title
Michael I. Jordan: Machine Learning, Recommender Systems, and Future of AI | Lex Fridman Podcast #74

description
Michael I Jordan is a professor at Berkeley, and one of the most influential people in the history of machine learning, statistics, and artificial intelligence. He has been cited over 170,000 times and has mentored many of the world-class researchers defining the field of AI today, including Andrew Ng, Zoubin Ghahramani, Ben Taskar, and Yoshua Bengio. This episode is presented by Cash App. Download it & use code "LexPodcast": Cash App (App Store): https://apple.co/2sPrUHe Cash App (Google Play): https://bit.ly/2MlvP5w PODCAST INFO: Podcast website: https://lexfridman.com/podcast Apple Podcasts: https://apple.co/2lwqZIr Spotify: https://spoti.fi/2nEwCF8 RSS: https://lexfridman.com/feed/podcast/ Full episodes playlist: https://www.youtube.com/playlist?list=PLrAXtmErZgOdP_8GztsuKi9nrraNbKKp4 Clips playlist: https://www.youtube.com/playlist?list=PLrAXtmErZgOeciFP3CBCIEElOJeitOr41 EPISODE LINKS: (Blog post) Artificial Intelligence -- The Revolution Hasn’t Happened Yet: https://hdsr.mitpress.mit.edu/pub/wot7mkc1 OUTLINE: 0:00 - Introduction 3:02 - How far are we in development of AI? 8:25 - Neuralink and brain-computer interfaces 14:49 - The term "artificial intelligence" 19:00 - Does science progress by ideas or personalities? 19:55 - Disagreement with Yann LeCun 23:53 - Recommender systems and distributed decision-making at scale 43:34 - Facebook, privacy, and trust 1:01:11 - Are human beings fundamentally good? 1:02:32 - Can a human life and society be modeled as an optimization problem? 1:04:27 - Is the world deterministic? 1:04:59 - Role of optimization in multi-agent systems 1:09:52 - Optimization of neural networks 1:16:08 - Beautiful idea in optimization: Nesterov acceleration 1:19:02 - What is statistics? 1:29:21 - What is intelligence? 1:37:01 - Advice for students 1:39:57 - Which language is more beautiful: English or French? CONNECT: - Subscribe to this YouTube channel - Twitter: https://twitter.com/lexfridman - LinkedIn: https://www.linkedin.com/in/lexfridman - Facebook: https://www.facebook.com/LexFridmanPage - Instagram: https://www.instagram.com/lexfridman - Medium: https://medium.com/@lexfridman - Support on Patreon: https://www.patreon.com/lexfridman

detail
{'title': 'Michael I. Jordan: Machine Learning, Recommender Systems, and Future of AI | Lex Fridman Podcast #74', 'heatmap': [{'end': 764.272, 'start': 697.914, 'weight': 1}], 'summary': "Michael i. jordan's influential career with over 170,000 citations advocates for broadening the scope of ai; discusses evolution, challenges, and limitations of ai; explores potential impact of ai in music market and recommends shift to direct consumer-creator connection; addresses challenges of advertising model and algorithmic advertising, and emphasizes need for transparency; discusses technology's impact on human connection, privacy, and societal behavior; delves into limitations of optimization, distributed optimization in multi-agent systems, and deep learning optimization; examines challenges in defining human intelligence and its potential impact on future computer systems.", 'chapters': [{'end': 140.165, 'segs': [{'end': 35.968, 'src': 'embed', 'start': 0.309, 'weight': 0, 'content': [{'end': 3.11, 'text': 'The following is a conversation with Michael I.', 'start': 0.309, 'duration': 2.801}, {'end': 10.814, 'text': 'Jordan, a professor at Berkeley and one of the most influential people in the history of machine learning, statistics, and artificial intelligence.', 'start': 3.11, 'duration': 7.704}, {'end': 20.458, 'text': 'He has been cited over 170,000 times and he has mentored many of the world-class researchers defining the field of AI today,', 'start': 11.574, 'duration': 8.884}, {'end': 26.321, 'text': 'including Andrew Ng Zubin, Garamani Ben Taskar and Yoshua Bengio.', 'start': 20.458, 'duration': 5.863}, {'end': 35.968, 'text': 'All this, to me, is as impressive as the over 32,000 points in the six NBA championships of the Michael J.', 'start': 27.563, 'duration': 8.405}], 'summary': "Michael i. jordan has been cited over 170,000 times and mentored world-class researchers in ai, akin to michael jordan's 32,000 points in six nba championships.", 'duration': 35.659, 'max_score': 0.309, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/EYIKy_FM9x0/pics/EYIKy_FM9x0309.jpg'}], 'start': 0.309, 'title': "Michael i. jordan's influential career", 'summary': 'Explores the influential career of michael i. jordan, with over 170,000 citations, and his advocacy for broadening the scope of artificial intelligence, emphasizing its deeply human nature.', 'chapters': [{'end': 140.165, 'start': 0.309, 'title': 'Michael i. jordan: ai pioneer', 'summary': 'Explores the influential career of michael i. jordan, with over 170,000 citations, and his advocacy for broadening the scope of artificial intelligence, emphasizing its deeply human nature.', 'duration': 139.856, 'highlights': ['Michael I. Jordan has been cited over 170,000 times, mentoring many world-class researchers in the field of AI today, including Andrew Ng, Zubin Gharamani, Ben Taskar, and Yoshua Bengio.', 'Michael I. Jordan advocates for broadening the scope of artificial intelligence to understand and empower human beings at all levels of abstraction, viewing it as a deeply human endeavor.', "The host describes Michael I. Jordan as 'the Miles Davis of machine learning,' emphasizing his significant influence in the field."]}], 'duration': 139.856, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/EYIKy_FM9x0/pics/EYIKy_FM9x0309.jpg', 'highlights': ['Michael I. Jordan has been cited over 170,000 times, mentoring many world-class researchers in the field of AI today, including Andrew Ng, Zubin Gharamani, Ben Taskar, and Yoshua Bengio.', "The host describes Michael I. Jordan as 'the Miles Davis of machine learning,' emphasizing his significant influence in the field.", 'Michael I. Jordan advocates for broadening the scope of artificial intelligence to understand and empower human beings at all levels of abstraction, viewing it as a deeply human endeavor.']}, {'end': 1450.817, 'segs': [{'end': 356.118, 'src': 'embed', 'start': 323.237, 'weight': 0, 'content': [{'end': 324.938, 'text': 'So a whole field was developed called electrical engineering.', 'start': 323.237, 'duration': 1.701}, {'end': 334.184, 'text': "I think that's what's happening right now is that we have a proto-field, which is statistics more of the theoretical side,", 'start': 326.162, 'duration': 8.022}, {'end': 335.645, 'text': 'of the algorithmic side of computer science.', 'start': 334.184, 'duration': 1.461}, {'end': 337.505, 'text': 'That was enough to start to build things.', 'start': 336.165, 'duration': 1.34}, {'end': 343.467, 'text': 'But what things? Systems that bring value to human beings and use human data and mix in human decisions.', 'start': 337.905, 'duration': 5.562}, {'end': 346.508, 'text': 'The engineering side of that is all ad hoc.', 'start': 344.067, 'duration': 2.441}, {'end': 348.393, 'text': "That's what's emerging.", 'start': 347.372, 'duration': 1.021}, {'end': 351.295, 'text': "In fact, if you want to call machine learning a field, I think that's what it is.", 'start': 348.413, 'duration': 2.882}, {'end': 356.118, 'text': "That's a proto-form of engineering based on statistical and computational ideas of previous generations.", 'start': 351.355, 'duration': 4.763}], 'summary': 'Emerging proto-field merging statistics and computer science to build systems bringing value to human beings.', 'duration': 32.881, 'max_score': 323.237, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/EYIKy_FM9x0/pics/EYIKy_FM9x0323237.jpg'}, {'end': 421.155, 'src': 'embed', 'start': 397.775, 'weight': 1, 'content': [{'end': 404.421, 'text': "And a lot of people, say neuroscientists say we're not very far in understanding the human brain, but you're saying we're in the dark here.", 'start': 397.775, 'duration': 6.646}, {'end': 405.982, 'text': "Well, no, I'm not unique.", 'start': 404.681, 'duration': 1.301}, {'end': 411.987, 'text': "I don't even think in the clarity, but if you talk to real neuroscientists that really study real synapses or real neurons, they agree.", 'start': 406.282, 'duration': 5.705}, {'end': 413.028, 'text': 'They agree.', 'start': 412.668, 'duration': 0.36}, {'end': 417.552, 'text': "It's a hundreds-of-year task, and they're building it up slowly and surely.", 'start': 413.389, 'duration': 4.163}, {'end': 421.155, 'text': 'What the signal is there is not clear.', 'start': 418.553, 'duration': 2.602}], 'summary': 'Neuroscientists agree that understanding the human brain is a slow, centuries-long task.', 'duration': 23.38, 'max_score': 397.775, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/EYIKy_FM9x0/pics/EYIKy_FM9x0397775.jpg'}, {'end': 764.272, 'src': 'heatmap', 'start': 697.914, 'weight': 1, 'content': [{'end': 702.758, 'text': "that had real principles and you really know what you're doing and you have a little scientific understanding, maybe not even complete.", 'start': 697.914, 'duration': 4.844}, {'end': 707.969, 'text': 'It became more predictable and it really gave value to human life because it was understood.', 'start': 703.745, 'duration': 4.224}, {'end': 718.117, 'text': "We don't want to muddle too much these waters of what we're able to do versus what we really can do in a way that's going to impress the next.", 'start': 709.25, 'duration': 8.867}, {'end': 726.812, 'text': "I don't need to be wowed, but I think that someone comes along in 20 years, a younger person who's absorbed all the technology,", 'start': 718.137, 'duration': 8.675}, {'end': 729.414, 'text': 'and for them to be wowed, I think they have to be more deeply impressed.', 'start': 726.812, 'duration': 2.602}, {'end': 733.778, 'text': 'A young Kolmogorov would not be wowed by some of the stunts that you see right now coming from the big companies.', 'start': 729.454, 'duration': 4.324}, {'end': 734.439, 'text': 'The demos.', 'start': 733.818, 'duration': 0.621}, {'end': 739.502, 'text': 'but do you think the breakthroughs from Kolmogorov would be and give this question a chance,', 'start': 734.439, 'duration': 5.063}, {'end': 743.584, 'text': "do you think they'll be in the scientific fundamental principles arena?", 'start': 739.502, 'duration': 4.082}, {'end': 750.466, 'text': "Or do you think it's possible to have fundamental breakthroughs in engineering? Meaning, you know I would say,", 'start': 744.104, 'duration': 6.362}, {'end': 757.969, 'text': 'some of the things that Elon Musk is working with SpaceX and then others sort of trying to revolutionize the fundamentals of engineering,', 'start': 750.466, 'duration': 7.503}, {'end': 764.272, 'text': "of manufacturing, of, of saying here's a problem we know how to do a demo of and actually taking it to scale.", 'start': 757.969, 'duration': 6.303}], 'summary': 'Predictable technology can bring value to human life, but future breakthroughs may require deeper impact.', 'duration': 66.358, 'max_score': 697.914, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/EYIKy_FM9x0/pics/EYIKy_FM9x0697914.jpg'}, {'end': 1046.756, 'src': 'embed', 'start': 1018.177, 'weight': 2, 'content': [{'end': 1019.738, 'text': "We're trying to keep value at that scale.", 'start': 1018.177, 'duration': 1.561}, {'end': 1023.4, 'text': 'And aspects of this will look intelligent.', 'start': 1020.919, 'duration': 2.481}, {'end': 1026.521, 'text': 'Computers were so dumb before.', 'start': 1024.339, 'duration': 2.182}, {'end': 1028.08, 'text': 'They will seem more intelligent.', 'start': 1027.001, 'duration': 1.079}, {'end': 1029.781, 'text': 'We will use that buzzword of intelligence.', 'start': 1028.101, 'duration': 1.68}, {'end': 1031.123, 'text': 'So we can use it in that sense.', 'start': 1029.801, 'duration': 1.322}, {'end': 1037.185, 'text': 'So machine learning, you can scope it narrowly as just learning from data and pattern recognition.', 'start': 1032.262, 'duration': 4.923}, {'end': 1042.631, 'text': 'But whatever, when I talk about these topics, maybe data science is another word you could throw in the mix.', 'start': 1037.885, 'duration': 4.746}, {'end': 1046.756, 'text': 'It really is important that the decisions are as part of it.', 'start': 1043.652, 'duration': 3.104}], 'summary': 'Efforts to enhance computer intelligence at scale through machine learning and data science are crucial for making more intelligent decisions.', 'duration': 28.579, 'max_score': 1018.177, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/EYIKy_FM9x0/pics/EYIKy_FM9x01018177.jpg'}, {'end': 1414.848, 'src': 'embed', 'start': 1384.809, 'weight': 3, 'content': [{'end': 1385.249, 'text': "It's important.", 'start': 1384.809, 'duration': 0.44}, {'end': 1387.571, 'text': 'And the words of pattern recognition and prediction.', 'start': 1385.549, 'duration': 2.022}, {'end': 1391.053, 'text': 'I think the distinction there not to linger on words,', 'start': 1387.571, 'duration': 3.482}, {'end': 1400.359, 'text': 'but the distinction there is more a constraint sort of in the lab data set versus decision-making is talking about consequential decisions in the real world,', 'start': 1391.053, 'duration': 9.306}, {'end': 1402.661, 'text': 'under the messiness and the uncertainty of the real world.', 'start': 1400.359, 'duration': 2.302}, {'end': 1409.205, 'text': 'and just the whole of it, the whole mess of it that actually touches human beings and scale, like you said, market forces.', 'start': 1403.501, 'duration': 5.704}, {'end': 1409.825, 'text': "that's the.", 'start': 1409.205, 'duration': 0.62}, {'end': 1410.725, 'text': "that's the distinction.", 'start': 1409.825, 'duration': 0.9}, {'end': 1411.986, 'text': 'yeah, it helps add those.', 'start': 1410.725, 'duration': 1.261}, {'end': 1414.848, 'text': "that perspective, that broader perspective, you're right, i totally agree.", 'start': 1411.986, 'duration': 2.862}], 'summary': 'Distinguishing lab data from real-world decisions with human impact and market forces.', 'duration': 30.039, 'max_score': 1384.809, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/EYIKy_FM9x0/pics/EYIKy_FM9x01384809.jpg'}], 'start': 140.987, 'title': 'Evolution of ai, understanding human brain, and ai challenges', 'summary': 'Explores the evolution of ai and machine learning, parallels with engineering development, discusses challenges in understanding the human brain, limitations of current ai and machine learning technologies, and the importance of decision-making in ai.', 'chapters': [{'end': 356.118, 'start': 140.987, 'title': 'Evolution of ai and machine learning', 'summary': 'Explores the evolution of ai and machine learning, drawing parallels with the development of chemical and electrical engineering, and emphasizes the transition from theoretical and algorithmic aspects to a proto-field of engineering that brings value to human beings through the use of human data and decisions.', 'duration': 215.131, 'highlights': ['The development of AI and machine learning is compared to the evolution of chemical and electrical engineering, emphasizing the emergence of a proto-field that brings value to human beings through the use of human data and decisions.', 'Analogies are drawn between the historical development of chemical engineering from chemistry and electrical engineering from electromagnetism, and the transition from statistics and theoretical aspects to a proto-form of engineering based on computational ideas in machine learning.', 'The speaker discusses the emergence of a proto-field of engineering in machine learning, emphasizing the ad hoc nature of the engineering side that is emerging, and describes it as a combination of statistical and computational ideas from previous generations.']}, {'end': 578.119, 'start': 356.518, 'title': 'Understanding the human brain', 'summary': 'Discusses the challenges in understanding the human brain, emphasizing that it is a complex, centuries-long task, and cautioning against overestimating our current knowledge and technology capabilities.', 'duration': 221.601, 'highlights': ["The brain's computation methods are not fully understood, and real neuroscientists agree that it is a task spanning hundreds of years, with the complexity of synapses and dendritic trees presenting significant challenges.", 'The speaker warns against overestimating our current understanding of the brain and technology capabilities, particularly in the context of brain-computer interfaces and computational neuroscience.', "The chapter emphasizes the need for realistic expectations in neuroscience research, highlighting the potential for advancements in addressing brain diseases such as Alzheimer's and the current lack of understanding in the biochemistry of the brain."]}, {'end': 1138.816, 'start': 578.119, 'title': 'Challenges in ai and machine learning', 'summary': 'Discusses the limitations of current ai and machine learning technologies, emphasizing the lack of scientific understanding and the focus on practical applications over fundamental breakthroughs, highlighting the need for new terminology and a shift in the perception of ai.', 'duration': 560.697, 'highlights': ['The focus is on practical applications over fundamental breakthroughs, lacking scientific understanding. The speaker emphasizes the limitations of current AI and machine learning technologies, highlighting the lack of scientific understanding and the focus on practical applications over fundamental breakthroughs.', 'The need for new terminology and a shift in the perception of AI is emphasized. The chapter emphasizes the need for new terminology and a shift in the perception of AI, suggesting a focus on practical working systems at planetary scale rather than the goal of achieving intelligence.', 'Criticism of the current state of AI and machine learning, downplaying the potential for deep breakthroughs. The speaker criticizes the current state of AI and machine learning, downplaying the potential for deep breakthroughs and expressing skepticism towards the achievement of true intelligence in the near future.']}, {'end': 1450.817, 'start': 1139.336, 'title': 'Importance of decision-making in ai', 'summary': 'Discusses the importance of decision-making in ai, highlighting the need for a balanced focus on prediction and decision-making, and the current emphasis on prediction in the field of ai, with a shift towards decision-making in the industry.', 'duration': 311.481, 'highlights': ['The field of AI emphasizes prediction, but decision-making is where human lives are at stake, risks are taken, and the real-world consequences and uncertainties are considered, with about half of the people in industry working on decision-making. The importance of decision-making in AI is emphasized, as it involves consequential decisions in the real world, under the messiness and uncertainty of real-world scenarios, where human lives, risks, and market forces are at stake. About half of the people in industry are working on decision-making.', 'The imbalance in the current emphasis on prediction in AI, leading to the neglect of decision-making, which is equally important and integral to the real-world application of AI. There is an imbalance in the current emphasis on prediction in AI, neglecting the equally important aspect of decision-making, which is integral to the real-world application of AI.', 'The need for a balanced focus on both prediction and decision-making in AI, as prediction alone does not account for the complexities, uncertainties, and consequences of real-world scenarios. There is a need for a balanced focus on both prediction and decision-making in AI, as prediction alone does not account for the complexities, uncertainties, and consequences of real-world scenarios.']}], 'duration': 1309.83, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/EYIKy_FM9x0/pics/EYIKy_FM9x0140987.jpg', 'highlights': ['The development of AI and machine learning is compared to the evolution of chemical and electrical engineering, emphasizing the emergence of a proto-field that brings value to human beings through the use of human data and decisions.', "The brain's computation methods are not fully understood, and real neuroscientists agree that it is a task spanning hundreds of years, with the complexity of synapses and dendritic trees presenting significant challenges.", 'The focus is on practical applications over fundamental breakthroughs, lacking scientific understanding. The speaker emphasizes the limitations of current AI and machine learning technologies, highlighting the lack of scientific understanding and the focus on practical applications over fundamental breakthroughs.', 'The field of AI emphasizes prediction, but decision-making is where human lives are at stake, risks are taken, and the real-world consequences and uncertainties are considered, with about half of the people in industry working on decision-making.']}, {'end': 1886.677, 'segs': [{'end': 1642.892, 'src': 'embed', 'start': 1618.668, 'weight': 2, 'content': [{'end': 1625.212, 'text': "Anyway, a real market would be, if you're a creator of music, that you actually are somebody who's good enough that people want to listen to you.", 'start': 1618.668, 'duration': 6.544}, {'end': 1627.748, 'text': 'You should have the data available to you.', 'start': 1626.227, 'duration': 1.521}, {'end': 1630.988, 'text': 'There should be a dashboard showing a map of the United States.', 'start': 1627.768, 'duration': 3.22}, {'end': 1634.189, 'text': "So in last week, here's all the places your songs were listened to.", 'start': 1631.429, 'duration': 2.76}, {'end': 1642.892, 'text': "It should be transparent, vetable, so that if someone down in Providence sees that you're being listened to 10,000 times in Providence,", 'start': 1634.629, 'duration': 8.263}], 'summary': 'A real market for music creators should provide transparent and vetable data, such as a dashboard showing song listenership across the united states.', 'duration': 24.224, 'max_score': 1618.668, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/EYIKy_FM9x0/pics/EYIKy_FM9x01618668.jpg'}, {'end': 1686.747, 'src': 'embed', 'start': 1653.159, 'weight': 0, 'content': [{'end': 1656.362, 'text': 'You do that three times a year, you start to have a career.', 'start': 1653.159, 'duration': 3.203}, {'end': 1658.904, 'text': 'So in this sense, AI creates jobs.', 'start': 1657.043, 'duration': 1.861}, {'end': 1660.586, 'text': "It's not about taking away human jobs.", 'start': 1659.525, 'duration': 1.061}, {'end': 1662.568, 'text': "It's creating new jobs because it creates a new market.", 'start': 1660.646, 'duration': 1.922}, {'end': 1666.286, 'text': "Once you've created a market, you've now connected up producers and consumers.", 'start': 1663.402, 'duration': 2.884}, {'end': 1671.833, 'text': "You know, the person who's making the music can say to someone who comes to their shows a lot, hey, I'll play your daughter's wedding for $10,000.", 'start': 1666.506, 'duration': 5.327}, {'end': 1673.094, 'text': "You'll say $8,000.", 'start': 1671.834, 'duration': 1.26}, {'end': 1673.516, 'text': "They'll say $9,000.", 'start': 1673.095, 'duration': 0.421}, {'end': 1676.039, 'text': 'Then you, again, you can now get an income up to $100,000.', 'start': 1673.516, 'duration': 2.523}, {'end': 1677.401, 'text': "You're not going to be a millionaire.", 'start': 1676.039, 'duration': 1.362}, {'end': 1686.747, 'text': 'And now even think about really the value of music is in these personal connections,', 'start': 1681.844, 'duration': 4.903}], 'summary': 'Ai creates new jobs and markets, fostering personal connections and income opportunities.', 'duration': 33.588, 'max_score': 1653.159, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/EYIKy_FM9x0/pics/EYIKy_FM9x01653159.jpg'}, {'end': 1811.479, 'src': 'embed', 'start': 1773.266, 'weight': 4, 'content': [{'end': 1780.71, 'text': 'is to have at least one million creators make a comfortable living, putting on Spotify.', 'start': 1773.266, 'duration': 7.444}, {'end': 1781.39, 'text': 'So in..', 'start': 1781.09, 'duration': 0.3}, {'end': 1792.899, 'text': 'I think you articulate a really nice vision of the world and the digital and the cyberspace of markets.', 'start': 1785.332, 'duration': 7.567}, {'end': 1803.898, 'text': 'What do you think companies like Spotify or YouTube or Netflix can do to create such markets?', 'start': 1793.479, 'duration': 10.419}, {'end': 1805.338, 'text': 'Is it an AI problem??', 'start': 1804.098, 'duration': 1.24}, {'end': 1806.858, 'text': 'Is it an interface problem?', 'start': 1805.418, 'duration': 1.44}, {'end': 1811.479, 'text': 'so interface design? Is it some other kind of?', 'start': 1806.858, 'duration': 4.621}], 'summary': 'Goal: 1 million creators to make a comfortable living on spotify; discussing companies like spotify, youtube, and netflix creating digital markets through ai and interface design.', 'duration': 38.213, 'max_score': 1773.266, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/EYIKy_FM9x0/pics/EYIKy_FM9x01773266.jpg'}, {'end': 1855.621, 'src': 'embed', 'start': 1827.784, 'weight': 3, 'content': [{'end': 1830.464, 'text': "it's everything you said, but really I think missing that kind of culture.", 'start': 1827.784, 'duration': 2.68}, {'end': 1831.445, 'text': 'All right.', 'start': 1831.205, 'duration': 0.24}, {'end': 1834.266, 'text': "So it's literally that 16 year old who's, who's able to create the songs.", 'start': 1831.465, 'duration': 2.801}, {'end': 1836.807, 'text': "You don't create that as a Silicon Valley entity.", 'start': 1834.746, 'duration': 2.061}, {'end': 1838.407, 'text': "You don't hire them per se.", 'start': 1836.947, 'duration': 1.46}, {'end': 1843.549, 'text': 'Right You have to create an ecosystem in which they are wanted and that there belong.', 'start': 1839.047, 'duration': 4.502}, {'end': 1846.919, 'text': 'And so you have to have some cultural credibility to do things like this.', 'start': 1844.298, 'duration': 2.621}, {'end': 1850.96, 'text': 'Netflix, to their credit, wanted some of that sort of credibility.', 'start': 1847.739, 'duration': 3.221}, {'end': 1852.881, 'text': 'They created shows, content.', 'start': 1851, 'duration': 1.881}, {'end': 1853.861, 'text': 'They call it content.', 'start': 1853.021, 'duration': 0.84}, {'end': 1855.621, 'text': "It's such a terrible word, but it's culture.", 'start': 1853.881, 'duration': 1.74}], 'summary': "Creating a cultural ecosystem for the young generation, as seen in netflix's effort to build cultural credibility through shows and content.", 'duration': 27.837, 'max_score': 1827.784, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/EYIKy_FM9x0/pics/EYIKy_FM9x01827784.jpg'}], 'start': 1451.457, 'title': "Ai's impact on the music market", 'summary': 'Explores the potential of ai in creating a music market, connecting creators with consumers, potentially creating new job opportunities. it also discusses the potential for musicians to earn up to $100,000 through direct interactions with fans and the role of companies like spotify, youtube, and netflix in shaping digital markets.', 'chapters': [{'end': 1666.286, 'start': 1451.457, 'title': 'Ai creating music market', 'summary': 'Explores the idea of ai creating a music market, where ai systems can help connect music creators with consumers, potentially creating new job opportunities and addressing the issue of a missing market for independent musicians.', 'duration': 214.829, 'highlights': ['AI can create a market for music creators, connecting them with consumers and potentially creating new job opportunities. AI has the potential to create a music market that connects independent music creators with consumers, potentially providing new job opportunities and addressing the missing market for independent musicians.', 'The current music market lacks transparency and data availability for independent music creators. The current music market lacks transparency and data availability for independent music creators, hindering their ability to build a career and make a living from their music.', "AI's role in creating a transparent music market where creators can access data on their audience and potentially make a living from their music. AI can enable the creation of a transparent music market where creators have access to data on their audience, potentially allowing them to make a living from their music by performing in locations where their music is popular."]}, {'end': 1886.677, 'start': 1666.506, 'title': 'Creating personal connections through music', 'summary': 'Discusses the potential of creating personal connections in the music industry, enabling musicians to earn up to $100,000 through direct interactions with fans, and the importance of fostering a cultural ecosystem for young creators. it also explores the challenges and principles involved in creating new markets and the crucial role of companies like spotify, youtube, and netflix in shaping digital markets.', 'duration': 220.171, 'highlights': ['Musicians can earn up to $100,000 through direct interactions with fans, fostering personal connections and eliminating the need for advertising. The chapter emphasizes the potential for musicians to generate income by engaging directly with their fans, illustrating the possibility of earning up to $100,000 through personal connections and eliminating the need for traditional advertising.', 'The importance of creating a cultural ecosystem for young creators to thrive and be wanted in the industry, emphasizing cultural credibility and fostering a sense of belonging. The discussion highlights the significance of establishing a cultural ecosystem that welcomes and supports young creators, emphasizing the need for cultural credibility and a sense of belonging within the industry.', 'Challenges and principles involved in creating new markets, including the need to address unwanted aspects such as bad actors and misuse of data. The chapter delves into the challenges and principles associated with creating new markets, highlighting the need to address unwanted aspects such as bad actors and data misuse, emphasizing the importance of considering potential failures and value delivery.', 'The role of companies like Spotify, YouTube, and Netflix in shaping digital markets and the need to consider various factors including AI, interface design, and economics to create sustainable markets. The chapter explores the role of major companies such as Spotify, YouTube, and Netflix in shaping digital markets, emphasizing the need to address factors like AI, interface design, and economics to establish sustainable and inclusive markets for creators.']}], 'duration': 435.22, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/EYIKy_FM9x0/pics/EYIKy_FM9x01451457.jpg', 'highlights': ['AI can create a market for music creators, connecting them with consumers and potentially creating new job opportunities.', 'Musicians can earn up to $100,000 through direct interactions with fans, fostering personal connections and eliminating the need for advertising.', 'The current music market lacks transparency and data availability for independent music creators.', 'The importance of creating a cultural ecosystem for young creators to thrive and be wanted in the industry.', 'The role of companies like Spotify, YouTube, and Netflix in shaping digital markets and the need to consider various factors including AI, interface design, and economics to create sustainable markets.']}, {'end': 2263.887, 'segs': [{'end': 1948.885, 'src': 'embed', 'start': 1923.433, 'weight': 0, 'content': [{'end': 1928.497, 'text': "There's no magic in the algorithms, but a good recommender system is way better than a bad recommender system.", 'start': 1923.433, 'duration': 5.064}, {'end': 1932.56, 'text': 'And recommender systems is a billion dollar industry back even 10, 20 years ago.', 'start': 1929.017, 'duration': 3.543}, {'end': 1937.082, 'text': 'And it continues to be extremely important going forward.', 'start': 1935.101, 'duration': 1.981}, {'end': 1939.723, 'text': "What's your favorite recommender system, just so we can put something?", 'start': 1937.402, 'duration': 2.321}, {'end': 1942.723, 'text': 'Well, just historically, I was one of the, you know, when I first went to Amazon.', 'start': 1939.763, 'duration': 2.96}, {'end': 1946.985, 'text': "you know I first didn't like Amazon because they put the book people out of business or the library.", 'start': 1942.723, 'duration': 4.262}, {'end': 1948.885, 'text': 'you know the local booksellers went out of business.', 'start': 1946.985, 'duration': 1.9}], 'summary': 'Recommender systems are crucial and a billion-dollar industry, with amazon as a notable example.', 'duration': 25.452, 'max_score': 1923.433, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/EYIKy_FM9x0/pics/EYIKy_FM9x01923433.jpg'}, {'end': 2025.147, 'src': 'embed', 'start': 1996.64, 'weight': 4, 'content': [{'end': 2000.742, 'text': "It doesn't work so well for other kind of entities, but that's just the complexity of human life, like shirts.", 'start': 1996.64, 'duration': 4.102}, {'end': 2005.565, 'text': "I'm not going to get recommendations on shirts, but that's interesting.", 'start': 2000.902, 'duration': 4.663}, {'end': 2011.748, 'text': "If you try to recommend restaurants, it's hard.", 'start': 2006.425, 'duration': 5.323}, {'end': 2014.39, 'text': "It's hard to do it at scale.", 'start': 2012.088, 'duration': 2.302}, {'end': 2017.872, 'text': 'But a blend of recommendation systems with other..', 'start': 2015.23, 'duration': 2.642}, {'end': 2025.147, 'text': 'Uh, economic ideas, uh, matchings and so on is really, really still very open research wise.', 'start': 2019.46, 'duration': 5.687}], 'summary': 'Challenges in recommending for diverse entities; blending recommendation systems with economic concepts remains open research.', 'duration': 28.507, 'max_score': 1996.64, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/EYIKy_FM9x0/pics/EYIKy_FM9x01996640.jpg'}, {'end': 2073.884, 'src': 'embed', 'start': 2040.264, 'weight': 1, 'content': [{'end': 2048.71, 'text': "I think Facebook just launched Facebook News, so they're having recommend the kind of news that are most likely for you to be interesting.", 'start': 2040.264, 'duration': 8.446}, {'end': 2052.393, 'text': 'Do you think this is an AI solvable?', 'start': 2048.73, 'duration': 3.663}, {'end': 2058.077, 'text': "again, whatever term you want to use, do you think it's a solvable problem for machines or is it a deeply human problem that's unsolvable?", 'start': 2052.393, 'duration': 5.684}, {'end': 2060.166, 'text': "So I don't even think about it at that level.", 'start': 2058.704, 'duration': 1.462}, {'end': 2064.492, 'text': "I think that what's broken with some of these companies, it's all monetization by advertising.", 'start': 2060.306, 'duration': 4.186}, {'end': 2067.074, 'text': "They're not, at least Facebook.", 'start': 2065.313, 'duration': 1.761}, {'end': 2072.641, 'text': "I want to critique them, but they didn't really try to connect a producer and a consumer in an economic way, right?", 'start': 2067.074, 'duration': 5.567}, {'end': 2073.884, 'text': 'No one wants to pay for anything.', 'start': 2072.681, 'duration': 1.203}], 'summary': 'Facebook launched facebook news to recommend personalized news; the issue of monetization by advertising remains unsolved.', 'duration': 33.62, 'max_score': 2040.264, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/EYIKy_FM9x0/pics/EYIKy_FM9x02040264.jpg'}, {'end': 2162.414, 'src': 'embed', 'start': 2132.83, 'weight': 3, 'content': [{'end': 2134.611, 'text': 'is there an actual market between the producer and consumer?', 'start': 2132.83, 'duration': 1.781}, {'end': 2140.573, 'text': "They're the producers, the person who created that video clip, the person that made that website, the person who could make more such things,", 'start': 2135.031, 'duration': 5.542}, {'end': 2142.954, 'text': 'the person who could adjust it as a function of demand.', 'start': 2140.573, 'duration': 2.381}, {'end': 2147.241, 'text': "The person on the other side who's asking for different kinds of things, you know.", 'start': 2143.595, 'duration': 3.646}, {'end': 2148.643, 'text': 'so you see glimmers of that.', 'start': 2147.241, 'duration': 1.402}, {'end': 2153.331, 'text': "now there's influencers and there's kind of a little glimmering of a market, but it should have been done 20 years ago.", 'start': 2148.643, 'duration': 4.688}, {'end': 2154.333, 'text': 'It should have been thought about.', 'start': 2153.371, 'duration': 0.962}, {'end': 2157.158, 'text': 'It should have been created in parallel with the advertising ecosystem.', 'start': 2154.353, 'duration': 2.805}, {'end': 2162.414, 'text': "And then Facebook inherited that and I think they also didn't think very much about that.", 'start': 2158.292, 'duration': 4.122}], 'summary': 'There are glimmers of a market between producers and consumers, but it should have been established 20 years ago and integrated with the advertising ecosystem.', 'duration': 29.584, 'max_score': 2132.83, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/EYIKy_FM9x0/pics/EYIKy_FM9x02132830.jpg'}, {'end': 2256.443, 'src': 'embed', 'start': 2230.092, 'weight': 5, 'content': [{'end': 2233.793, 'text': "I'm not going to find, you know, I'll have lots of advertisers in my face.", 'start': 2230.092, 'duration': 3.701}, {'end': 2243.81, 'text': "Right. What I really want to do is broadcast to the world that I am going to Mumbai and have someone on the other side of a market look at me and there's a recommendation system there.", 'start': 2234.193, 'duration': 9.617}, {'end': 2245.912, 'text': "So they're not looking at all possible people coming to Mumbai.", 'start': 2243.83, 'duration': 2.082}, {'end': 2247.554, 'text': "They're looking at the people who are relevant to them.", 'start': 2245.932, 'duration': 1.622}, {'end': 2250.777, 'text': 'So someone my age group, someone who kind of knows me at some level.', 'start': 2247.614, 'duration': 3.163}, {'end': 2256.443, 'text': "I give up a little privacy by that, but I'm happy because what I'm going to get back is this person can make a little video for me.", 'start': 2251.958, 'duration': 4.485}], 'summary': 'A person seeks relevant recommendations for a mumbai trip via a privacy-compromising system, yielding personalized videos.', 'duration': 26.351, 'max_score': 2230.092, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/EYIKy_FM9x0/pics/EYIKy_FM9x02230092.jpg'}], 'start': 1886.777, 'title': 'Recommender systems and consumer connection', 'summary': 'Delves into the significance of recommender systems in linking consumers to creators, highlighting their billion-dollar industry status and impact on learning and economic ideas. it also addresses the broken monetization model of companies relying on advertising, emphasizing the need for a shift to a direct producer-consumer connection.', 'chapters': [{'end': 2060.166, 'start': 1886.777, 'title': 'Role of recommender systems in technology', 'summary': 'Discusses the importance of recommender systems in connecting consumers to creators, emphasizing their billion-dollar industry status and impact on learning and economic ideas.', 'duration': 173.389, 'highlights': ['Recommender systems are crucial in connecting consumers to creators and have a significant impact on learning and economic ideas, with the industry being valued at a billion dollars even 10 to 20 years ago.', "Good recommender systems are emphasized over bad ones, with a mention of Amazon's effective system that led to increased book sales and learning from peers indirectly.", 'The complexity of human life poses challenges for recommendation systems, particularly in suggesting entities like restaurants and shirts, with the need for open research in combining recommendation systems with economic ideas.', 'The discussion delves into the difficulty of recommendation systems in dealing with political content and news, posing the question of whether it is an AI-solvable problem or a deeply human, unsolvable issue.']}, {'end': 2263.887, 'start': 2060.306, 'title': 'Monetization and consumer connection', 'summary': 'Discusses the broken monetization model of companies relying on advertising, the missed opportunity to connect producers and consumers in an economic way, and the potential for a market between them, emphasizing the need for a shift from the advertising ecosystem to a direct producer-consumer connection.', 'duration': 203.581, 'highlights': ['The broken monetization model of companies relying on advertising and the missed opportunity to connect producers and consumers in an economic way. The chapter criticizes the reliance on advertising for monetization and highlights the missed opportunity to establish an economic connection between producers and consumers.', 'The potential for a market between producers and consumers and the need for a shift from the advertising ecosystem to a direct producer-consumer connection. The discussion emphasizes the potential for a market between producers and consumers and underscores the necessity to shift from the advertising ecosystem to a direct connection between producers and consumers for economic value liberation.', 'The example of a consumer seeking recommendations for activities in Mumbai and the idea of a personalized recommendation system based on relevant demographics. The example illustrates the concept of a consumer seeking recommendations in Mumbai and the potential for a personalized recommendation system based on relevant demographics, emphasizing the value of a direct connection between producers and consumers.']}], 'duration': 377.11, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/EYIKy_FM9x0/pics/EYIKy_FM9x01886777.jpg', 'highlights': ['Recommender systems are crucial in connecting consumers to creators and have a significant impact on learning and economic ideas, with the industry being valued at a billion dollars even 10 to 20 years ago.', 'The broken monetization model of companies relying on advertising and the missed opportunity to connect producers and consumers in an economic way.', "Good recommender systems are emphasized over bad ones, with a mention of Amazon's effective system that led to increased book sales and learning from peers indirectly.", 'The potential for a market between producers and consumers and the need for a shift from the advertising ecosystem to a direct producer-consumer connection.', 'The complexity of human life poses challenges for recommendation systems, particularly in suggesting entities like restaurants and shirts, with the need for open research in combining recommendation systems with economic ideas.', 'The example of a consumer seeking recommendations for activities in Mumbai and the idea of a personalized recommendation system based on relevant demographics.', 'The discussion delves into the difficulty of recommendation systems in dealing with political content and news, posing the question of whether it is an AI-solvable problem or a deeply human, unsolvable issue.', 'The chapter criticizes the reliance on advertising for monetization and highlights the missed opportunity to establish an economic connection between producers and consumers.']}, {'end': 2872.062, 'segs': [{'end': 2353.955, 'src': 'embed', 'start': 2329.808, 'weight': 0, 'content': [{'end': 2337.315, 'text': "But Silicon Valley has been dominated by the Google Facebook kind of mentality and the subscription and advertising and that's the core problem.", 'start': 2329.808, 'duration': 7.507}, {'end': 2345.532, 'text': "The fake news actually rides on top of that, because it means that you're monetizing with clip-through rate, and that is the core problem.", 'start': 2338.269, 'duration': 7.263}, {'end': 2346.452, 'text': "You've got to remove that.", 'start': 2345.552, 'duration': 0.9}, {'end': 2351.154, 'text': "So advertisement, if we're going to linger on that, I mean, that's an interesting thesis.", 'start': 2346.812, 'duration': 4.342}, {'end': 2353.955, 'text': "I don't know if everyone really deeply thinks about that.", 'start': 2351.234, 'duration': 2.721}], 'summary': 'Silicon valley dominated by google facebook mentality; core problem is monetizing through ads and subscriptions.', 'duration': 24.147, 'max_score': 2329.808, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/EYIKy_FM9x0/pics/EYIKy_FM9x02329808.jpg'}, {'end': 2563.263, 'src': 'embed', 'start': 2532.326, 'weight': 5, 'content': [{'end': 2536.61, 'text': "because it's just dominated by these tiny little things that add up to big money for the company.", 'start': 2532.326, 'duration': 4.284}, {'end': 2544.275, 'text': "So I think it will change because societies just don't stick with things that annoy a lot of people.", 'start': 2537.69, 'duration': 6.585}, {'end': 2547.618, 'text': 'And advertising currently annoys people more than it provides information.', 'start': 2544.696, 'duration': 2.922}, {'end': 2555.041, 'text': "And I think that Google probably is smart enough to figure out that this is a bad model, even though it's a huge amount of money,", 'start': 2548.419, 'duration': 6.622}, {'end': 2557.661, 'text': "and they'll have to figure out how to pull it away from it slowly.", 'start': 2555.041, 'duration': 2.62}, {'end': 2563.263, 'text': "And I'm sure the CEO there will figure it out, but they need to do it.", 'start': 2558.001, 'duration': 5.262}], 'summary': 'Societal change will push companies like google to reduce annoying advertising to retain users and revenue.', 'duration': 30.937, 'max_score': 2532.326, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/EYIKy_FM9x0/pics/EYIKy_FM9x02532326.jpg'}, {'end': 2588.028, 'src': 'embed', 'start': 2564.143, 'weight': 3, 'content': [{'end': 2571.584, 'text': 'So if you reduce advertising not to zero but you reduce it at the same time, you bring up producer-consumer actual real value being delivered.', 'start': 2564.143, 'duration': 7.441}, {'end': 2574.605, 'text': 'so real money is being paid and they take a 5% cut.', 'start': 2571.584, 'duration': 3.021}, {'end': 2580.006, 'text': 'that 5% could start to get big enough to cancel out the lost revenue from the poor kind of advertising.', 'start': 2574.605, 'duration': 5.401}, {'end': 2583.007, 'text': 'And I think that a good company will do that, will realize that.', 'start': 2580.026, 'duration': 2.981}, {'end': 2588.028, 'text': 'And Facebook, again, God bless them.', 'start': 2584.627, 'duration': 3.401}], 'summary': 'By reducing advertising and focusing on real value, a 5% cut could offset lost revenue, benefiting companies like facebook.', 'duration': 23.885, 'max_score': 2564.143, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/EYIKy_FM9x0/pics/EYIKy_FM9x02564143.jpg'}, {'end': 2842.994, 'src': 'embed', 'start': 2811.352, 'weight': 1, 'content': [{'end': 2812.493, 'text': "We don't want to do creepy things.", 'start': 2811.352, 'duration': 1.141}, {'end': 2818.457, 'text': "We really want people to trust us to actually only use information in ways that they really would approve of, that we don't decide.", 'start': 2813.113, 'duration': 5.344}, {'end': 2827.605, 'text': "And I'm just kind of adding that the health of a market is that when I connect to someone who producer-consumer,", 'start': 2820.258, 'duration': 7.347}, {'end': 2830.528, 'text': "it's not just a random producer-consumer, it's people who see each other.", 'start': 2827.605, 'duration': 2.923}, {'end': 2835.332, 'text': "They don't like each other, but they sense that if they transact, some happiness will go up on both sides.", 'start': 2830.548, 'duration': 4.784}, {'end': 2842.994, 'text': 'If a company helps me to do that in moments that I choose, of my choosing, then fine.', 'start': 2835.952, 'duration': 7.042}], 'summary': 'Ensuring trust by using information in approved ways for mutual happiness in transactions.', 'duration': 31.642, 'max_score': 2811.352, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/EYIKy_FM9x0/pics/EYIKy_FM9x02811352.jpg'}], 'start': 2263.947, 'title': 'Challenges of advertising model and algorithmic advertising', 'summary': 'Discusses challenges of the advertising model, emphasizing the need for a direct consumer-creator connection and addressing fake news. it also explores trust issues related to algorithmic advertising and the importance of transparency in company-customer relationships.', 'chapters': [{'end': 2658.397, 'start': 2263.947, 'title': 'Challenges of advertising model', 'summary': 'Discusses the challenges of the advertising model in online platforms, emphasizing the need to transition towards a direct connection between consumers and creators in order to provide real value and tackle the issue of fake news. it also highlights the impact of advertising dominance on market dynamics and the potential for platforms to shift away from the advertising model.', 'duration': 394.45, 'highlights': ['Online platforms need to transition towards a direct connection between consumers and creators to provide real value and combat fake news. The chapter emphasizes the importance of moving away from the advertising model to establish a direct connection between consumers and creators, in order to provide real value and address the issue of fake news.', 'Impact of advertising dominance on market dynamics and innovation levels. The dominance of advertising in online platforms is noted to have a negative impact on market dynamics and innovation levels, potentially hindering the emergence of new business models and reducing innovation.', 'Potential for platforms to shift away from the advertising model and prioritize real value. There is potential for online platforms to shift away from the advertising model and prioritize the delivery of real value by enabling a direct connection between consumers and creators, thereby reducing the reliance on advertising revenue.']}, {'end': 2872.062, 'start': 2658.397, 'title': 'Algorithmic advertising and trust', 'summary': 'Explores the trust issues related to algorithmic advertising, emphasizing the importance of transparency and trust in company-customer relationships and the impact on market health.', 'duration': 213.665, 'highlights': ['The importance of transparency and trust in company-customer relationships is emphasized, with the impact on market health being discussed.', 'The chapter delves into the trust issues related to algorithmic advertising, highlighting the role of caring and trusting in companies and their interactions with consumers.', 'The discussion touches on the need for companies to use consumer information in ways that are approved of, in contrast to exploiting it for undisclosed purposes.', 'The difference between moments when consumers want to buy something and moments when they simply want to browse is examined in relation to advertising strategies.']}], 'duration': 608.115, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/EYIKy_FM9x0/pics/EYIKy_FM9x02263947.jpg', 'highlights': ['Online platforms need to transition towards a direct connection between consumers and creators to provide real value and combat fake news.', 'The importance of transparency and trust in company-customer relationships is emphasized, with the impact on market health being discussed.', 'The chapter delves into the trust issues related to algorithmic advertising, highlighting the role of caring and trusting in companies and their interactions with consumers.', 'Potential for platforms to shift away from the advertising model and prioritize real value.', 'The discussion touches on the need for companies to use consumer information in ways that are approved of, in contrast to exploiting it for undisclosed purposes.', 'Impact of advertising dominance on market dynamics and innovation levels.']}, {'end': 3745.815, 'segs': [{'end': 2901.962, 'src': 'embed', 'start': 2872.502, 'weight': 0, 'content': [{'end': 2875.443, 'text': 'I want to walk around and see things and see people and be exposed to stuff.', 'start': 2872.502, 'duration': 2.941}, {'end': 2876.744, 'text': 'So I want control over that, though.', 'start': 2875.503, 'duration': 1.241}, {'end': 2879.125, 'text': "I don't want the company's algorithms to decide for me.", 'start': 2876.804, 'duration': 2.321}, {'end': 2880.931, 'text': "I think that's the thing.", 'start': 2880.231, 'duration': 0.7}, {'end': 2882.452, 'text': "It's a total loss of control.", 'start': 2881.252, 'duration': 1.2}, {'end': 2886.794, 'text': 'if Facebook thinks they should take the control from us of deciding when we want to have certain kinds of information.', 'start': 2882.452, 'duration': 4.342}, {'end': 2892.557, 'text': "when we don't what information that is, how much it relates to what they know about us that we didn't really want them to know about us.", 'start': 2886.794, 'duration': 5.763}, {'end': 2895.679, 'text': "I don't want them to be helping me in that way.", 'start': 2893.758, 'duration': 1.921}, {'end': 2897.96, 'text': "I don't want them to be helping them, but they decide.", 'start': 2895.979, 'duration': 1.981}, {'end': 2901.962, 'text': 'They have control over what I want and when.', 'start': 2898.3, 'duration': 3.662}], 'summary': "Desire for control over information exposure, opposed to company's algorithms' influence.", 'duration': 29.46, 'max_score': 2872.502, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/EYIKy_FM9x0/pics/EYIKy_FM9x02872502.jpg'}, {'end': 2995.451, 'src': 'embed', 'start': 2966.641, 'weight': 1, 'content': [{'end': 2970.265, 'text': "Right That's a company that had the right business model in mind from the get go.", 'start': 2966.641, 'duration': 3.624}, {'end': 2972.407, 'text': 'Right Executed on that.', 'start': 2970.285, 'duration': 2.122}, {'end': 2975.049, 'text': 'And from day one, there was value brought to.', 'start': 2972.687, 'duration': 2.362}, {'end': 2978.993, 'text': 'So here you have a kid who made some songs who suddenly their songs are on the NBA website.', 'start': 2975.37, 'duration': 3.623}, {'end': 2982.895, 'text': "Right That's real economic value to people.", 'start': 2979.814, 'duration': 3.081}, {'end': 2984.757, 'text': 'And so, you know.', 'start': 2983.275, 'duration': 1.482}, {'end': 2995.451, 'text': 'So you and I differ on the optimism of being able to sort of change the direction of the Titanic, right? Yeah.', 'start': 2985.438, 'duration': 10.013}], 'summary': 'Company with a successful business model brought real economic value, with songs featured on nba website.', 'duration': 28.81, 'max_score': 2966.641, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/EYIKy_FM9x0/pics/EYIKy_FM9x02966641.jpg'}, {'end': 3319.354, 'src': 'embed', 'start': 3287.668, 'weight': 2, 'content': [{'end': 3289.689, 'text': "They're not directly interacting with you.", 'start': 3287.668, 'duration': 2.021}, {'end': 3296.973, 'text': 'Sort of your close friends and family, the way they know you is by having conversation, by actually having interactions back and forth.', 'start': 3290.35, 'duration': 6.623}, {'end': 3304.357, 'text': "Do you think there's a place for recommender systems, sort of to step, because you just emphasized the value of human to human connection?", 'start': 3297.313, 'duration': 7.044}, {'end': 3307.318, 'text': "but let's give it a chance AI human connection.", 'start': 3304.357, 'duration': 2.961}, {'end': 3314.746, 'text': 'Is there a role for an AI system to have conversations with you in terms of to try to figure out what kind of music you like,', 'start': 3307.818, 'duration': 6.928}, {'end': 3319.354, 'text': 'not by just watching what you listen to, but actually having a conversation, natural language or otherwise?', 'start': 3314.746, 'duration': 4.608}], 'summary': 'Can ai have meaningful conversations to understand preferences like music?', 'duration': 31.686, 'max_score': 3287.668, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/EYIKy_FM9x0/pics/EYIKy_FM9x03287668.jpg'}, {'end': 3447.536, 'src': 'embed', 'start': 3420.462, 'weight': 3, 'content': [{'end': 3424.526, 'text': 'What does control over privacy look like? Do you think you should be able to just view all the data that.', 'start': 3420.462, 'duration': 4.064}, {'end': 3425.967, 'text': "No, it's much more than that.", 'start': 3425.046, 'duration': 0.921}, {'end': 3427.509, 'text': 'I mean, first of all, it should be an individual decision.', 'start': 3425.987, 'duration': 1.522}, {'end': 3429.19, 'text': "Some people don't want privacy.", 'start': 3427.829, 'duration': 1.361}, {'end': 3430.532, 'text': 'They want their whole life out there.', 'start': 3429.21, 'duration': 1.322}, {'end': 3431.433, 'text': "Other people's want it.", 'start': 3430.672, 'duration': 0.761}, {'end': 3435.416, 'text': 'Privacy is not a zero one.', 'start': 3433.334, 'duration': 2.082}, {'end': 3436.878, 'text': "It's not a legal thing.", 'start': 3435.977, 'duration': 0.901}, {'end': 3438.82, 'text': "It's not just about which data is available, which is not.", 'start': 3436.918, 'duration': 1.902}, {'end': 3445.936, 'text': 'I like to recall to people that, you know, a couple hundred years ago, everyone, there was not really big cities.', 'start': 3440.635, 'duration': 5.301}, {'end': 3447.536, 'text': 'Everyone lived in the countryside and villages.', 'start': 3445.996, 'duration': 1.54}], 'summary': "Privacy is not a binary concept; it's about individual choice and varies among people.", 'duration': 27.074, 'max_score': 3420.462, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/EYIKy_FM9x0/pics/EYIKy_FM9x03420462.jpg'}], 'start': 2872.502, 'title': "Personal information and technology's future", 'summary': "Discusses individuals' loss of control over personal information by companies like facebook and the potential positive impact through the right approach. it also explores the challenges and potential of technology in enhancing human connection, privacy, and societal behavior, emphasizing the limitations of recommender systems and the need for more transparent and controlled interactions.", 'chapters': [{'end': 3009.569, 'start': 2872.502, 'title': 'Control over personal information', 'summary': 'Discusses the loss of control over personal information by individuals in the hands of companies like facebook, emphasizing the lack of a value-driven business model and the potential for positive impact through the right approach.', 'duration': 137.067, 'highlights': ['The loss of control over personal information by individuals in the hands of companies like Facebook is emphasized, with a focus on the negative impact of algorithms dictating what information users receive and when.', 'The discussion touches on the potential for positive impact through the right approach, highlighting the value brought to individuals when companies have a value-driven business model from the beginning, as exemplified by a music market company with 100,000 artists signed on and their music featured on the NBA website.', 'The contrasting perspectives on the potential for change and the difficulty of addressing the issue are elaborated, with a reference to past experiences impacting the level of optimism.', 'The conversation delves into the need for a value-driven business model, emphasizing the economic value brought to individuals when companies execute on the right approach from the beginning, as demonstrated by the success of a company called United Masters.']}, {'end': 3745.815, 'start': 3009.569, 'title': 'Future of technology and human interaction', 'summary': 'Discusses the challenges and potential of technology in enhancing human connection, privacy, and societal behavior, emphasizing the limitations of recommender systems and the need for more transparent and controlled interactions.', 'duration': 736.246, 'highlights': ['The limitations of recommender systems and the need for human-to-human connections are emphasized, suggesting a more controlled and transparent approach to technology interactions. limitations of recommender systems, need for human-to-human connections, controlled and transparent technology interactions', 'The complexities of privacy and the need for individuals to have agency in managing their personal information are highlighted, stressing the importance of new structures and control over privacy. complexities of privacy, agency in managing personal information, importance of new structures and control over privacy', 'The potential for technology to open up perspectives and understanding, while acknowledging the fundamental limitations of human behavior and the impact of technology on societal interactions, is discussed. potential for technology in opening up perspectives, fundamental limitations of human behavior, impact of technology on societal interactions']}], 'duration': 873.313, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/EYIKy_FM9x0/pics/EYIKy_FM9x02872502.jpg', 'highlights': ['The loss of control over personal information by individuals in the hands of companies like Facebook is emphasized, with a focus on the negative impact of algorithms dictating what information users receive and when.', 'The need for a value-driven business model is emphasized, highlighting the economic value brought to individuals when companies execute on the right approach from the beginning.', 'The limitations of recommender systems and the need for human-to-human connections are emphasized, suggesting a more controlled and transparent approach to technology interactions.', 'The complexities of privacy and the need for individuals to have agency in managing their personal information are highlighted, stressing the importance of new structures and control over privacy.']}, {'end': 4134.504, 'segs': [{'end': 3799.54, 'src': 'embed', 'start': 3767.537, 'weight': 1, 'content': [{'end': 3770.358, 'text': 'Which, individual human life or society as a whole? Both.', 'start': 3767.537, 'duration': 2.821}, {'end': 3774.24, 'text': 'Both I mean, individual human life is amazingly complex.', 'start': 3770.898, 'duration': 3.342}, {'end': 3779.921, 'text': 'And so optimization is kind of just one branch of mathematics that talks about certain kinds of things,', 'start': 3774.32, 'duration': 5.601}, {'end': 3784.083, 'text': 'and it just feels way too limited for the complexity of such things.', 'start': 3779.921, 'duration': 4.162}, {'end': 3792.392, 'text': 'What properties of optimization problems do you think most interesting problems that could be solved through optimization?', 'start': 3784.403, 'duration': 7.989}, {'end': 3794.274, 'text': 'what kind of properties does that surface have?', 'start': 3792.392, 'duration': 1.882}, {'end': 3799.54, 'text': 'Non-convexity, convexity, linearity, all those kinds of things, saddle points.', 'start': 3794.374, 'duration': 5.166}], 'summary': 'Optimization is limited for the complexity of human life and society as a whole.', 'duration': 32.003, 'max_score': 3767.537, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/EYIKy_FM9x0/pics/EYIKy_FM9x03767537.jpg'}, {'end': 3827.098, 'src': 'embed', 'start': 3802.043, 'weight': 2, 'content': [{'end': 3810.808, 'text': "Even in our era, we're aware that, say, sampling It's coming up with examples of something, coming up with a distribution.", 'start': 3802.043, 'duration': 8.765}, {'end': 3811.589, 'text': "What's optimization??", 'start': 3810.828, 'duration': 0.761}, {'end': 3812.369, 'text': "What's sampling??", 'start': 3811.649, 'duration': 0.72}, {'end': 3818.553, 'text': "Well, if you're a certain kind of mathematician, you can try to blend them and make them seem to be sort of the same thing.", 'start': 3812.95, 'duration': 5.603}, {'end': 3827.098, 'text': 'But optimization is, roughly speaking, trying to find a point, a single point that is the optimum of a criterion function of some kind.', 'start': 3818.593, 'duration': 8.505}], 'summary': 'Optimization is finding the optimum point of a function.', 'duration': 25.055, 'max_score': 3802.043, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/EYIKy_FM9x0/pics/EYIKy_FM9x03802043.jpg'}, {'end': 3880.207, 'src': 'embed', 'start': 3855.486, 'weight': 0, 'content': [{'end': 3862.694, 'text': 'But in mathematics, if you go too high up that kind of abstraction hierarchy, you start to lose the ability to do the interesting theorems.', 'start': 3855.486, 'duration': 7.208}, {'end': 3865.637, 'text': "So you kind of don't try to overly over-abstract.", 'start': 3862.834, 'duration': 2.803}, {'end': 3876.845, 'text': "So as a small tangent, what kind of world do you find more appealing? One that is deterministic or stochastic? Well, that's easy.", 'start': 3866.898, 'duration': 9.947}, {'end': 3877.986, 'text': "I mean, I'm a statistician.", 'start': 3876.905, 'duration': 1.081}, {'end': 3880.207, 'text': 'The world is highly stochastic.', 'start': 3878.406, 'duration': 1.801}], 'summary': 'Mathematics requires balance between abstraction and practicality. the world is highly stochastic.', 'duration': 24.721, 'max_score': 3855.486, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/EYIKy_FM9x0/pics/EYIKy_FM9x03855486.jpg'}, {'end': 3926.336, 'src': 'embed', 'start': 3898.203, 'weight': 3, 'content': [{'end': 3906.966, 'text': 'So how does the distributed at scale when you have multi-agent systems look like?', 'start': 3898.203, 'duration': 8.763}, {'end': 3909.806, 'text': 'So optimization can optimize.', 'start': 3907.006, 'duration': 2.8}, {'end': 3916.208, 'text': 'sort of it makes a lot more sense, sort of, at least from a robotics perspective, for a single robot,', 'start': 3909.806, 'duration': 6.402}, {'end': 3919.049, 'text': 'for a single agent trying to optimize some objective function.', 'start': 3916.208, 'duration': 2.841}, {'end': 3926.336, 'text': 'When you start to enter the real world, this game theoretic concept starts popping up.', 'start': 3920.931, 'duration': 5.405}], 'summary': 'Distributed multi-agent systems optimize game theoretic concepts in the real world.', 'duration': 28.133, 'max_score': 3898.203, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/EYIKy_FM9x0/pics/EYIKy_FM9x03898203.jpg'}, {'end': 4128.428, 'src': 'embed', 'start': 4098.02, 'weight': 5, 'content': [{'end': 4100.902, 'text': 'So, you know, one example is a Stackelberg equilibrium.', 'start': 4098.02, 'duration': 2.882}, {'end': 4105.724, 'text': 'So, you know, Nash, you and I are both playing this game against each other or for each other.', 'start': 4101.122, 'duration': 4.602}, {'end': 4106.604, 'text': "Maybe it's cooperative.", 'start': 4105.764, 'duration': 0.84}, {'end': 4108.444, 'text': "And we're both going to think it through.", 'start': 4107.144, 'duration': 1.3}, {'end': 4111.645, 'text': "Then we're going to decide and we're going to do our thing simultaneously.", 'start': 4108.545, 'duration': 3.1}, {'end': 4114.612, 'text': "In a Stackelberg, no, I'm going to be the first mover.", 'start': 4112.669, 'duration': 1.943}, {'end': 4115.653, 'text': "I'm going to make a move.", 'start': 4114.631, 'duration': 1.022}, {'end': 4117.875, 'text': "You're going to look at my move, and then you're going to make yours.", 'start': 4115.792, 'duration': 2.083}, {'end': 4123.482, 'text': "Now, since I know you're going to look at my move, I anticipate what you're going to do, and so I don't do something stupid.", 'start': 4118.356, 'duration': 5.126}, {'end': 4128.428, 'text': "But then I know that you are also anticipating me, so we're kind of going back and forth in our mind.", 'start': 4124.323, 'duration': 4.105}], 'summary': 'Explaining the dynamics of stackelberg equilibrium and decision-making in a game scenario.', 'duration': 30.408, 'max_score': 4098.02, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/EYIKy_FM9x0/pics/EYIKy_FM9x04098020.jpg'}], 'start': 3745.815, 'title': 'Optimization and technology', 'summary': 'Delves into the limitations of optimization in modeling human life and society, the relationship between optimization and sampling, and the preference for a stochastic world due to uncertainty. it also explores distributed optimization in multi-agent systems, discussing game theory, market dynamics, algorithmic equilibria, and varied approaches in achieving optimization.', 'chapters': [{'end': 3896.303, 'start': 3745.815, 'title': 'Technology and optimization', 'summary': 'Discusses the limitations of optimization in modeling individual human life and society as a whole, the relationship between optimization and sampling, and the preference for a stochastic world due to massive uncertainty.', 'duration': 150.488, 'highlights': ['The world is highly stochastic, leading to massive uncertainty, and as a statistician, the preference is for a stochastic world. The world is highly stochastic, leading to massive uncertainty, and as a statistician, the preference is for a stochastic world.', 'Optimization is limited in modeling individual human life and society as a whole due to their complexity. Optimization is limited in modeling individual human life and society as a whole due to their complexity.', 'The relationship between optimization and sampling, and the attempt to blend them as similar concepts. The chapter discusses the relationship between optimization and sampling, and the attempt to blend them as similar concepts.']}, {'end': 4134.504, 'start': 3898.203, 'title': 'Distributed optimization in multi-agent systems', 'summary': 'Explores the concept of distributed optimization in multi-agent systems, discussing the interplay of game theory, market dynamics, and algorithmic equilibria, highlighting the challenges and varied approaches in achieving optimization.', 'duration': 236.301, 'highlights': ['The interplay of game theory, market dynamics, and algorithmic equilibria The chapter discusses the interplay of game theory, market dynamics, and algorithmic equilibria in the context of distributed optimization.', 'Challenges and varied approaches in achieving optimization It highlights the challenges and varied approaches involved in achieving optimization in multi-agent systems, encompassing game theory, market dynamics, and algorithmic equilibria.', 'Exploration of Nash equilibria and Stackelberg equilibrium The exploration of Nash equilibria and Stackelberg equilibrium provides insights into the different equilibria and decision-making dynamics in multi-agent systems, illustrating the complexity of achieving optimization.']}], 'duration': 388.689, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/EYIKy_FM9x0/pics/EYIKy_FM9x03745815.jpg', 'highlights': ['The world is highly stochastic, leading to massive uncertainty, and as a statistician, the preference is for a stochastic world.', 'Optimization is limited in modeling individual human life and society as a whole due to their complexity.', 'The relationship between optimization and sampling, and the attempt to blend them as similar concepts.', 'The interplay of game theory, market dynamics, and algorithmic equilibria in the context of distributed optimization.', 'Challenges and varied approaches involved in achieving optimization in multi-agent systems, encompassing game theory, market dynamics, and algorithmic equilibria.', 'Exploration of Nash equilibria and Stackelberg equilibrium provides insights into the different equilibria and decision-making dynamics in multi-agent systems, illustrating the complexity of achieving optimization.']}, {'end': 5331.475, 'segs': [{'end': 4165.948, 'src': 'embed', 'start': 4134.764, 'weight': 7, 'content': [{'end': 4139.948, 'text': 'And so just mathematically yeah, these things have certain topologies and certain shapes that are like salivates.', 'start': 4134.764, 'duration': 5.184}, {'end': 4142.749, 'text': 'And then algorithmically or dynamically, how do you move towards them??', 'start': 4139.968, 'duration': 2.781}, {'end': 4144.149, 'text': 'How do you move away from things??', 'start': 4142.81, 'duration': 1.339}, {'end': 4147.493, 'text': 'You know, so some of these questions have answers.', 'start': 4145.77, 'duration': 1.723}, {'end': 4148.292, 'text': "They've been studied.", 'start': 4147.513, 'duration': 0.779}, {'end': 4151.075, 'text': 'Others do not, and especially if it becomes stochastic.', 'start': 4148.533, 'duration': 2.542}, {'end': 4154.344, 'text': "especially if there's large numbers of decentralized things.", 'start': 4151.923, 'duration': 2.421}, {'end': 4159.145, 'text': "There's just, you know, young people getting in this field who kind of think it's all done because we have, you know, TensorFlow.", 'start': 4154.404, 'duration': 4.741}, {'end': 4163.508, 'text': "Well, no, these are all open problems and they're really important and interesting.", 'start': 4159.746, 'duration': 3.762}, {'end': 4165.048, 'text': "It's about strategic settings.", 'start': 4163.568, 'duration': 1.48}, {'end': 4165.948, 'text': 'How do I collect data?', 'start': 4165.087, 'duration': 0.861}], 'summary': 'Mathematical topologies and shapes inform movements, with open problems in stochastic and decentralized systems.', 'duration': 31.184, 'max_score': 4134.764, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/EYIKy_FM9x0/pics/EYIKy_FM9x04134764.jpg'}, {'end': 4255.605, 'src': 'embed', 'start': 4225.511, 'weight': 2, 'content': [{'end': 4230.415, 'text': "And secondly, if it's overparameterized, there's lots of paths down to reasonable optima.", 'start': 4225.511, 'duration': 4.904}, {'end': 4238.042, 'text': 'And so the getting downhill to an optimum is viewed as not as hard as you might have expected in high dimensions.', 'start': 4231.436, 'duration': 6.606}, {'end': 4248.059, 'text': 'fact that some optima tend to be really good ones and others not so good, and sometimes you find the good ones still needs explanation.', 'start': 4240.072, 'duration': 7.987}, {'end': 4255.605, 'text': 'But the particular surfaces coming from the particular generation of neural nets, I kind of suspect those will change.', 'start': 4249.92, 'duration': 5.685}], 'summary': 'High-dimensional overparameterized models have multiple paths to optima, some good, some not, with evolving neural net surfaces.', 'duration': 30.094, 'max_score': 4225.511, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/EYIKy_FM9x0/pics/EYIKy_FM9x04225511.jpg'}, {'end': 4349.083, 'src': 'embed', 'start': 4324.626, 'weight': 1, 'content': [{'end': 4336.69, 'text': "So if you just ask if we stay with the kind of architectures we have now not just neural nets but phase retrieval architectures or matrix completion architectures and so on I think we've kind of come to a place where,", 'start': 4324.626, 'duration': 12.064}, {'end': 4342.692, 'text': 'yeah, stochastic gradient algorithms are dominant and there are versions.', 'start': 4336.69, 'duration': 6.002}, {'end': 4349.083, 'text': "they're a little better than others, they have more guarantees, they're more robust, and so on.", 'start': 4344.5, 'duration': 4.583}], 'summary': 'Stochastic gradient algorithms dominate in various architectures, with some versions being more robust and having more guarantees.', 'duration': 24.457, 'max_score': 4324.626, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/EYIKy_FM9x0/pics/EYIKy_FM9x04324626.jpg'}, {'end': 4395.432, 'src': 'embed', 'start': 4365.272, 'weight': 3, 'content': [{'end': 4370.155, 'text': "So that I can't really anticipate that co-evolution process, but..", 'start': 4365.272, 'duration': 4.883}, {'end': 4373.597, 'text': 'You know, gradients are amazing mathematical objects.', 'start': 4370.755, 'duration': 2.842}, {'end': 4383.304, 'text': 'They have a lot of people who start to study them more deeply mathematically are kind of shocked about what they are and what they can do.', 'start': 4374.438, 'duration': 8.866}, {'end': 4386.186, 'text': 'I mean, think about it this way.', 'start': 4383.324, 'duration': 2.862}, {'end': 4395.432, 'text': 'Suppose that I tell you, if you move along the x-axis, you go uphill in some objective by three units, whereas if you move along the y-axis,', 'start': 4386.306, 'duration': 9.126}], 'summary': 'Gradients are powerful mathematical objects studied for their surprising capabilities and applications.', 'duration': 30.16, 'max_score': 4365.272, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/EYIKy_FM9x0/pics/EYIKy_FM9x04365272.jpg'}, {'end': 4696.481, 'src': 'embed', 'start': 4672.669, 'weight': 0, 'content': [{'end': 4679.93, 'text': "And so I think it's a surprise to most that Nesterov discovered a new algorithm that has got two pieces to it.", 'start': 4672.669, 'duration': 7.261}, {'end': 4680.871, 'text': 'It uses two gradients.', 'start': 4679.97, 'duration': 0.901}, {'end': 4686.113, 'text': 'and puts those together in a certain kind of obscure way.', 'start': 4683.05, 'duration': 3.063}, {'end': 4689.195, 'text': "And the thing doesn't even move downhill all the time.", 'start': 4686.453, 'duration': 2.742}, {'end': 4690.376, 'text': 'It sometimes goes back uphill.', 'start': 4689.235, 'duration': 1.141}, {'end': 4693.058, 'text': "And if you're a physicist, that kind of makes some sense.", 'start': 4691.457, 'duration': 1.601}, {'end': 4696.481, 'text': "You're building up some momentum and that is kind of the right intuition.", 'start': 4693.078, 'duration': 3.403}], 'summary': "Nesterov's new algorithm uses two gradients, occasionally moving uphill, aligning with physicist intuition.", 'duration': 23.812, 'max_score': 4672.669, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/EYIKy_FM9x0/pics/EYIKy_FM9x04672669.jpg'}, {'end': 4947.414, 'src': 'embed', 'start': 4920.541, 'weight': 4, 'content': [{'end': 4928.384, 'text': 'in most advanced statistical curricula you teach decision theory as the starting point and then it branches out into the two branches of Bayesian and frequentist.', 'start': 4920.541, 'duration': 7.843}, {'end': 4930.245, 'text': "but it's all about decisions.", 'start': 4928.384, 'duration': 1.861}, {'end': 4943.062, 'text': "In statistics, what is the most beautiful, mysterious, maybe surprising idea that you've come across? Yeah, good question.", 'start': 4931.781, 'duration': 11.281}, {'end': 4947.414, 'text': "I mean, there's a bunch of surprising ones.", 'start': 4945.353, 'duration': 2.061}], 'summary': 'Statistical curricula typically start with decision theory, branching into bayesian and frequentist approaches, all centered around decisions.', 'duration': 26.873, 'max_score': 4920.541, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/EYIKy_FM9x0/pics/EYIKy_FM9x04920541.jpg'}, {'end': 5205.376, 'src': 'embed', 'start': 5178.072, 'weight': 5, 'content': [{'end': 5180.813, 'text': "You're writing software, people are using it out there, you want to say something.", 'start': 5178.072, 'duration': 2.741}, {'end': 5183.895, 'text': 'So these two things have got to fight each other a little bit, but they have to blend.', 'start': 5180.833, 'duration': 3.062}, {'end': 5188.791, 'text': "So long story short, there's a set of ideas that are right in the middle that are called empirical Bayes.", 'start': 5184.81, 'duration': 3.981}, {'end': 5192.552, 'text': 'And empirical Bayes sort of starts with the Bayesian framework.', 'start': 5189.732, 'duration': 2.82}, {'end': 5199.575, 'text': "It's kind of arguably philosophically more, you know, reasonable and kosher.", 'start': 5193.593, 'duration': 5.982}, {'end': 5205.376, 'text': "Write down a bunch of the math that kind of flows from that and then realize there's a bunch of things you don't know,", 'start': 5199.595, 'duration': 5.781}], 'summary': 'Empirical bayes blends bayesian framework with math for software', 'duration': 27.304, 'max_score': 5178.072, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/EYIKy_FM9x0/pics/EYIKy_FM9x05178072.jpg'}, {'end': 5301.602, 'src': 'embed', 'start': 5273.229, 'weight': 6, 'content': [{'end': 5277.512, 'text': "you'd like the fraction of your false discoveries among your discoveries to be small.", 'start': 5273.229, 'duration': 4.283}, {'end': 5282.915, 'text': "that's a different criterion than accuracy or precision, or recall, or sensitivity and specificity.", 'start': 5277.512, 'duration': 5.403}, {'end': 5284.797, 'text': "it's a different quantity.", 'start': 5282.915, 'duration': 1.882}, {'end': 5289.859, 'text': 'those latter ones are almost all of them, um, have more of a frequentist flavor.', 'start': 5284.797, 'duration': 5.062}, {'end': 5295.161, 'text': "They say, given the truth is that the null hypothesis is true, here's what accuracy I would get.", 'start': 5289.919, 'duration': 5.242}, {'end': 5297.841, 'text': "Or given that the alternative is true, here's what I would get.", 'start': 5295.721, 'duration': 2.12}, {'end': 5301.602, 'text': "So it's kind of going forward from the state of nature to the data.", 'start': 5297.861, 'duration': 3.741}], 'summary': 'Minimize false discoveries among discoveries for better results.', 'duration': 28.373, 'max_score': 5273.229, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/EYIKy_FM9x0/pics/EYIKy_FM9x05273229.jpg'}], 'start': 4134.764, 'title': 'Understanding data analysis and deep learning optimization', 'summary': 'Discusses mathematical aspects of data analysis, emphasizing open problems and strategic settings, and delves into deep learning optimization, highlighting loss function surfaces, stochastic gradient algorithms, and faster convergence rates with nesterov acceleration.', 'chapters': [{'end': 4183.118, 'start': 4134.764, 'title': 'Understanding topologies and shapes in data analysis', 'summary': 'Discusses the mathematical aspects of data analysis, emphasizing the significance of open problems and strategic settings in collecting data, especially in stochastic and decentralized scenarios.', 'duration': 48.354, 'highlights': ['The significance of open problems and strategic settings in collecting data, especially in stochastic and decentralized scenarios, is emphasized, highlighting the importance of understanding topologies and shapes in data analysis.', 'The chapter discusses the mathematical aspects of data analysis, addressing questions related to moving towards or away from certain shapes and topologies, and the challenges posed by stochastic and decentralized scenarios.', 'The importance of collecting data strategically in scenarios where there is uncertainty about the behavior of individuals, emphasizing the need to understand the strategic aspects of data collection and analysis.']}, {'end': 4881.546, 'start': 4183.919, 'title': 'Optimizing deep learning: surfaces, algorithms, and stochasticity', 'summary': 'Delves into the optimization of deep learning, highlighting the smooth surfaces of loss functions, the potential for co-evolution of algorithms and architectures, the effectiveness of stochastic gradient algorithms, and the profound concept of nesterov acceleration in achieving faster convergence rates in optimization.', 'duration': 697.627, 'highlights': ["Nesterov acceleration offers faster convergence rates in optimization, achieving 1 over K squared compared to gradient descent's 1 over K, presenting a new algorithmic approach to utilizing gradients and exploring mathematical complexities. Nesterov acceleration achieves faster convergence rates of 1 over K squared, surpassing gradient descent's 1 over K; introduces a new algorithmic approach utilizing gradients and exploring mathematical complexities.", 'Stochastic gradient algorithms are dominant in the current deep learning architectures, with ongoing research to determine the best algorithm for specific situations, leading to potential co-evolution of algorithms and architectures. Stochastic gradient algorithms dominate current deep learning architectures; ongoing research aims to identify the best algorithm for specific situations, potentially leading to co-evolution of algorithms and architectures.', 'The smoothness of the surface and the overparameterization of deep learning optimization spaces contribute to the relative ease of finding reasonable optima in high dimensions, while stochasticity is shown to be effective in injecting randomness for better optimization. The smoothness of the surface and overparameterization ease finding reasonable optima in high dimensions; stochasticity effectively injects randomness for better optimization.', 'Gradients are emphasized as powerful mathematical objects with surprising properties, and the concept of stochasticity is explored for its ability to mitigate particular features of surfaces, especially in higher dimensions. Gradients are highlighted as powerful mathematical objects with surprising properties; stochasticity mitigates particular features of surfaces, especially in higher dimensions.']}, {'end': 5331.475, 'start': 4883.067, 'title': 'Evolution of decision theory and statistics', 'summary': 'Traces the development of decision theory and statistics, highlighting the fundamental concepts of bayesian and frequentist approaches, the paradoxes in statistics, and the emergence of empirical bayes and false discovery rate.', 'duration': 448.408, 'highlights': ['The chapter discusses the development of decision theory and statistics, emphasizing the fundamental concepts of Bayesian and frequentist approaches. Provides insights into the historical evolution of decision theory and statistics, focusing on the foundational concepts of Bayesian and frequentist approaches.', 'It explores the paradoxes in statistics and the challenges in reconciling the Bayesian and frequentist perspectives. Discusses the paradoxes in statistics, particularly the challenges in reconciling the Bayesian and frequentist perspectives, drawing parallels with the wave and particle duality in physics.', 'The emergence of empirical Bayes as a middle ground and its integration of human expertise and mathematical guidance. Explores the concept of empirical Bayes as a middle ground between Bayesian and frequentist approaches, emphasizing its integration of human expertise and mathematical guidance in decision-making.', 'An introduction to the false discovery rate, a criterion for evaluating the fraction of false discoveries among the announced discoveries. Introduces the false discovery rate as a criterion for evaluating the fraction of false discoveries among the announced discoveries, distinguishing it from traditional accuracy and precision measures.']}], 'duration': 1196.711, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/EYIKy_FM9x0/pics/EYIKy_FM9x04134764.jpg', 'highlights': ["Nesterov acceleration achieves faster convergence rates of 1 over K squared, surpassing gradient descent's 1 over K; introduces a new algorithmic approach utilizing gradients and exploring mathematical complexities.", 'Stochastic gradient algorithms dominate current deep learning architectures; ongoing research aims to identify the best algorithm for specific situations, potentially leading to co-evolution of algorithms and architectures.', 'The smoothness of the surface and overparameterization ease finding reasonable optima in high dimensions; stochasticity effectively injects randomness for better optimization.', 'Gradients are highlighted as powerful mathematical objects with surprising properties; stochasticity mitigates particular features of surfaces, especially in higher dimensions.', 'The chapter discusses the development of decision theory and statistics, emphasizing the fundamental concepts of Bayesian and frequentist approaches.', 'Explores the concept of empirical Bayes as a middle ground between Bayesian and frequentist approaches, emphasizing its integration of human expertise and mathematical guidance in decision-making.', 'Introduces the false discovery rate as a criterion for evaluating the fraction of false discoveries among the announced discoveries, distinguishing it from traditional accuracy and precision measures.', 'The significance of open problems and strategic settings in collecting data, especially in stochastic and decentralized scenarios, is emphasized, highlighting the importance of understanding topologies and shapes in data analysis.', 'The chapter discusses the mathematical aspects of data analysis, addressing questions related to moving towards or away from certain shapes and topologies, and the challenges posed by stochastic and decentralized scenarios.', 'The importance of collecting data strategically in scenarios where there is uncertainty about the behavior of individuals, emphasizing the need to understand the strategic aspects of data collection and analysis.']}, {'end': 6339.951, 'segs': [{'end': 5367.335, 'src': 'embed', 'start': 5332.543, 'weight': 4, 'content': [{'end': 5334.124, 'text': "And so it's a beautiful set of ideas.", 'start': 5332.543, 'duration': 1.581}, {'end': 5336.424, 'text': 'So this kind of line of argument has come out.', 'start': 5334.164, 'duration': 2.26}, {'end': 5342.266, 'text': "It's not certainly mine, but it sort of came out from Robbins around 1960.", 'start': 5336.724, 'duration': 5.542}, {'end': 5346.067, 'text': 'Brad Efron has written beautifully about this in various papers and books.', 'start': 5342.266, 'duration': 3.801}, {'end': 5354.67, 'text': 'And the FDRs, Ben Yamini in Israel, John Story did this Bayesian interpretation and so on.', 'start': 5346.327, 'duration': 8.343}, {'end': 5359.472, 'text': "So I've just absorbed these things over the years and find it a very healthy way to think about statistics.", 'start': 5354.71, 'duration': 4.762}, {'end': 5367.335, 'text': 'Let me ask you about intelligence to jump slightly back out into philosophy perhaps.', 'start': 5361.172, 'duration': 6.163}], 'summary': 'Absorbed and finds it healthy to think about statistics from ideas by efron and others.', 'duration': 34.792, 'max_score': 5332.543, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/EYIKy_FM9x0/pics/EYIKy_FM9x05332543.jpg'}, {'end': 5552.786, 'src': 'embed', 'start': 5524.777, 'weight': 1, 'content': [{'end': 5526.958, 'text': "it's been working for thousands of years.", 'start': 5524.777, 'duration': 2.181}, {'end': 5529.898, 'text': "uh, it works rain or shine, so it's adaptive.", 'start': 5526.958, 'duration': 2.94}, {'end': 5534.779, 'text': 'um so, all the kind of you know, those are adjectives one tends to apply to intelligent systems.', 'start': 5529.898, 'duration': 4.881}, {'end': 5537.7, 'text': "robust, adaptive, you know, you don't need to keep adjusting it.", 'start': 5534.779, 'duration': 2.921}, {'end': 5540.86, 'text': "it's self self healing, whatever plus, not perfect.", 'start': 5537.7, 'duration': 3.16}, {'end': 5543.521, 'text': 'you know, intelligences are never perfect and markets are not perfect.', 'start': 5540.86, 'duration': 2.661}, {'end': 5550.525, 'text': 'But I do not believe in this era that you can say well, our computers, our humans are smart, but no, markets are not.', 'start': 5544.581, 'duration': 5.944}, {'end': 5552.786, 'text': 'Well, markets are, so they are intelligent.', 'start': 5550.565, 'duration': 2.221}], 'summary': 'Intelligent systems are robust and adaptive, working for thousands of years rain or shine, self-healing and not perfect.', 'duration': 28.009, 'max_score': 5524.777, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/EYIKy_FM9x0/pics/EYIKy_FM9x05524777.jpg'}, {'end': 5700.911, 'src': 'embed', 'start': 5676.256, 'weight': 0, 'content': [{'end': 5686.358, 'text': 'But say as an axiom, one day we create a human level, superhuman level intelligent, not the scale of markets, but the scale of an individual.', 'start': 5676.256, 'duration': 10.102}, {'end': 5691.582, 'text': 'What do you think it would take to do that?', 'start': 5687.399, 'duration': 4.183}, {'end': 5700.911, 'text': 'Or maybe to ask another question is how would that system be different than the biological human beings that we see around us today?', 'start': 5691.722, 'duration': 9.189}], 'summary': 'Creating superhuman-level intelligence requires understanding its differences from biological humans.', 'duration': 24.655, 'max_score': 5676.256, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/EYIKy_FM9x0/pics/EYIKy_FM9x05676256.jpg'}, {'end': 5995.471, 'src': 'embed', 'start': 5966.298, 'weight': 3, 'content': [{'end': 5971.08, 'text': "So I'm always willing to reveal my ignorance to people around me so they can teach me things.", 'start': 5966.298, 'duration': 4.782}, {'end': 5974.174, 'text': 'And I think a lot of us feel that way about our field.', 'start': 5971.972, 'duration': 2.202}, {'end': 5975.055, 'text': "So it's very cooperative.", 'start': 5974.194, 'duration': 0.861}, {'end': 5978.918, 'text': "I might add, it's also very international because it's so cooperative.", 'start': 5976.015, 'duration': 2.903}, {'end': 5980.419, 'text': 'We see no barriers.', 'start': 5979.098, 'duration': 1.321}, {'end': 5984.162, 'text': 'And so the nationalism that you see, especially in the current era,', 'start': 5980.679, 'duration': 3.483}, {'end': 5988.005, 'text': "and everything is just at odds with the way that most of us think about what we're doing here,", 'start': 5984.162, 'duration': 3.843}, {'end': 5995.471, 'text': 'where this is a human endeavor and we cooperate and are very much trying to do it together for the benefit of everybody.', 'start': 5988.005, 'duration': 7.466}], 'summary': 'Cooperative and international environment, focused on human endeavor and benefit of everybody.', 'duration': 29.173, 'max_score': 5966.298, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/EYIKy_FM9x0/pics/EYIKy_FM9x05966298.jpg'}, {'end': 6118.611, 'src': 'embed', 'start': 6094.966, 'weight': 2, 'content': [{'end': 6101.77, 'text': "And the ability to also see myself in other people's shoes and have empathy and kind of work on that language is part of that.", 'start': 6094.966, 'duration': 6.804}, {'end': 6111.128, 'text': 'So after that kind of experience, and also embedding myself in French culture, which is quite amazing.', 'start': 6104.225, 'duration': 6.903}, {'end': 6115.91, 'text': "Languages are rich, not just because there is something inherently beautiful about it, but it's all the creativity that went into it.", 'start': 6111.408, 'duration': 4.502}, {'end': 6118.611, 'text': 'So I learned a lot of songs, read poems, read books.', 'start': 6115.93, 'duration': 2.681}], 'summary': 'Learning about empathy and language through cultural immersion.', 'duration': 23.645, 'max_score': 6094.966, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/EYIKy_FM9x0/pics/EYIKy_FM9x06094966.jpg'}], 'start': 5332.543, 'title': 'Human intelligence and ai', 'summary': 'Delves into the challenges of defining human intelligence, the intelligence of market systems, the value of philosophy and science fiction in ai, and the importance of learning languages. it also highlights the potential for increased intelligence in human civilization and the impact on future computer systems.', 'chapters': [{'end': 5484.912, 'start': 5332.543, 'title': 'Understanding human intelligence', 'summary': 'Discusses the challenges in defining human intelligence, the humble aspirations of psychologists to understand human intelligence, and the complexity of human reasoning, emphasizing the yet-to-occur ai revolution.', 'duration': 152.369, 'highlights': ['The complexity of human reasoning and the challenge in understanding human intelligence is emphasized, with psychologists having limited aspirations for the near future.', 'The discussion on understanding human intelligence involves experiments with babies and reasoning patterns, highlighting the limited understanding of how humans reason.', 'The absorption of ideas from various sources, including Robbins and Efron, is mentioned as a healthy way to think about statistics.']}, {'end': 5753.896, 'start': 5484.912, 'title': 'Intelligence in markets and economic systems', 'summary': 'Discusses the intelligence of market systems, comparing them to neurons in a brain, highlighting their adaptability, robustness, and relevance to future computer systems, while also exploring the potential for increased intelligence in human civilization. it also addresses the hypothetical creation of superhuman intelligence and its differences from biological human beings.', 'duration': 268.984, 'highlights': ['Market systems are compared to neurons in a brain, showcasing their decentralized decision-making and adaptability, working at various scales and being relevant to future computer systems. The comparison between market systems and neurons in a brain highlights their decentralized decision-making and adaptability, working at various scales, from small villages to big cities. It also emphasizes their relevance to future computer systems.', 'The intelligence of market systems is described as robust, adaptive, and self-healing, similar to intelligent systems, and it is noted that markets are not perfect, akin to human intelligence. The intelligence of market systems is described as robust, adaptive, and self-healing, exhibiting qualities similar to intelligent systems. It is also acknowledged that markets are not perfect, similar to human intelligence.', 'The discussion delves into the potential for increased intelligence in human civilization, posing questions about the linear or exponential increase in intelligence and the future of this intelligence. The chapter explores the potential for increased intelligence in human civilization, posing questions about the linear or exponential increase in intelligence and the future of this intelligence.', 'The hypothetical creation of superhuman intelligence is addressed, contemplating the differences between such a system and biological human beings, while acknowledging the speculative nature of the discussion. The chapter addresses the hypothetical creation of superhuman intelligence, contemplating the differences between such a system and biological human beings, while acknowledging the speculative nature of the discussion.']}, {'end': 5984.162, 'start': 5754.496, 'title': 'Philosophy, science fiction, and career advice', 'summary': 'Discusses the value of philosophy, science fiction, and career advice in the field of ai and machine learning, emphasizing the importance of apprenticeship, broad learning, and cooperation within the scientific community.', 'duration': 229.666, 'highlights': ['The importance of philosophy and thinking about societal challenges is discussed, emphasizing the need for a broader focus on present challenges and real-world problems over science fiction.', 'Advice for undergraduates interested in AI and machine learning is given, highlighting the significance of apprenticeship, hard work, and broad learning as essential steps in their journey.', 'The cooperative and international nature of the scientific community is emphasized, with a focus on mutual teaching and learning, revealing ignorance, and breaking down barriers.']}, {'end': 6339.951, 'start': 5984.162, 'title': 'Learning languages and human experience', 'summary': 'Discusses the importance of learning languages in understanding human experience, emphasizing the richness of diversity and the potential of natural language for ai, as advocated by michael i. jordan.', 'duration': 355.789, 'highlights': ["Michael I. Jordan emphasizes the importance of learning languages to understand and embed oneself in other people's experiences. Jordan believes in the significance of learning languages to immerse oneself in diverse human experiences, cultivating empathy and understanding.", 'Jordan highlights the potential of natural language understanding for AI, considering it a significant scientific challenge and the core of AI. Jordan advocates for deep understanding of natural language as a crucial aspect of AI, recognizing its potential to enhance human communication and provide insights into the human mind.', "Jordan advocates for broadening the scope of AI and conceiving a human-centric engineering discipline, recognizing the limitations of the term 'AI' and calling for a realistic approach to the challenges ahead. Jordan urges for a broader perspective on AI, suggesting the creation of a human-centric engineering discipline and acknowledging the limitations of the term 'AI,' emphasizing the need to address the serious challenges in the field."]}], 'duration': 1007.408, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/EYIKy_FM9x0/pics/EYIKy_FM9x05332543.jpg', 'highlights': ['The hypothetical creation of superhuman intelligence is addressed, contemplating the differences between such a system and biological human beings, while acknowledging the speculative nature of the discussion.', 'The intelligence of market systems is described as robust, adaptive, and self-healing, exhibiting qualities similar to intelligent systems. It is also acknowledged that markets are not perfect, similar to human intelligence.', "The importance of learning languages to understand and embed oneself in other people's experiences is emphasized by Michael I. Jordan.", 'The cooperative and international nature of the scientific community is emphasized, with a focus on mutual teaching and learning, revealing ignorance, and breaking down barriers.', 'The absorption of ideas from various sources, including Robbins and Efron, is mentioned as a healthy way to think about statistics.']}], 'highlights': ['Michael I. Jordan has been cited over 170,000 times, mentoring many world-class researchers in the field of AI today, including Andrew Ng, Zubin Gharamani, Ben Taskar, and Yoshua Bengio.', "The host describes Michael I. Jordan as 'the Miles Davis of machine learning,' emphasizing his significant influence in the field.", 'Michael I. Jordan advocates for broadening the scope of artificial intelligence to understand and empower human beings at all levels of abstraction, viewing it as a deeply human endeavor.', 'The development of AI and machine learning is compared to the evolution of chemical and electrical engineering, emphasizing the emergence of a proto-field that brings value to human beings through the use of human data and decisions.', "The brain's computation methods are not fully understood, and real neuroscientists agree that it is a task spanning hundreds of years, with the complexity of synapses and dendritic trees presenting significant challenges.", 'AI can create a market for music creators, connecting them with consumers and potentially creating new job opportunities.', 'The importance of creating a cultural ecosystem for young creators to thrive and be wanted in the industry.', 'Recommender systems are crucial in connecting consumers to creators and have a significant impact on learning and economic ideas, with the industry being valued at a billion dollars even 10 to 20 years ago.', 'The broken monetization model of companies relying on advertising and the missed opportunity to connect producers and consumers in an economic way.', 'Online platforms need to transition towards a direct connection between consumers and creators to provide real value and combat fake news.', 'The loss of control over personal information by individuals in the hands of companies like Facebook is emphasized, with a focus on the negative impact of algorithms dictating what information users receive and when.', 'The world is highly stochastic, leading to massive uncertainty, and as a statistician, the preference is for a stochastic world.', 'Optimization is limited in modeling individual human life and society as a whole due to their complexity.', "Nesterov acceleration achieves faster convergence rates of 1 over K squared, surpassing gradient descent's 1 over K; introduces a new algorithmic approach utilizing gradients and exploring mathematical complexities.", 'The hypothetical creation of superhuman intelligence is addressed, contemplating the differences between such a system and biological human beings, while acknowledging the speculative nature of the discussion.', 'The intelligence of market systems is described as robust, adaptive, and self-healing, exhibiting qualities similar to intelligent systems. It is also acknowledged that markets are not perfect, similar to human intelligence.']}