510.225, 'text': "because you basically don't care about a lot of things that other companies care about.", 'start': 4506.223, 'duration': 4.002}, {'end': 4516.628, 'text': "You don't try to bullshit in a sense, like make up stuff.", 'start': 4510.945, 'duration': 5.683}, {'end': 4521.751, 'text': "So to drive up valuation, you're really very real and you're trying to solve the problem and admire that a lot.", 'start': 4516.688, 'duration': 5.063}, {'end': 4531.999, 'text': "What I don't necessarily fully can't trust you on, with all due respect, is how good it is, right? I can only, but I also know how bad others are.", 'start': 4522.311, 'duration': 9.688}, {'end': 4537.583, 'text': "And so- I'll say two things about, trust but verify, right? I'll say two things about that.", 'start': 4532.019, 'duration': 5.564}, {'end': 4545.189, 'text': 'One is try, get in a 2020 Corolla and try OpenPilot 0.6 when it comes out next month.', 'start': 4538.224, 'duration': 6.965}, {'end': 4551.015, 'text': "I think already you'll look at this and you'll be like, this is already really good.", 'start': 4546.731, 'duration': 4.284}, {'end': 4557.061, 'text': 'And then I could be doing that all with hand labelers and all with like the same approach that like Mobileye uses.', 'start': 4551.416, 'duration': 5.645}, {'end': 4568.112, 'text': 'When we release a model that no longer has the lanes in it, that only outputs a path, then think about how we did that machine learning.', 'start': 4558.001, 'duration': 10.111}], 'summary': "Toyota research institute's technology surpasses tesla's, with a challenge to test openpilot 0.6 in a 2020 corolla.", 'duration': 88.408, 'max_score': 4479.704, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/iwcYp-XT7UI/pics/iwcYp-XT7UI4479704.jpg'}], 'start': 4160.819, 'title': "Waymo's and openpilot's impact on driving", 'summary': "Discusses waymo's potential to redefine driving with a majority of self-driving cars, and openpilot's progress with 100% driver monitoring and data monetization opportunities, including comma ai's profitability goal and plans to sell a new consumer product.", 'chapters': [{'end': 4234.943, 'start': 4160.819, 'title': "Waymo's impact on future driving", 'summary': "Discusses waymo's potential to redefine driving by having a majority of self-driving cars in an area, emphasizing the need for human adaptation. comma ai aims for profitability by the next year and plans to sell a consumer product different from comma one.", 'duration': 74.124, 'highlights': ["Waymo's potential to redefine driving by having a majority of self-driving cars in an area, emphasizing the need for human adaptation.", "Comma AI's aim for profitability by the next year.", 'Plans to sell a consumer product different from Comma One.']}, {'end': 4581.387, 'start': 4236.3, 'title': 'Openpilot progress and data monetization', 'summary': 'Discusses the significant improvements in openpilot over the last two years, with 100% driver monitoring and the potential for monetizing data by selling access to automakers, highlighting the challenges and opportunities in doing so.', 'duration': 345.087, 'highlights': ['OpenPilot now offers 100% driver monitoring, showcasing significant progress in the last two years.', "The project involves analyzing a data set to identify AEB triggers, where the team claims to have a better data set than the manufacturers, surpassing Toyota's 10 million miles of real-world driving data for AEB triggers.", 'Suggests the monetization of data by creating a consortia where automakers would pay to access the data, potentially generating significant revenue.', 'Expresses reluctance to engage in extensive business negotiations with automakers, highlighting the need for a simplified and direct approach to selling data.', 'Shares the idea of selling data subscriptions to automakers for 100k a month, indicating a potential revenue model for the data.', 'Acknowledges the challenges of communication with traditional automakers and the need to navigate their old-school approach, highlighting the difficulties in working with such companies.', "Expresses confidence in the advanced technology of OpenPilot, suggesting that it is years ahead of others, except possibly Tesla, and far superior to the Toyota Research Institute's capabilities.", 'Emphasizes the realism and straightforward approach of the company, applauding the focus on solving problems rather than inflating valuations.', 'Challenges others to verify the quality of OpenPilot by testing it in a 2020 Corolla, and hints at the advanced machine learning approach used, showcasing confidence in the technology.', 'Highlights the upcoming release of a model in OpenPilot that only outputs a path, emphasizing the sophistication of the machine learning approach and the accuracy of the claims made.']}], 'duration': 420.568, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/iwcYp-XT7UI/pics/iwcYp-XT7UI4160819.jpg', 'highlights': ['Waymo aims to redefine driving with a majority of self-driving cars, emphasizing the need for human adaptation.', 'Comma AI targets profitability by the next year.', 'Comma AI plans to sell a new consumer product different from Comma One.', 'OpenPilot now offers 100% driver monitoring, showcasing significant progress in the last two years.', "The project involves analyzing a data set to identify AEB triggers, surpassing Toyota's 10 million miles of real-world driving data for AEB triggers.", 'Suggests the monetization of data by creating a consortia where automakers would pay to access the data, potentially generating significant revenue.', 'Expresses reluctance to engage in extensive business negotiations with automakers, highlighting the need for a simplified and direct approach to selling data.', 'Shares the idea of selling data subscriptions to automakers for 100k a month, indicating a potential revenue model for the data.', "Expresses confidence in the advanced technology of OpenPilot, suggesting that it is years ahead of others, except possibly Tesla, and far superior to the Toyota Research Institute's capabilities.", 'Emphasizes the realism and straightforward approach of the company, applauding the focus on solving problems rather than inflating valuations.', 'Challenges others to verify the quality of OpenPilot by testing it in a 2020 Corolla, showcasing confidence in the technology.', 'Highlights the upcoming release of a model in OpenPilot that only outputs a path, emphasizing the sophistication of the machine learning approach and the accuracy of the claims made.']}, {'end': 5849.375, 'segs': [{'end': 4632.434, 'src': 'embed', 'start': 4582.989, 'weight': 1, 'content': [{'end': 4593.821, 'text': 'I listened to your talk with Kyle, and Kyle was originally building the aftermarket system, and he gave up on it because of technical challenges.', 'start': 4582.989, 'duration': 10.832}, {'end': 4599.108, 'text': "Because of the fact that he's going to have to support 20 to 50 cars.", 'start': 4594.983, 'duration': 4.125}, {'end': 4600.55, 'text': 'We support 45.', 'start': 4599.188, 'duration': 1.362}, {'end': 4606.096, 'text': 'Because what is he going to do when the manufacturer ABS system triggers? We have alerts and warnings to deal with all of that and all the cars.', 'start': 4600.55, 'duration': 5.546}, {'end': 4610.222, 'text': 'And how is he going to formally verify it? Well, I got 10 million miles of data.', 'start': 4606.697, 'duration': 3.525}, {'end': 4612.064, 'text': "It's probably better verified than the spec.", 'start': 4610.302, 'duration': 1.762}, {'end': 4616.605, 'text': "I'm glad you're here talking to me.", 'start': 4613.723, 'duration': 2.882}, {'end': 4621.067, 'text': "I'll remember this day because it's interesting.", 'start': 4616.625, 'duration': 4.442}, {'end': 4627.771, 'text': "If you look at Kyle's from Cruise, I'm sure they have a large number of business development folks.", 'start': 4621.127, 'duration': 6.644}, {'end': 4629.732, 'text': "He's working with GM.", 'start': 4628.111, 'duration': 1.621}, {'end': 4632.434, 'text': 'He could work with Argo AI, working with Ford.', 'start': 4630.212, 'duration': 2.222}], 'summary': 'Kyle faced technical challenges with aftermarket system for 20-50 cars. our system supports 45 cars with alerts and warnings for abs triggers, and 10 million miles of data for formal verification.', 'duration': 49.445, 'max_score': 4582.989, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/iwcYp-XT7UI/pics/iwcYp-XT7UI4582989.jpg'}, {'end': 4691.674, 'src': 'embed', 'start': 4659.649, 'weight': 3, 'content': [{'end': 4663.193, 'text': "ways to make money or create value that you don't expect will pop up.", 'start': 4659.649, 'duration': 3.544}, {'end': 4667.938, 'text': "Oh, I've known how to do it since kind of 2017 is the first time I said it.", 'start': 4664.014, 'duration': 3.924}, {'end': 4672.028, 'text': 'Which part to not do, how to do which part? Our long-term plan is to be a car insurance company.', 'start': 4668.667, 'duration': 3.361}, {'end': 4673.709, 'text': 'Insurance Yeah, I love it.', 'start': 4672.548, 'duration': 1.161}, {'end': 4674.849, 'text': 'Yep, yep.', 'start': 4674.069, 'duration': 0.78}, {'end': 4676.389, 'text': 'I make driving twice as safe.', 'start': 4675.309, 'duration': 1.08}, {'end': 4679.71, 'text': 'Not only that, I have the best data set to know who statistically is the safest drivers.', 'start': 4676.65, 'duration': 3.06}, {'end': 4683.652, 'text': 'And oh, oh, we see you driving unsafely.', 'start': 4680.091, 'duration': 3.561}, {'end': 4684.712, 'text': "We're not going to insure you.", 'start': 4683.732, 'duration': 0.98}, {'end': 4691.674, 'text': "And that causes a bifurcation in the market because the only people who can't get common insurance are the bad drivers.", 'start': 4685.372, 'duration': 6.302}], 'summary': 'Plan to disrupt car insurance market by identifying and insuring safest drivers.', 'duration': 32.025, 'max_score': 4659.649, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/iwcYp-XT7UI/pics/iwcYp-XT7UI4659649.jpg'}, {'end': 4931.052, 'src': 'embed', 'start': 4888.79, 'weight': 0, 'content': [{'end': 4890.41, 'text': "I'll say nice things about both Waymo and Cruise.", 'start': 4888.79, 'duration': 1.62}, {'end': 4893.591, 'text': "Let's do it.", 'start': 4892.511, 'duration': 1.08}, {'end': 4894.211, 'text': 'Nice is good.', 'start': 4893.691, 'duration': 0.52}, {'end': 4898.453, 'text': 'Waymo is by far the furthest along with technology.', 'start': 4895.912, 'duration': 2.541}, {'end': 4902.614, 'text': 'Waymo has a three to five-year lead on all the competitors.', 'start': 4899.393, 'duration': 3.221}, {'end': 4909.499, 'text': 'If the Waymo looking stack works, maybe three-year lead.', 'start': 4904.115, 'duration': 5.384}, {'end': 4912.221, 'text': 'If the Waymo looking stack works, they have a three-year lead.', 'start': 4909.819, 'duration': 2.402}, {'end': 4919.925, 'text': 'Now, I argue that Waymo has spent too much money to recapitalize, to gain back their losses in those three years.', 'start': 4912.961, 'duration': 6.964}, {'end': 4923.027, 'text': 'Also, self-driving cars have no network effect like that.', 'start': 4920.266, 'duration': 2.761}, {'end': 4924.728, 'text': 'Uber has a network effect.', 'start': 4923.748, 'duration': 0.98}, {'end': 4925.369, 'text': 'You have a market.', 'start': 4924.848, 'duration': 0.521}, {'end': 4926.95, 'text': 'You have drivers and you have riders.', 'start': 4925.529, 'duration': 1.421}, {'end': 4929.691, 'text': 'Self-driving cars, you have capital and you have riders.', 'start': 4927.23, 'duration': 2.461}, {'end': 4931.052, 'text': "There's no network effect.", 'start': 4929.952, 'duration': 1.1}], 'summary': 'Waymo is ahead by 3-5 years in technology, but self-driving cars lack network effect.', 'duration': 42.262, 'max_score': 4888.79, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/iwcYp-XT7UI/pics/iwcYp-XT7UI4888790.jpg'}, {'end': 5288.658, 'src': 'embed', 'start': 5256.117, 'weight': 2, 'content': [{'end': 5259.298, 'text': 'The static driving problem is assuming you are the only car on the road.', 'start': 5256.117, 'duration': 3.181}, {'end': 5263.039, 'text': 'And this problem can be solved 100% with mapping and localization.', 'start': 5260.178, 'duration': 2.861}, {'end': 5265.3, 'text': 'This is why farms work the way they do.', 'start': 5263.999, 'duration': 1.301}, {'end': 5270.124, 'text': 'If all you have to deal with is the static problem and you can statically schedule your machines right?', 'start': 5265.82, 'duration': 4.304}, {'end': 5272.446, 'text': "It's the same as like statically scheduling processes.", 'start': 5270.144, 'duration': 2.302}, {'end': 5277.169, 'text': 'You can statically schedule your tractors to never hit each other on their paths, right? Because you know the speed they go at.', 'start': 5272.706, 'duration': 4.463}, {'end': 5279.791, 'text': "So that's the static driving problem.", 'start': 5277.55, 'duration': 2.241}, {'end': 5282.914, 'text': 'Maps only helps you with the static driving problem.', 'start': 5280.292, 'duration': 2.622}, {'end': 5288.658, 'text': "Yeah The question about static driving, you've just made it sound like it's really easy.", 'start': 5284.195, 'duration': 4.463}], 'summary': 'Solving the static driving problem with mapping and localization, ensuring 100% avoidance of collisions, as seen in farming operations.', 'duration': 32.541, 'max_score': 5256.117, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/iwcYp-XT7UI/pics/iwcYp-XT7UI5256117.jpg'}, {'end': 5364.962, 'src': 'embed', 'start': 5338.662, 'weight': 7, 'content': [{'end': 5344.119, 'text': "Building a perfect localizer without LiDAR, It's going to be hard.", 'start': 5338.662, 'duration': 5.457}, {'end': 5345.681, 'text': 'You can get 10 centimeters without LiDAR.', 'start': 5344.319, 'duration': 1.362}, {'end': 5346.962, 'text': 'You can get one centimeter with LiDAR.', 'start': 5345.701, 'duration': 1.261}, {'end': 5349.205, 'text': "I'm not even concerned about the one or 10 centimeters.", 'start': 5347.243, 'duration': 1.962}, {'end': 5352.288, 'text': "I'm concerned if every once in a while you're just way off.", 'start': 5349.265, 'duration': 3.023}, {'end': 5359.576, 'text': "Yeah So this is why you have to carefully make sure you're always tracking your position.", 'start': 5352.749, 'duration': 6.827}, {'end': 5364.962, 'text': 'You want to use LiDAR camera fusion, but you can get the reliability of that system up to..', 'start': 5360.036, 'duration': 4.926}], 'summary': 'Challenging to achieve precise localization without lidar, aiming for 10cm accuracy.', 'duration': 26.3, 'max_score': 5338.662, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/iwcYp-XT7UI/pics/iwcYp-XT7UI5338662.jpg'}, {'end': 5437.528, 'src': 'embed', 'start': 5415.011, 'weight': 9, 'content': [{'end': 5424.76, 'text': 'So you have to detect that car in real time and then you have to do the appropriate action, right? Also, that car is not a fixed object.', 'start': 5415.011, 'duration': 9.749}, {'end': 5427.342, 'text': 'That car may move, and you have to predict what that car will do.', 'start': 5424.8, 'duration': 2.542}, {'end': 5429.663, 'text': 'So this is the dynamic problem.', 'start': 5428.722, 'duration': 0.941}, {'end': 5432.285, 'text': 'So you have to deal with this.', 'start': 5431.464, 'duration': 0.821}, {'end': 5437.528, 'text': "This involves, again, you're going to need models of other people's behavior.", 'start': 5432.805, 'duration': 4.723}], 'summary': "Real-time car detection and prediction require dynamic problem-solving and modeling others' behavior.", 'duration': 22.517, 'max_score': 5415.011, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/iwcYp-XT7UI/pics/iwcYp-XT7UI5415011.jpg'}, {'end': 5615.986, 'src': 'embed', 'start': 5574.115, 'weight': 5, 'content': [{'end': 5579.399, 'text': "Gotcha But the problem is you first don't know very much, and so you're going to make mistakes.", 'start': 5574.115, 'duration': 5.284}, {'end': 5581.781, 'text': 'So the learning, the exploration happens through mistakes.', 'start': 5579.64, 'duration': 2.141}, {'end': 5582.322, 'text': "We're all ready.", 'start': 5581.801, 'duration': 0.521}, {'end': 5583.523, 'text': 'Yeah, but okay.', 'start': 5582.382, 'duration': 1.141}, {'end': 5585.504, 'text': 'So the consequences of a mistake.', 'start': 5583.823, 'duration': 1.681}, {'end': 5589.685, 'text': 'OpenPilot and Autopilot are making mistakes left and right.', 'start': 5586.164, 'duration': 3.521}, {'end': 5593.666, 'text': 'We have 700 daily active users, 1,000 weekly active users.', 'start': 5589.745, 'duration': 3.921}, {'end': 5598.108, 'text': 'OpenPilot makes tens of thousands of mistakes a week.', 'start': 5594.127, 'duration': 3.981}, {'end': 5600.589, 'text': 'These mistakes have zero consequences.', 'start': 5598.948, 'duration': 1.641}, {'end': 5606.759, 'text': 'These mistakes are, oh, I wanted to take this exit and it went straight.', 'start': 5601.195, 'duration': 5.564}, {'end': 5608.52, 'text': "So I'm just going to carefully touch the wheel.", 'start': 5606.819, 'duration': 1.701}, {'end': 5609.261, 'text': 'The humans catch them.', 'start': 5608.54, 'duration': 0.721}, {'end': 5610.262, 'text': 'The humans catch them.', 'start': 5609.281, 'duration': 0.981}, {'end': 5615.986, 'text': 'And the human disengagement is labeling that reinforcement learning in a completely consequence-free way.', 'start': 5610.742, 'duration': 5.244}], 'summary': 'Openpilot makes tens of thousands of mistakes a week with zero consequences, caught by human disengagement.', 'duration': 41.871, 'max_score': 5574.115, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/iwcYp-XT7UI/pics/iwcYp-XT7UI5574115.jpg'}], 'start': 4582.989, 'title': 'Challenges and innovations in autonomous vehicles', 'summary': 'Discusses challenges in building an aftermarket system for 20 to 50 cars, the competition between waymo and cruise in autonomous vehicles, breakthrough innovation in autonomous driving, and reinforcement learning in autonomous vehicles, with insights on technological lead, financial backing, strategic moves, and potential breakthrough innovations.', 'chapters': [{'end': 4857.618, 'start': 4582.989, 'title': 'Challenges in building an aftermarket system', 'summary': 'Discusses the challenges faced in building an aftermarket system for supporting 20 to 50 cars, the potential of using 10 million miles of data for verification, and the long-term plan of transforming into a car insurance company to improve driving safety and win the car insurance market.', 'duration': 274.629, 'highlights': ['The potential of using 10 million miles of data for verification The speaker mentions having 10 million miles of data and believes it to be better verified than the spec, showcasing a strong potential for verification.', 'Long-term plan to transform into a car insurance company The speaker discusses the plan to transform into a car insurance company, aiming to make driving twice as safe and offering lower premiums based on driving behavior, indicating a strategic long-term vision for the company.', 'Challenges faced in building an aftermarket system for supporting 20 to 50 cars The speaker highlights the technical challenges faced by Kyle in building an aftermarket system for supporting 20 to 50 cars, suggesting the complexity and scale of the task at hand.']}, {'end': 5205.397, 'start': 4859.11, 'title': 'Waymo vs. cruise in autonomous vehicles', 'summary': "Discusses the competition between waymo and cruise in the autonomous vehicle industry, highlighting waymo's technological lead of 3-5 years, financial backing, and the absence of network effects in self-driving cars, while also considering the first mover advantage of uber. it also emphasizes the challenges of creating value and beating uber, and the strategic move of gm in acquiring cruise as an insurance policy against waymo.", 'duration': 346.287, 'highlights': ["Waymo has a three to five-year lead on all the competitors in autonomous vehicle technology. Waymo's significant technological lead of 3-5 years over its competitors.", "Waymo's financial backing positions it as the only company likely to stand the 50-year journey to L5 autonomy. Highlighting Waymo's financial strength and its potential to outlast competitors in the journey to L5 autonomy.", "The absence of network effects in self-driving cars compared to Uber's model is emphasized, highlighting Uber's first mover advantage due to network effects. Emphasizing the absence of network effects in self-driving cars and highlighting Uber's first mover advantage due to network effects.", "GM's acquisition of Cruise for $1 billion is seen as an insurance policy against Waymo, highlighting Cruise's strategic value in providing a potential monopoly on autonomous vehicle technology. Highlighting GM's strategic move in acquiring Cruise as an insurance policy against Waymo, potentially securing a monopoly on autonomous vehicle technology."]}, {'end': 5573.695, 'start': 5205.437, 'title': 'Breakthrough innovation in autonomous driving', 'summary': 'Discusses the potential for breakthrough innovation in autonomous driving, particularly focusing on the static and dynamic driving problems, and the use of lidar technology for building a perfect localizer.', 'duration': 368.258, 'highlights': ['The static driving problem can be solved 100% with mapping and localization, especially with LiDAR and good HD maps. The static driving problem can be solved 100% with mapping and localization, especially with LiDAR and good HD maps, making it a feasible solution for autonomous driving.', 'The potential of using LiDAR technology to build a perfect localizer, achieving a high level of precision in tracking position, with the ability to get one centimeter accuracy. The potential of using LiDAR technology to build a perfect localizer, achieving a high level of precision in tracking position, with the ability to get one centimeter accuracy, showcasing the technological advancements in autonomous driving.', 'The discussion of the dynamic driving problem, which involves real-time detection of dynamic objects and predicting their behavior, highlighting the challenges in autonomous driving. The discussion of the dynamic driving problem, which involves real-time detection of dynamic objects and predicting their behavior, highlighting the challenges in autonomous driving and the need for advanced detection and prediction models.']}, {'end': 5849.375, 'start': 5574.115, 'title': 'Reinforcement learning in autonomous vehicles', 'summary': 'Delves into the challenges and strategies for implementing reinforcement learning in autonomous vehicles, highlighting the significance of human intervention and the number of daily and weekly active users, including tens of thousands of mistakes made by the system weekly, with zero consequences, and the messaging and safety model of the openpilot system.', 'duration': 275.26, 'highlights': ['OpenPilot makes tens of thousands of mistakes a week with zero consequences, caught by human intervention, highlighting the significance of human oversight in the reinforcement learning process. Tens of thousands of mistakes made by OpenPilot system weekly with zero consequences, emphasizing the role of human intervention in catching these mistakes.', 'The system has 700 daily active users and 1,000 weekly active users, providing quantifiable data on user engagement and system usage. OpenPilot system has 700 daily active users and 1,000 weekly active users, indicating user engagement and system usage.', "The messaging and safety model of the OpenPilot system are discussed, emphasizing the importance of clear communication and enforcing safety measures, including the level of the system and the need for new hardware for future upgrades. Discussion on the messaging and safety model of the OpenPilot system, highlighting the need for clear communication and the importance of safety measures, including the system's level and future hardware upgrades."]}], 'duration': 1266.386, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/iwcYp-XT7UI/pics/iwcYp-XT7UI4582989.jpg', 'highlights': ['Waymo has a three to five-year lead on all the competitors in autonomous vehicle technology.', 'The potential of using 10 million miles of data for verification.', 'The static driving problem can be solved 100% with mapping and localization, especially with LiDAR and good HD maps.', 'Long-term plan to transform into a car insurance company.', "The absence of network effects in self-driving cars compared to Uber's model is emphasized.", 'The system has 700 daily active users and 1,000 weekly active users.', 'Challenges faced in building an aftermarket system for supporting 20 to 50 cars.', 'The potential of using LiDAR technology to build a perfect localizer, achieving a high level of precision in tracking position, with the ability to get one centimeter accuracy.', 'OpenPilot makes tens of thousands of mistakes a week with zero consequences, caught by human intervention.', 'The discussion of the dynamic driving problem, which involves real-time detection of dynamic objects and predicting their behavior.']}, {'end': 7156.26, 'segs': [{'end': 5938.275, 'src': 'embed', 'start': 5867.405, 'weight': 0, 'content': [{'end': 5874.986, 'text': "I'm the president, but I'm one of those like unelected presidents of like a small dictatorship country, not one of those like elected presidents.", 'start': 5867.405, 'duration': 7.581}, {'end': 5878.067, 'text': "Oh, so you're like Putin when he was like the, yeah, I got you.", 'start': 5875.286, 'duration': 2.781}, {'end': 5882.092, 'text': "So what's the governance structure??", 'start': 5880.292, 'duration': 1.8}, {'end': 5887.254, 'text': "What's the future of CommAI as a business??", 'start': 5882.132, 'duration': 5.122}, {'end': 5896.376, 'text': 'Are you just focused on getting things right now, making some small amount of money in the meantime, and then, when it works,', 'start': 5887.274, 'duration': 9.102}, {'end': 5897.276, 'text': 'it works and you scale?', 'start': 5896.376, 'duration': 0.9}, {'end': 5902.177, 'text': 'Our burn rate is about 200K a month and our revenue is about 100K a month.', 'start': 5897.696, 'duration': 4.481}, {'end': 5907.858, 'text': "So we need to 4X our revenue, but we haven't tried very hard at that yet.", 'start': 5903.077, 'duration': 4.781}, {'end': 5910.106, 'text': 'And the revenue is basically selling stuff online.', 'start': 5908.223, 'duration': 1.883}, {'end': 5911.748, 'text': 'Yeah, we sell stuff at shop.com.ai.', 'start': 5910.146, 'duration': 1.602}, {'end': 5913.391, 'text': 'Is there other..', 'start': 5912.449, 'duration': 0.942}, {'end': 5913.791, 'text': 'Well, okay.', 'start': 5913.391, 'duration': 0.4}, {'end': 5915.754, 'text': "So you'll have to figure out..", 'start': 5913.951, 'duration': 1.803}, {'end': 5916.515, 'text': "That's our only..", 'start': 5915.754, 'duration': 0.761}, {'end': 5919.459, 'text': "See, but to me, that's like respectable revenues.", 'start': 5916.515, 'duration': 2.944}, {'end': 5922.604, 'text': 'We make it by selling products to consumers.', 'start': 5920.481, 'duration': 2.123}, {'end': 5924.687, 'text': "We're honest and transparent about what they are.", 'start': 5922.664, 'duration': 2.023}, {'end': 5928.729, 'text': 'most actually level four companies right?', 'start': 5926.187, 'duration': 2.542}, {'end': 5938.275, 'text': 'Because you could easily start blowing up like smoke, like overselling the hype and feeding into getting some fundraisers.', 'start': 5930.65, 'duration': 7.625}], 'summary': 'Commai aims to 4x revenue, with a burn rate of 200k and revenue of 100k a month, selling products at shop.com.ai.', 'duration': 70.87, 'max_score': 5867.405, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/iwcYp-XT7UI/pics/iwcYp-XT7UI5867405.jpg'}, {'end': 6017.116, 'src': 'embed', 'start': 5984.051, 'weight': 4, 'content': [{'end': 5991.115, 'text': "So are these like Elon Musk type characters? Or actually, you know what? I'll make it more broad than that.", 'start': 5984.051, 'duration': 7.064}, {'end': 5992.356, 'text': "I won't make it about a person.", 'start': 5991.295, 'duration': 1.061}, {'end': 5993.637, 'text': 'I respect skill.', 'start': 5992.716, 'duration': 0.921}, {'end': 5996.698, 'text': 'I respect people who have skills right?', 'start': 5994.117, 'duration': 2.581}, {'end': 6005.043, 'text': "And I would like to like be I'm not going to say famous, but be like known among more people who have like real skills.", 'start': 5996.998, 'duration': 8.045}, {'end': 6017.116, 'text': 'Who in cars do you think have skill, not do you respect? oh, Kyle Vogt has skill.', 'start': 6006.984, 'duration': 10.132}], 'summary': 'Seeking recognition for real skills, admires people like kyle vogt.', 'duration': 33.065, 'max_score': 5984.051, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/iwcYp-XT7UI/pics/iwcYp-XT7UI5984051.jpg'}, {'end': 6174.604, 'src': 'embed', 'start': 6143.838, 'weight': 5, 'content': [{'end': 6146.339, 'text': 'Another company that I really respect is Nauto.', 'start': 6143.838, 'duration': 2.501}, {'end': 6147.76, 'text': 'Actually, I respect their business model.', 'start': 6146.62, 'duration': 1.14}, {'end': 6153.324, 'text': 'Nauto sells a driver monitoring camera and they sell it to fleet owners.', 'start': 6148.401, 'duration': 4.923}, {'end': 6164.467, 'text': 'If I owned a fleet of cars and I could pay 40 bucks a month to monitor my employees, This is going to like reduce his accidents 18%.', 'start': 6154.004, 'duration': 10.463}, {'end': 6174.604, 'text': "Yeah, it's so like that in the space that is like the business model that I like most respect because they're creating value today.", 'start': 6164.467, 'duration': 10.137}], 'summary': 'Nauto sells driver monitoring cameras to fleet owners, reducing accidents by 18%.', 'duration': 30.766, 'max_score': 6143.838, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/iwcYp-XT7UI/pics/iwcYp-XT7UI6143838.jpg'}, {'end': 6343.637, 'src': 'embed', 'start': 6311.336, 'weight': 6, 'content': [{'end': 6314.078, 'text': 'we have a solution to the trolley problem at com.ai.', 'start': 6311.336, 'duration': 2.742}, {'end': 6318.101, 'text': 'well, so there is actually an alert in our code ethical dilemma detected.', 'start': 6314.078, 'duration': 4.023}, {'end': 6318.961, 'text': "it's not triggered yet.", 'start': 6318.101, 'duration': 0.86}, {'end': 6322.343, 'text': "we don't we don't know how yet to detect the ethical dilemmas, but we're a level two system.", 'start': 6318.961, 'duration': 3.382}, {'end': 6324.525, 'text': "so we're going to disengage and leave that decision to the humans.", 'start': 6322.343, 'duration': 2.182}, {'end': 6326.146, 'text': "You're such a troll.", 'start': 6325.305, 'duration': 0.841}, {'end': 6328.667, 'text': 'No, but the trolley problem deserves to be trolled.', 'start': 6326.686, 'duration': 1.981}, {'end': 6330.328, 'text': "Yeah, you're right.", 'start': 6328.747, 'duration': 1.581}, {'end': 6331.869, 'text': "That's a beautiful answer, actually.", 'start': 6330.408, 'duration': 1.461}, {'end': 6332.37, 'text': 'I know.', 'start': 6332.07, 'duration': 0.3}, {'end': 6336.492, 'text': 'I gave it to someone who was like, sometimes people ask, like you asked about the trolley problem.', 'start': 6332.49, 'duration': 4.002}, {'end': 6338.013, 'text': 'You can have a kind of discussion about it.', 'start': 6336.693, 'duration': 1.32}, {'end': 6343.637, 'text': "But when you get someone who's really earnest about it, because it's the kind of thing where,", 'start': 6338.034, 'duration': 5.603}], 'summary': 'Ai system detects ethical dilemmas and disengages, leaving decisions to humans.', 'duration': 32.301, 'max_score': 6311.336, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/iwcYp-XT7UI/pics/iwcYp-XT7UI6311336.jpg'}, {'end': 6492.883, 'src': 'embed', 'start': 6463.432, 'weight': 7, 'content': [{'end': 6467.095, 'text': "If you're designing systems that are unreliable, they're definitely not secure.", 'start': 6463.432, 'duration': 3.663}, {'end': 6472.627, 'text': 'Your car should always do something safe using its local sensors.', 'start': 6468.464, 'duration': 4.163}, {'end': 6475.129, 'text': 'And then the local sensor should be hardwired.', 'start': 6473.388, 'duration': 1.741}, {'end': 6480.473, 'text': 'And then could somebody hack into your canvas and turn your steering wheel on your brakes? Yes, but they could do it before AI too.', 'start': 6475.149, 'duration': 5.324}, {'end': 6484.596, 'text': "So let's think out of the box on some things.", 'start': 6480.533, 'duration': 4.063}, {'end': 6489.34, 'text': 'So do you think teleoperation has a role in any of this??', 'start': 6484.757, 'duration': 4.583}, {'end': 6492.883, 'text': 'So, remotely stepping in and controlling the cars??', 'start': 6489.38, 'duration': 3.503}], 'summary': 'Unreliable systems are not secure; local sensors for safety; potential for remote control.', 'duration': 29.451, 'max_score': 6463.432, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/iwcYp-XT7UI/pics/iwcYp-XT7UI6463432.jpg'}, {'end': 6829.807, 'src': 'embed', 'start': 6794.993, 'weight': 8, 'content': [{'end': 6796.975, 'text': 'And I seem to be pretty good at this stuff.', 'start': 6794.993, 'duration': 1.982}, {'end': 6798.175, 'text': "Let's try my best.", 'start': 6797.255, 'duration': 0.92}, {'end': 6805.84, 'text': "What's the worst that happens? But the best possible future I see is me sort of merging with the machine.", 'start': 6798.636, 'duration': 7.204}, {'end': 6811.263, 'text': 'And the way that I personify this is in a long-term monogamous relationship with a machine.', 'start': 6806.9, 'duration': 4.363}, {'end': 6820.859, 'text': "Oh, you don't think there's room for another human in your life if you really truly merge with another machine? I mean, I see merging.", 'start': 6812.992, 'duration': 7.867}, {'end': 6829.807, 'text': 'I see the best interface to my brain is the same relationship interface to merge with an AI.', 'start': 6820.959, 'duration': 8.848}], 'summary': 'Individual envisions merging with ai in a monogamous relationship.', 'duration': 34.814, 'max_score': 6794.993, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/iwcYp-XT7UI/pics/iwcYp-XT7UI6794993.jpg'}, {'end': 7086.178, 'src': 'embed', 'start': 7063.244, 'weight': 9, 'content': [{'end': 7070.548, 'text': 'Well, the ideal thing, mathematically you can go back to, like Schmidhuber, theories about this is to build a compressive model of the world,', 'start': 7063.244, 'duration': 7.304}, {'end': 7078.413, 'text': 'to build a maximally compressive to explore the world such that your exploration function maximizes the derivative of compression of the past.', 'start': 7070.548, 'duration': 7.865}, {'end': 7080.194, 'text': 'Schmidhuber has a paper about this.', 'start': 7078.934, 'duration': 1.26}, {'end': 7083.597, 'text': 'And like, I took that kind of as like a personal goal function.', 'start': 7080.815, 'duration': 2.782}, {'end': 7086.178, 'text': 'So what I mean to win?', 'start': 7085.418, 'duration': 0.76}], 'summary': "Building a maximally compressive model of the world for exploring, inspired by schmidhuber's theory.", 'duration': 22.934, 'max_score': 7063.244, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/iwcYp-XT7UI/pics/iwcYp-XT7UI7063244.jpg'}], 'start': 5849.835, 'title': 'Conversations on ai and tech', 'summary': 'Delves into leadership transition within calm ai, tech entrepreneur insights on revenue and ethical practices, ethical dilemmas in ai and autonomous vehicles, and the concept of ai girlfriends and the singularity, with predictions for the future and discussions on ethical, security, and business aspects.', 'chapters': [{'end': 5887.254, 'start': 5849.835, 'title': "Conversation about calm ai's future", 'summary': 'Discusses the leadership transition within calm ai, with the protagonist becoming the president, and the conversation revolves around the future governance structure and business prospects of calm ai.', 'duration': 37.419, 'highlights': ['The protagonist has transitioned to the role of president within Calm AI, resembling an unelected president of a small dictatorship country.', 'The conversation revolves around the future governance structure and business prospects of Calm AI, hinting at potential changes in leadership and strategic direction.']}, {'end': 6267.819, 'start': 5887.274, 'title': 'Tech entrepreneur insights', 'summary': "Covers a tech entrepreneur discussing revenue, burn rate, ethical business practices, and his respect for skilled individuals in the industry, emphasizing the importance of creating value in today's market.", 'duration': 380.545, 'highlights': ["The entrepreneur's company has a burn rate of 200K a month and a revenue of 100K a month, indicating a need to 4X the revenue. The burn rate of 200K a month and revenue of 100K a month highlight the financial situation of the entrepreneur's company, emphasizing the need to 4X the revenue.", 'The entrepreneur emphasizes the importance of honesty and transparency in selling products to consumers, expressing a dislike for overselling and hype. The entrepreneur values honesty and transparency in selling products to consumers and expresses disdain for overselling and hype in business practices.', 'The entrepreneur respects individuals with real skills and aims to be recognized among people with genuine expertise, mentioning specific names like Elon Musk, Kyle Vogt, and individuals at Waymo. The entrepreneur expresses respect for skilled individuals like Elon Musk, Kyle Vogt, and Waymo members, emphasizing his desire to be known among people with genuine expertise.', 'Nauto, a company the entrepreneur respects, sells driver monitoring cameras to fleet owners, reducing accidents by 18% and creating value in the market. The entrepreneur respects Nauto for its business model, which involves selling driver monitoring cameras to fleet owners, resulting in an 18% reduction in accidents and creating value in the market.', 'The entrepreneur spends significant time watching videos to learn from pedestrians and driver behaviors, highlighting the importance of continuous learning and observation in the tech industry. The entrepreneur emphasizes the significance of continuous learning and observation by spending time watching videos to learn from pedestrian and driver behaviors, showcasing the importance of ongoing education in the tech industry.']}, {'end': 6624.007, 'start': 6267.819, 'title': 'Ethical dilemmas in ai and autonomous vehicles', 'summary': 'Discusses ethical dilemmas in ai decision-making and the security concerns with autonomous vehicles, highlighting the trolley problem, ethical detection in ai, adversarial examples, gps spoofing, and the role of teleoperation and v2i communication in safety.', 'duration': 356.188, 'highlights': ['The trolley problem and ethical questions arise in AI decision-making, with a solution to detect ethical dilemmas at com.ai and disengage to leave decisions to humans.', 'Security concerns include adversarial examples affecting human behavior and GPS spoofing fooling Navigate on Autopilot, highlighting the importance of reliable local sensors in autonomous vehicles.', 'V2V communication reliance on accurate communication and the preference for V2I communication for non-safety critical functions in autonomous vehicles.', 'The importance of balancing hard work and relaxation, with a focus on working smarter, not harder, to maximize results and finding meaning in work.', 'The discussion on personal relationships as another source of meaning beyond work.', 'The chapter also delves into the role of teleoperation in the safety of autonomous vehicles and the potential limitations of constant link requirements.']}, {'end': 7156.26, 'start': 6624.587, 'title': 'Ai girlfriends and the singularity', 'summary': 'Discusses the concept of ai girlfriends, the potential for merging with machines, the singularity, and the meaning of winning in life, with a prediction of the singularity happening by 2038 and the goal of building a compressive model of the world.', 'duration': 531.673, 'highlights': ['The chapter delves into the concept of AI girlfriends and the potential for forming deep connections with AI, including the idea of VR brothels and falling in love with a machine. Discussion on the potential for AI girlfriends and forming deep emotional connections with AI, mentioning the possibility of VR brothels and falling in love with a machine.', "The speaker predicts the singularity to happen by 2038, based on the exponential growth of machine flops, and discusses the potential for merging with machines, envisioning a long-term monogamous relationship with a machine. Prediction of the singularity happening by 2038 based on the exponential growth of machine flops, and the speaker's vision of merging with machines in a long-term monogamous relationship.", 'The speaker describes the meaning of winning in life as building a compressive model of the world and maximizing the derivative of compression of the past, while acknowledging the uncertainty around the reward function and the ongoing process of discovering it. Discussion on the meaning of winning in life as building a compressive model of the world and maximizing the derivative of compression of the past, with acknowledgment of uncertainty around the reward function and the ongoing process of discovering it.']}], 'duration': 1306.425, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/iwcYp-XT7UI/pics/iwcYp-XT7UI5849835.jpg', 'highlights': ['The protagonist transitions to the role of president within Calm AI, resembling an unelected president of a small dictatorship country.', 'The conversation revolves around the future governance structure and business prospects of Calm AI, hinting at potential changes in leadership and strategic direction.', "The entrepreneur's company has a burn rate of 200K a month and a revenue of 100K a month, indicating a need to 4X the revenue.", 'The entrepreneur emphasizes the importance of honesty and transparency in selling products to consumers, expressing a dislike for overselling and hype.', 'The entrepreneur respects individuals with real skills and aims to be recognized among people with genuine expertise, mentioning specific names like Elon Musk, Kyle Vogt, and individuals at Waymo.', 'Nauto, a company the entrepreneur respects, sells driver monitoring cameras to fleet owners, reducing accidents by 18% and creating value in the market.', 'The trolley problem and ethical questions arise in AI decision-making, with a solution to detect ethical dilemmas at com.ai and disengage to leave decisions to humans.', 'Security concerns include adversarial examples affecting human behavior and GPS spoofing fooling Navigate on Autopilot, highlighting the importance of reliable local sensors in autonomous vehicles.', 'The speaker predicts the singularity to happen by 2038, based on the exponential growth of machine flops, and discusses the potential for merging with machines, envisioning a long-term monogamous relationship with a machine.', 'The speaker describes the meaning of winning in life as building a compressive model of the world and maximizing the derivative of compression of the past, while acknowledging the uncertainty around the reward function and the ongoing process of discovering it.']}], 'highlights': ['George Hotz is the founder of Kama AI, a machine learning based vehicle automation company.', 'He gained recognition for being the first person to carrier unlock an iPhone.', 'He has made significant contributions at the intersection of hardware and software.', 'He is known for being an outspoken personality in the field of AI and technology.', 'George Hotz unlocked the first iPhone in August 2007, marking his entry into hacking.', 'The concept of living in a simulation is discussed, highlighting the challenge of proving its existence due to a closed system design.', "Project Zero's 90-day deadlines for fixing vulnerabilities advanced the industry.", 'Contrast in coding styles: methodical vs. fast and chaotic approach.', 'Efficiency in coding approach: fast problem-solving with undocumented APIs.', 'Development of a 200-line distributed file system showcases an iterative and efficient approach.', 'Limitations of Python for large code bases due to lack of type checking and slower performance.', 'Advantages of Go over Python for large code bases, including strong typing and ecosystem.', 'Complexities and challenges of JavaScript, ES6, and TypeScript development explored.', 'Involvement in autonomous vehicle automation, including founding of Karma AI and work on a vision system for autopilot.', "Tesla's lead in data collection through real user experience puts them ahead in the race for level 5 autonomy, outpacing other companies like GM and Ford.", "The significance of driver monitoring to ensure safety and attention, with Tesla's focus on achieving 100% driver monitoring before releasing a consumer product for driving cars.", 'The need for driver monitoring systems in autonomous vehicles is emphasized, with a focus on the inefficiencies of current systems and the potential for improvement.', "The inefficiencies of Super Cruise's transition to engagement and disengagement are pointed out, indicating the need for smoother and more intuitive operation.", "The shortcomings of Tesla's autopilot experience, particularly the annoyance of the wheel engagement requirement and the potential for improvement through driver monitoring, are discussed.", 'The difficulty in achieving natural longitudinal driving and the need to solve the learning problem of lane changes', 'Waymo aims to redefine driving with a majority of self-driving cars, emphasizing the need for human adaptation.', 'Comma AI targets profitability by the next year.', 'OpenPilot now offers 100% driver monitoring, showcasing significant progress in the last two years.', "The project involves analyzing a data set to identify AEB triggers, surpassing Toyota's 10 million miles of real-world driving data for AEB triggers.", 'Waymo has a three to five-year lead on all the competitors in autonomous vehicle technology.', 'The protagonist transitions to the role of president within Calm AI, resembling an unelected president of a small dictatorship country.', "The entrepreneur's company has a burn rate of 200K a month and a revenue of 100K a month, indicating a need to 4X the revenue.", 'The entrepreneur emphasizes the importance of honesty and transparency in selling products to consumers, expressing a dislike for overselling and hype.', 'The speaker predicts the singularity to happen by 2038, based on the exponential growth of machine flops, and discusses the potential for merging with machines, envisioning a long-term monogamous relationship with a machine.']}