title
Douglas Lenat: Cyc and the Quest to Solve Common Sense Reasoning in AI | Lex Fridman Podcast #221

description
Douglas Lenat is the founder of Cyc, a 37 year project aiming to solve common-sense knowledge and reasoning in AI. Please support this podcast by checking out our sponsors: - Squarespace: https://lexfridman.com/squarespace and use code LEX to get 10% off - BiOptimizers: http://www.magbreakthrough.com/lex to get 10% off - Stamps.com: https://stamps.com and use code LEX to get free postage & scale - LMNT: https://drinkLMNT.com/lex to get free sample pack - ExpressVPN: https://expressvpn.com/lexpod and use code LexPod to get 3 months free EPISODE LINKS: Douglas's Twitter: https://twitter.com/cycorpai Cyc's Website: https://cyc.com PODCAST INFO: Podcast website: https://lexfridman.com/podcast Apple Podcasts: https://apple.co/2lwqZIr Spotify: https://spoti.fi/2nEwCF8 RSS: https://lexfridman.com/feed/podcast/ Full episodes playlist: https://www.youtube.com/playlist?list=PLrAXtmErZgOdP_8GztsuKi9nrraNbKKp4 Clips playlist: https://www.youtube.com/playlist?list=PLrAXtmErZgOeciFP3CBCIEElOJeitOr41 OUTLINE: 0:00 - Introduction 1:11 - What is Cyc? 9:17 - How to form a knowledge base of the universe 19:43 - How to train an AI knowledge base 24:04 - Global consistency versus local consistency 48:25 - Automated reasoning 54:05 - Direct uses of AI and machine learning 1:06:43 - The semantic web 1:17:16 - Tools to help Cyc interpret data 1:26:26 - The most beautiful idea about Cyc 1:32:25 - Love and consciousness in AI 1:39:24 - The greatness of Marvin Minsky 1:44:18 - Is Cyc just a beautiful dream? 1:49:03 - What is OpenCyc and how was it born? 1:54:53 - The open source community and OpenCyc 2:05:20 - The inference problem 2:07:03 - Cyc's programming language 2:14:37 - Ontological engineering 2:22:02 - Do machines think? 2:30:47 - Death and consciousness 2:40:48 - What would you say to AI? 2:45:24 - Advice to young people 2:47:20 - Mortality SOCIAL: - Twitter: https://twitter.com/lexfridman - LinkedIn: https://www.linkedin.com/in/lexfridman - Facebook: https://www.facebook.com/lexfridman - Instagram: https://www.instagram.com/lexfridman - Medium: https://medium.com/@lexfridman - Reddit: https://reddit.com/r/lexfridman - Support on Patreon: https://www.patreon.com/lexfridman

detail
{'title': 'Douglas Lenat: Cyc and the Quest to Solve Common Sense Reasoning in AI | Lex Fridman Podcast #221', 'heatmap': [], 'summary': "Discusses the psyche ai project aiming to acquire common sense knowledge for superintelligence, challenges in representing common sense in ais, ai's impact on human intelligence, abductive reasoning in machine learning, automating knowledge acquisition, language understanding limitations, semantic web, teaching ai, ai research challenges, open psych strategies, ai impact on reasoning and general intelligence tests, ethics, morality, societal resistance, agi's role in global problems, and paradigm shifts in science.", 'chapters': [{'end': 700.679, 'segs': [{'end': 24.412, 'src': 'embed', 'start': 0.149, 'weight': 0, 'content': [{'end': 9.219, 'text': 'The following is a conversation with Doug Lennett, creator of Psyche, a system that for close to 40 years and still today,', 'start': 0.149, 'duration': 9.07}, {'end': 18.169, 'text': 'has sought to solve the core problem of artificial intelligence the acquisition of common sense knowledge and the use of that knowledge to think,', 'start': 9.219, 'duration': 8.95}, {'end': 20.212, 'text': 'to reason and to understand the world.', 'start': 18.169, 'duration': 2.043}, {'end': 24.412, 'text': 'To support this podcast, please check out our sponsors in the description.', 'start': 21.271, 'duration': 3.141}], 'summary': 'Psyche, a system for 40 years, aims to solve the core problem of ai - acquiring common sense knowledge and using it to reason and understand the world.', 'duration': 24.263, 'max_score': 0.149, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/3wMKoSRbGVs/pics/3wMKoSRbGVs149.jpg'}, {'end': 113.633, 'src': 'embed', 'start': 78.153, 'weight': 1, 'content': [{'end': 83.378, 'text': 'whose goal is to assemble a knowledge base that spans the basic concepts and rules about how the world works.', 'start': 78.153, 'duration': 5.225}, {'end': 89.504, 'text': 'In other words, it hopes to capture common sense knowledge, which is a lot harder than it sounds.', 'start': 84.159, 'duration': 5.345}, {'end': 95.25, 'text': 'Can you elaborate on this mission and maybe perhaps speak to the various sub goals within this mission?', 'start': 90.245, 'duration': 5.005}, {'end': 102.024, 'text': 'When I was a faculty member in the computer science department at Stanford,', 'start': 96.26, 'duration': 5.764}, {'end': 113.633, 'text': 'my colleagues and I did research in all sorts of artificial intelligence programs, so natural language understanding programs, robots,', 'start': 102.024, 'duration': 11.609}], 'summary': 'Assembling a knowledge base for common sense knowledge in artificial intelligence research at stanford.', 'duration': 35.48, 'max_score': 78.153, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/3wMKoSRbGVs/pics/3wMKoSRbGVs78153.jpg'}, {'end': 164.634, 'src': 'embed', 'start': 136.442, 'weight': 2, 'content': [{'end': 141.866, 'text': "But if you're really trying to get AI, then you have to somehow get past the brick wall.", 'start': 136.442, 'duration': 5.424}, {'end': 147.129, 'text': "And the brick wall was the programs didn't have what we would call common sense.", 'start': 141.926, 'duration': 5.203}, {'end': 149.031, 'text': "They didn't have general world knowledge.", 'start': 147.189, 'duration': 1.842}, {'end': 154.552, 'text': "They didn't really understand what they were doing, what they were saying, what they were being asked.", 'start': 149.591, 'duration': 4.961}, {'end': 164.634, 'text': 'And so very much like a clever dog performing tricks, we could get them to do tricks, but they never really understood what they were doing.', 'start': 155.672, 'duration': 8.962}], 'summary': 'Ai programs lacked common sense and world knowledge, similar to clever dogs performing tricks.', 'duration': 28.192, 'max_score': 136.442, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/3wMKoSRbGVs/pics/3wMKoSRbGVs136442.jpg'}, {'end': 395.08, 'src': 'embed', 'start': 368.206, 'weight': 3, 'content': [{'end': 375.731, 'text': 'But see when you sit on that mountain or that ground that goes deep of knowledge.', 'start': 368.206, 'duration': 7.525}, {'end': 383.476, 'text': 'in order to make a split second decision about fast food trash or random trash from the back of a trash truck,', 'start': 375.731, 'duration': 7.745}, {'end': 389.398, 'text': 'You need to be able to leverage that ground you stand on in some way.', 'start': 386.077, 'duration': 3.321}, {'end': 395.08, 'text': "It's not merely, you know, it's not enough to just have a lot of ground to stand on.", 'start': 389.418, 'duration': 5.662}], 'summary': 'Leverage knowledge to make quick decisions about trash.', 'duration': 26.874, 'max_score': 368.206, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/3wMKoSRbGVs/pics/3wMKoSRbGVs368206.jpg'}, {'end': 594.9, 'src': 'embed', 'start': 564.648, 'weight': 4, 'content': [{'end': 565.928, 'text': 'So like what?', 'start': 564.648, 'duration': 1.28}, {'end': 566.649, 'text': 'in order?', 'start': 565.928, 'duration': 0.721}, {'end': 576.994, 'text': 'you mentioned philosophers in order to capture this world and represent it in a logical way and with a form of logic like how many statements are required?', 'start': 566.649, 'duration': 10.345}, {'end': 577.894, 'text': 'Is it five??', 'start': 577.294, 'duration': 0.6}, {'end': 578.735, 'text': 'Is it 10??', 'start': 578.034, 'duration': 0.701}, {'end': 580.095, 'text': 'Is it 10 trillion??', 'start': 578.735, 'duration': 1.36}, {'end': 581.135, 'text': 'Is it like that??', 'start': 580.535, 'duration': 0.6}, {'end': 585.277, 'text': "That's, as far as I understand, is probably still an open question.", 'start': 581.195, 'duration': 4.082}, {'end': 593.159, 'text': 'It may forever be an open question just to say like definitively about, to describe the universe perfectly.', 'start': 585.757, 'duration': 7.402}, {'end': 594.9, 'text': 'How many facts do you need?', 'start': 593.179, 'duration': 1.721}], 'summary': 'Philosophers aim to represent the world logically, posing the question of how many statements or facts are needed, with the answer remaining an open question.', 'duration': 30.252, 'max_score': 564.648, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/3wMKoSRbGVs/pics/3wMKoSRbGVs564648.jpg'}], 'start': 0.149, 'title': 'Psyche ai and knowledge for decisions', 'summary': 'Introduces psyche ai, a project aiming to solve the core problem of ai by acquiring common sense knowledge for general superintelligence. it also discusses the importance of leveraging knowledge for split-second decisions and the challenges of processing knowledge in logical form.', 'chapters': [{'end': 262.235, 'start': 0.149, 'title': 'Psyche ai: solving common sense knowledge', 'summary': 'Introduces doug lennett, creator of psyche, a long-standing project aiming to solve the core problem of artificial intelligence by acquiring common sense knowledge to achieve general superintelligence, which would also help humans understand their own minds and be more rational and kind to each other.', 'duration': 262.086, 'highlights': ['Doug Lennett created Psyche, a project that spans 40 years, aiming to solve the core problem of AI by acquiring common sense knowledge.', 'Acquiring common sense knowledge would contribute to achieving general superintelligence and help humans understand their own minds, the nature of truth, and be more rational and kind to each other.', 'The brick wall in AI was the lack of common sense knowledge, impeding systems from truly understanding what they were doing or being asked.', 'Understanding in AI is akin to a solid foundation that combines little things together through inference, similar to the way humans gain wisdom over time.', 'Computers need to be able to answer questions beyond providing the right answer, such as explaining why or considering potential errors, similar to human interaction.']}, {'end': 700.679, 'start': 262.255, 'title': 'Leveraging knowledge for split-second decisions', 'summary': 'Discusses the importance of leveraging a solid foundation of knowledge for making split-second decisions, as exemplified by a real-life scenario, and the challenges of representing and efficiently processing knowledge in logical form for practical use.', 'duration': 438.424, 'highlights': ['The importance of leveraging a solid foundation of knowledge for making split-second decisions is exemplified by a real-life scenario of a split-second decision while driving, emphasizing the need for common sense and quick decision-making in unforeseen situations. Real-life scenario of making a split-second decision while driving', 'Challenges of representing knowledge in logical form and efficiently processing it are discussed, along with the question of the quantity of statements required to capture and represent the world in a logical way. Discussion on the quantity of statements required to capture and represent the world in a logical way', 'The chapter also touches upon a meeting held in 1984 with prominent individuals to address the representation of information in logical form, showcasing the collaborative efforts to tackle complex knowledge-related questions. Meeting held in 1984 with prominent individuals to address the representation of information in logical form']}], 'duration': 700.53, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/3wMKoSRbGVs/pics/3wMKoSRbGVs149.jpg', 'highlights': ['Doug Lennett created Psyche, a 40-year project to acquire common sense knowledge for AI.', 'Acquiring common sense knowledge contributes to achieving general superintelligence and human understanding.', 'The lack of common sense knowledge is the brick wall in AI, impeding true understanding.', 'Leveraging knowledge for split-second decisions is crucial, exemplified by a real-life driving scenario.', 'Challenges of representing knowledge in logical form and the quantity of statements required are discussed.']}, {'end': 2051.012, 'segs': [{'end': 726.621, 'src': 'embed', 'start': 700.679, 'weight': 0, 'content': [{'end': 708.343, 'text': "which is if it's important to represent common sense, knowledge and world knowledge in order for AIs to not be brittle,", 'start': 700.679, 'duration': 7.664}, {'end': 712.385, 'text': 'in order for AIs not to just have the veneer of intelligence.', 'start': 708.343, 'duration': 4.042}, {'end': 718.607, 'text': 'well, how many pieces of common sense, how many if-then rules, For instance,', 'start': 712.385, 'duration': 6.222}, {'end': 726.621, 'text': 'would we have to actually write in order to essentially cover what people expect perfect strangers to already know about the world?', 'start': 718.607, 'duration': 8.014}], 'summary': 'Ai must incorporate common sense to avoid brittleness. requires vast knowledge base for human-like understanding.', 'duration': 25.942, 'max_score': 700.679, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/3wMKoSRbGVs/pics/3wMKoSRbGVs700679.jpg'}, {'end': 1049.452, 'src': 'embed', 'start': 1025.987, 'weight': 1, 'content': [{'end': 1038.45, 'text': 'But the good news was that, thanks to initially the fifth generation effort and then later US government agency funding and so on,', 'start': 1025.987, 'duration': 12.463}, {'end': 1046.691, 'text': 'we were able to get enough funding not for a couple person centuries of time, but for a couple person millennia of time,', 'start': 1038.45, 'duration': 8.241}, {'end': 1049.452, 'text': "which is what we've spent since 1984..", 'start': 1046.691, 'duration': 2.761}], 'summary': 'Secured funding for couple person millennia since 1984', 'duration': 23.465, 'max_score': 1025.987, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/3wMKoSRbGVs/pics/3wMKoSRbGVs1025987.jpg'}, {'end': 1689.105, 'src': 'embed', 'start': 1665.205, 'weight': 2, 'content': [{'end': 1672.753, 'text': "You want context to be first-class objects in your system's knowledge base, in particular in Psyche's knowledge base.", 'start': 1665.205, 'duration': 7.548}, {'end': 1688.545, 'text': 'And by first-class object I mean that we should be able to have Psyche think about and talk about and reason about one context or another context the same way it reasons about coffee cups and tables and people and fishing,', 'start': 1673.313, 'duration': 15.232}, {'end': 1689.105, 'text': 'and so on.', 'start': 1688.545, 'duration': 0.56}], 'summary': 'Make context a first-class object in the knowledge base, enabling psyche to reason about it like other objects.', 'duration': 23.9, 'max_score': 1665.205, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/3wMKoSRbGVs/pics/3wMKoSRbGVs1665205.jpg'}, {'end': 1865.299, 'src': 'embed', 'start': 1837.971, 'weight': 3, 'content': [{'end': 1845.013, 'text': 'And that was one of the things which they never were able to overcome, and I think that was one of the main reasons that that project failed.', 'start': 1837.971, 'duration': 7.042}, {'end': 1853.115, 'text': "So we'll return to some of the doors you've opened, but if we can go back to that room in 1984 around there with Marvin Minsky and Stanford.", 'start': 1845.433, 'duration': 7.682}, {'end': 1853.956, 'text': 'By the way,', 'start': 1853.575, 'duration': 0.381}, {'end': 1865.299, 'text': "I should mention that Marvin wouldn't do his estimate until someone brought him an envelope so that he could literally do a back of the envelope calculation to come up with his number.", 'start': 1853.956, 'duration': 11.343}], 'summary': "The project failed due to unresolved issues, including marvin minsky's estimate method.", 'duration': 27.328, 'max_score': 1837.971, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/3wMKoSRbGVs/pics/3wMKoSRbGVs1837971.jpg'}], 'start': 700.679, 'title': 'Representing common sense in ais', 'summary': 'Discusses the challenge of representing common sense in ais, exploring perspectives on the number of if-then rules required, with estimates ranging around a million based on factors such as human long-term memory, language fluency, and content of encyclopedic articles. it also covers the development of a common sense knowledge base involving tens of millions of rules, the complexities of truth in different contexts, and the importance of common sense knowledge representation.', 'chapters': [{'end': 828.148, 'start': 700.679, 'title': 'Ais and common sense representation', 'summary': 'Discusses the challenge of representing common sense in ais and explores different perspectives on the number of if-then rules required, with estimates ranging around a million based on factors such as human long-term memory, language fluency, and content of encyclopedic articles.', 'duration': 127.469, 'highlights': ['The challenge of representing common sense in AIs is explored, with estimates of around a million if-then rules required based on factors such as human long-term memory, language fluency, and content of encyclopedic articles.', 'Different perspectives on the number of if-then rules required are presented, including considerations of human long-term memory capacity, language fluency, and the content of encyclopedic articles.', 'Estimates for the number of if-then rules required for AIs to represent common sense range around a million, with factors such as human long-term memory, language fluency, and the content of encyclopedic articles influencing these estimations.']}, {'end': 1605.218, 'start': 828.849, 'title': 'Building common sense knowledge', 'summary': 'Discusses the development of a common sense knowledge base, spanning over decades and involving tens of millions of rules, aimed at encoding domain-specific applications and handling exceptions and inconsistencies.', 'duration': 776.369, 'highlights': ['The US government provided funding for a couple of person millennia of time, enabling the development of a common sense knowledge base since 1984. The US government funding and initial fifth generation effort allowed for the acquisition of funding for a couple of person millennia of time, which has been spent since 1984 in developing a common sense knowledge base.', 'Over 38 years, the project has been focused on containing tens of millions of rules, capturing common human knowledge, and specific domain applications. The project has been ongoing for 38 years, focused on containing tens of millions of rules to capture common human knowledge and specific domain applications, signifying substantial effort and dedication.', 'The knowledge base was designed with local consistency, akin to tectonic plates, to handle small inconsistencies between individual contexts. The knowledge base was developed with local consistency, similar to tectonic plates, to manage small inconsistencies between individual contexts, ensuring practical usability despite small inconsistencies.']}, {'end': 1837.751, 'start': 1605.218, 'title': 'Context and knowledge representation', 'summary': "Discusses the complexities of truth in different contexts and the importance of representing knowledge hierarchically through contexts, specifically emphasizing the need for context to be first-class objects in the system's knowledge base, and the drawbacks of representing knowledge as a tree rather than a graph.", 'duration': 232.533, 'highlights': ["The importance of context as first-class objects in the system's knowledge base Emphasizes the need for context to be first-class objects in the system's knowledge base, enabling reasoning and hierarchical arrangement of contexts.", 'Representation of knowledge as a graph rather than a tree Discusses the drawbacks of representing knowledge as a tree rather than a graph, highlighting the limitations it imposes on representing complex relationships and attributes of objects in the world.', 'Complexities of truth in different contexts Discusses the complexities of truth in different contexts, including variations based on time, region, belief system, and level of abstraction.']}, {'end': 2051.012, 'start': 1837.971, 'title': 'Common sense knowledge representation', 'summary': 'Explores the importance of common sense knowledge representation, discussing the failure of a project, the impact of the drake equation, and the distinction between facts and rules of thumb in knowledge representation.', 'duration': 213.041, 'highlights': ['The failure of the project was due to their inability to overcome certain challenges. The project failed due to the inability to overcome certain challenges.', 'The impact of the Drake Equation on motivating scientists and its potential value when aliens show up. The Drake Equation motivates scientists and will be valuable when aliens show up.', 'Discussion on representing rules of thumb and the distinction between facts and rules of thumb. The conversation involves representing rules of thumb and distinguishing them from facts.']}], 'duration': 1350.333, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/3wMKoSRbGVs/pics/3wMKoSRbGVs700679.jpg', 'highlights': ['Different perspectives on the number of if-then rules required are presented, including considerations of human long-term memory capacity, language fluency, and the content of encyclopedic articles.', 'The US government funding and initial fifth generation effort allowed for the acquisition of funding for a couple of person millennia of time, which has been spent since 1984 in developing a common sense knowledge base.', "Emphasizes the need for context to be first-class objects in the system's knowledge base, enabling reasoning and hierarchical arrangement of contexts.", 'The project failed due to the inability to overcome certain challenges.']}, {'end': 3533.211, 'segs': [{'end': 2135.844, 'src': 'embed', 'start': 2104.967, 'weight': 0, 'content': [{'end': 2106.707, 'text': "We're not very good at estimating that.", 'start': 2104.967, 'duration': 1.74}, {'end': 2112.189, 'text': 'And you are somebody who has really stayed with this question for decades.', 'start': 2107.448, 'duration': 4.741}, {'end': 2117.693, 'text': "What's your sense from 1984 to today??", 'start': 2114.391, 'duration': 3.302}, {'end': 2123.477, 'text': 'Have you gotten a stronger sense of just how much knowledge is required?', 'start': 2117.713, 'duration': 5.764}, {'end': 2129.861, 'text': "So you've kind of said, with some level of certainty, that it's still on the order of magnitude of tens of millions.", 'start': 2123.497, 'duration': 6.364}, {'end': 2135.844, 'text': 'Right, for the first several years, I would have said that it was on the order of one or two million.', 'start': 2130.241, 'duration': 5.603}], 'summary': 'It is estimated that tens of millions of knowledge is required since 1984.', 'duration': 30.877, 'max_score': 2104.967, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/3wMKoSRbGVs/pics/3wMKoSRbGVs2104967.jpg'}, {'end': 2323.663, 'src': 'embed', 'start': 2299.365, 'weight': 2, 'content': [{'end': 2307.127, 'text': 'And I think this is really an important point, which is if we have artificial intelligence, real general artificial intelligence,', 'start': 2299.365, 'duration': 7.762}, {'end': 2312.932, 'text': 'human-level artificial intelligence, then people will become smarter.', 'start': 2307.127, 'duration': 5.805}, {'end': 2317.016, 'text': "It's not so much that it'll be us versus the AIs.", 'start': 2314.013, 'duration': 3.003}, {'end': 2323.663, 'text': "It's more like us, and the AIs together will be able to do things that require more creativity.", 'start': 2317.076, 'duration': 6.587}], 'summary': 'Ai will make people smarter, enabling more creativity.', 'duration': 24.298, 'max_score': 2299.365, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/3wMKoSRbGVs/pics/3wMKoSRbGVs2299365.jpg'}, {'end': 2598.722, 'src': 'embed', 'start': 2569.617, 'weight': 1, 'content': [{'end': 2580.723, 'text': 'involving poverty and starvation, and global warming and overcrowding all the other problems that are besetting the planet.', 'start': 2569.617, 'duration': 11.106}, {'end': 2582.184, 'text': 'We really need to be smarter.', 'start': 2580.803, 'duration': 1.381}, {'end': 2584.747, 'text': 'And there are really only two routes to being smarter.', 'start': 2582.525, 'duration': 2.222}, {'end': 2589.831, 'text': 'One is through biochemistry and genetics.', 'start': 2584.927, 'duration': 4.904}, {'end': 2592.28, 'text': 'genetic engineering.', 'start': 2591.38, 'duration': 0.9}, {'end': 2598.722, 'text': 'The other route is through having general AIs that augment our intelligence.', 'start': 2592.6, 'duration': 6.122}], 'summary': 'Addressing global challenges through biochemistry, genetics, and ai for smarter solutions.', 'duration': 29.105, 'max_score': 2569.617, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/3wMKoSRbGVs/pics/3wMKoSRbGVs2569617.jpg'}, {'end': 2839.831, 'src': 'embed', 'start': 2810.423, 'weight': 4, 'content': [{'end': 2818.065, 'text': 'but educating in the sense of arming people with the ability to do good critical thinking is enormously powerful.', 'start': 2810.423, 'duration': 7.642}, {'end': 2824.907, 'text': "The education system that we have in the US and worldwide generally don't do a good job of that.", 'start': 2818.645, 'duration': 6.262}, {'end': 2832.269, 'text': "But I believe that the AI's will, the AI's can and will.", 'start': 2825.907, 'duration': 6.362}, {'end': 2839.831, 'text': 'In the same way that everyone can have their own Alexa or Siri or Google Assistant, or whatever.', 'start': 2832.329, 'duration': 7.502}], 'summary': 'Education should focus on empowering critical thinking, which current systems lack. ai can help achieve this.', 'duration': 29.408, 'max_score': 2810.423, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/3wMKoSRbGVs/pics/3wMKoSRbGVs2810423.jpg'}, {'end': 3169.464, 'src': 'embed', 'start': 3142.551, 'weight': 3, 'content': [{'end': 3148.396, 'text': 'And you say, oh, yeah, the answer to this question is this one, not the one you came up with.', 'start': 3142.551, 'duration': 5.845}, {'end': 3155.345, 'text': 'then what the system can do is it can use everything it already knows about common sense, general knowledge,', 'start': 3149.056, 'duration': 6.289}, {'end': 3164.698, 'text': "the domain you've already been telling it about and context like we talked about, and so on and say Well, here are seven alternatives,", 'start': 3155.345, 'duration': 9.353}, {'end': 3169.464, 'text': 'each of which I believe is plausible given everything I already know.', 'start': 3165.319, 'duration': 4.145}], 'summary': 'The system can provide seven plausible alternatives based on common sense and context.', 'duration': 26.913, 'max_score': 3142.551, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/3wMKoSRbGVs/pics/3wMKoSRbGVs3142551.jpg'}], 'start': 2051.212, 'title': 'Challenges in ai, ai impact on human intelligence, and abductive reasoning in machine learning', 'summary': 'Explores challenges in understanding the human mind for engineering intelligence, estimating the knowledge needed for super intelligent systems, discusses the future impact of ai on human intelligence, and explores the potential of ai to empower critical thinking. additionally, it delves into abductive reasoning in machine learning, which reduces the mental effort of human teachers in projects like genome-wide association studies.', 'chapters': [{'end': 2160.137, 'start': 2051.212, 'title': 'Challenges in understanding human mind and engineering intelligence', 'summary': 'Discusses the challenges in understanding the human mind and engineering intelligence, highlighting the difficulty in estimating knowledge required for creating super intelligent systems, with the speaker indicating the requirement of tens of millions of knowledge, showcasing a shift from an earlier estimate of one or two million.', 'duration': 108.925, 'highlights': ['The difficulty in estimating the knowledge required for creating super intelligent systems is highlighted, with the speaker indicating the requirement of tens of millions of knowledge, showcasing a shift from an earlier estimate of one or two million.', 'The speaker acknowledges the challenge of truly understanding ourselves when thinking about the human mind and engineering intelligence, emphasizing the difficulty in introspecting and solving intelligence.', 'The discussion revolves around the historical perspective of artificial intelligence, mentioning seminal figures like Marvin Minsky and Ed Feigenbaum, and the misconception in creating super intelligent systems.']}, {'end': 2542.705, 'start': 2160.137, 'title': 'Future of ai and human intelligence', 'summary': 'Discusses the potential impact of artificial intelligence on human intelligence, emphasizing the exponential learning potential and collaborative capabilities, while questioning the implications of widespread access to information and the evolution of human intelligence.', 'duration': 382.568, 'highlights': ['The development of artificial intelligence can lead to a collective increase in human intelligence, enabling parallel problem-solving and reduced misunderstandings. The collaboration between human intelligence and AI can result in enhanced creativity, parallel processing, and reduced misunderstandings, contributing to a collective increase in intelligence.', 'Access to extensive information through the internet and search engines may lead to a decline in independent knowledge and critical thinking skills. Widespread access to information through the internet and search engines may lead to a decline in independent knowledge, critical thinking skills, and the ability to estimate and understand numbers without external aid.', 'The chapter questions the potential negative impact of ubiquitous access to information, drawing parallels to the evolution of language and its criticism. The discussion draws parallels between the criticism of widespread access to information and the historical criticism of the evolution of language, highlighting the potential drawbacks of relying solely on external sources for knowledge and communication.']}, {'end': 3063.527, 'start': 2542.765, 'title': 'The impact of technology on society', 'summary': 'Discusses the impact of technology on society, emphasizing the need for smarter individuals and ai-augmented human species to overcome global problems, exploring the potential of ai to empower critical thinking and emotional growth, and addressing the challenges and opportunities associated with technology-driven knowledge acquisition.', 'duration': 520.762, 'highlights': ['The need for smarter individuals and AI-augmented human species to overcome global problems involving poverty, starvation, global warming, and overcrowding is emphasized, highlighting the potential of genetic engineering and general AIs to enhance intelligence.', 'The internet has enabled the creation of echo chambers, promoting the proliferation of conspiracy theories, fake news, and reinforcement of like-minded opinions, posing challenges to reasoned decision-making and critical thinking.', 'AI is expected to empower individuals by enhancing critical thinking skills, pointing out logical fallacies, and arming people with the ability to make reasoned decisions, potentially addressing the shortcomings of the current education system.', 'The potential of AI to act as a personal assistant, fostering emotional growth and happiness, in addition to intellectual development, is highlighted, indicating a broader impact of AI on human life.', 'The challenges and opportunities associated with technology-driven knowledge acquisition are discussed, including the need for automated reasoning and knowledge capture tools to expand and extend systems in areas where human input is limited.']}, {'end': 3533.211, 'start': 3063.527, 'title': 'Abductive reasoning in machine learning', 'summary': 'Discusses the concept of abductive reasoning in machine learning, which allows the system to make reasonable leaps and reduce the mental effort of human teachers, as demonstrated in a project involving genome-wide association studies and causal explanations for medical conditions.', 'duration': 469.684, 'highlights': ['The system can use everything it already knows and context to generate plausible alternatives, significantly reducing the mental effort of human teachers. The system can use common sense, general knowledge, and provided context to offer plausible alternatives, thus increasing the efficiency of the human teacher.', 'A project involving genome-wide association studies used abductive reasoning to provide human-interpretable causal explanations for medical conditions, leading to improved predictions and validation of causal chains. The project applied abductive reasoning to genome-wide association studies, offering causal explanations for medical conditions, which allowed for improved predictions and validation of causal chains.', "The concept of abductive reasoning in machine learning enables the system to think deeply and slowly, providing a balance to the quick thinking of machine learning techniques such as pattern recognition and statistical inference. Abductive reasoning provides a balance to the quick thinking of machine learning techniques, allowing the system to think deeply and slowly, similar to the left brain hemisphere's function in humans."]}], 'duration': 1481.999, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/3wMKoSRbGVs/pics/3wMKoSRbGVs2051212.jpg', 'highlights': ['The difficulty in estimating the knowledge required for creating super intelligent systems is highlighted, with the speaker indicating the requirement of tens of millions of knowledge, showcasing a shift from an earlier estimate of one or two million.', 'The need for smarter individuals and AI-augmented human species to overcome global problems involving poverty, starvation, global warming, and overcrowding is emphasized, highlighting the potential of genetic engineering and general AIs to enhance intelligence.', 'The development of artificial intelligence can lead to a collective increase in human intelligence, enabling parallel problem-solving and reduced misunderstandings. The collaboration between human intelligence and AI can result in enhanced creativity, parallel processing, and reduced misunderstandings, contributing to a collective increase in intelligence.', 'The system can use everything it already knows and context to generate plausible alternatives, significantly reducing the mental effort of human teachers. The system can use common sense, general knowledge, and provided context to offer plausible alternatives, thus increasing the efficiency of the human teacher.', 'AI is expected to empower individuals by enhancing critical thinking skills, pointing out logical fallacies, and arming people with the ability to make reasoned decisions, potentially addressing the shortcomings of the current education system.']}, {'end': 4167.971, 'segs': [{'end': 3581.441, 'src': 'embed', 'start': 3557.946, 'weight': 2, 'content': [{'end': 3565.371, 'text': 'So think of it like this kind of synergy where you want the right brain machine learning to quickly come up with possible answers.', 'start': 3557.946, 'duration': 7.425}, {'end': 3576.518, 'text': 'You want the left brain psych-like AI to think about that and think about why that might have been the case and what else would be the case if that were true,', 'start': 3565.711, 'duration': 10.807}, {'end': 3581.441, 'text': 'and so on, and then suggest things back to the right brain to quickly check out again.', 'start': 3576.518, 'duration': 4.923}], 'summary': 'Synergizing right brain machine learning with left brain ai to quickly generate and evaluate possible answers and suggestions.', 'duration': 23.495, 'max_score': 3557.946, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/3wMKoSRbGVs/pics/3wMKoSRbGVs3557946.jpg'}, {'end': 3630.653, 'src': 'embed', 'start': 3603.935, 'weight': 6, 'content': [{'end': 3610.159, 'text': 'This is very early days in the machine learning space of this, but self-supervised learning methods.', 'start': 3603.935, 'duration': 6.224}, {'end': 3620.927, 'text': 'You have these language models GPT-3 and so on that just read the internet, and they form representations that can then be mapped to something useful.', 'start': 3610.62, 'duration': 10.307}, {'end': 3622.848, 'text': 'The question is what is the useful thing??', 'start': 3621.027, 'duration': 1.821}, {'end': 3630.653, 'text': "Like, they're now playing with a pretty cool thing called OpenAI Codex, which is generating programs from documentation.", 'start': 3623.789, 'duration': 6.864}], 'summary': 'Early days in machine learning with self-supervised methods like gpt-3, openai codex generates programs from documentation.', 'duration': 26.718, 'max_score': 3603.935, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/3wMKoSRbGVs/pics/3wMKoSRbGVs3603935.jpg'}, {'end': 3681.648, 'src': 'embed', 'start': 3650.345, 'weight': 0, 'content': [{'end': 3653.627, 'text': 'one of our goals is, how can we harness machine learning,', 'start': 3650.345, 'duration': 3.282}, {'end': 3662.293, 'text': 'how can we harness natural language processing to increasingly automate the knowledge acquisition process, the growth of psych?', 'start': 3653.627, 'duration': 8.666}, {'end': 3671.478, 'text': "And that's what I meant by priming the pump that if you sort of learn things at the fringe of what you know already, you learn.", 'start': 3662.313, 'duration': 9.165}, {'end': 3677.022, 'text': 'this new thing is similar to what you know already, and here are the differences and the new things you had to learn about it, and so on.', 'start': 3671.478, 'duration': 5.544}, {'end': 3681.648, 'text': 'The more you know, the more and more easily you can learn new things.', 'start': 3677.682, 'duration': 3.966}], 'summary': 'Leverage machine learning and natural language processing to automate knowledge acquisition and facilitate easier learning.', 'duration': 31.303, 'max_score': 3650.345, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/3wMKoSRbGVs/pics/3wMKoSRbGVs3650345.jpg'}, {'end': 3829.517, 'src': 'embed', 'start': 3799.421, 'weight': 3, 'content': [{'end': 3806.127, 'text': 'Okay. However, if your goal is to save the life of your child who has some medical problem,', 'start': 3799.421, 'duration': 6.706}, {'end': 3814.851, 'text': 'or your goal is to be able to drive for the next 10,000 hours of driving without getting into a fatal accident and so on,', 'start': 3806.127, 'duration': 8.724}, {'end': 3821.373, 'text': 'then error rates down at the 10% level or even the 1% level are not really acceptable.', 'start': 3814.851, 'duration': 6.522}, {'end': 3829.517, 'text': 'I like the model of where that learning happens at the edge, and then you kind of think of knowledge as this sphere.', 'start': 3822.252, 'duration': 7.265}], 'summary': "High error rates are unacceptable for critical goals like saving a child's life or driving safely for 10,000 hours.", 'duration': 30.096, 'max_score': 3799.421, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/3wMKoSRbGVs/pics/3wMKoSRbGVs3799421.jpg'}, {'end': 3965.186, 'src': 'embed', 'start': 3933.501, 'weight': 4, 'content': [{'end': 3938.184, 'text': 'But the difference was because it was a knowledge-based system, a rule-based system.', 'start': 3933.501, 'duration': 4.683}, {'end': 3946.79, 'text': 'it was able to tell us, step by step by step, why this was the diagnosis and, step by step,', 'start': 3938.184, 'duration': 8.606}, {'end': 3953.875, 'text': 'why this was the best therapy and the best procedure to to do for her, and so on.', 'start': 3946.79, 'duration': 7.085}, {'end': 3957.899, 'text': 'And there was a real epiphany because that made all the difference in the world.', 'start': 3954.355, 'duration': 3.544}, {'end': 3965.186, 'text': 'Instead of blindly having to trust in authority, we were able to understand what was actually going on.', 'start': 3958.259, 'duration': 6.927}], 'summary': 'Knowledge-based system provides step-by-step diagnosis and therapy, leading to greater understanding and trust.', 'duration': 31.685, 'max_score': 3933.501, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/3wMKoSRbGVs/pics/3wMKoSRbGVs3933501.jpg'}, {'end': 4051.476, 'src': 'embed', 'start': 4028.12, 'weight': 5, 'content': [{'end': 4037.166, 'text': 'So think of the semantic web as a kind of knowledge graph, and Google already has something they call knowledge graph, for example,', 'start': 4028.12, 'duration': 9.046}, {'end': 4040.649, 'text': 'which is sort of like a node and link diagram.', 'start': 4037.166, 'duration': 3.483}, {'end': 4051.476, 'text': 'So you have these nodes that represent concepts or words or terms, and then there are some arcs that connect them that might be labeled.', 'start': 4041.089, 'duration': 10.387}], 'summary': "Semantic web is like a knowledge graph, similar to google's knowledge graph, with nodes representing concepts and terms connected by labeled arcs.", 'duration': 23.356, 'max_score': 4028.12, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/3wMKoSRbGVs/pics/3wMKoSRbGVs4028120.jpg'}], 'start': 3533.211, 'title': 'Automating knowledge acquisition with machine learning and challenges in language understanding', 'summary': 'Explores the synergy of machine learning and ai for automating knowledge acquisition, emphasizing the use of machine learning and nlu to expand a knowledge base for general ai. it also discusses the limitations of current language understanding models, stressing the importance of accurate understanding in critical scenarios such as medical diagnosis and driving safety.', 'chapters': [{'end': 3715.65, 'start': 3533.211, 'title': 'Automating knowledge acquisition with machine learning', 'summary': 'Discusses the synergy of machine learning and ai in automating knowledge acquisition, focusing on harnessing machine learning and nlu to expand a knowledge base for general ai.', 'duration': 182.439, 'highlights': ['The synergy between right brain machine learning and left brain psych-like AI is emphasized for general AI, with a focus on a small percentage of correlations with explanations and independent predictions.', 'The potential of self-supervised learning methods and language models like GPT-3 and OpenAI Codex in automating knowledge acquisition and growth of psych through harnessing machine learning and NLU.', 'The importance of priming the pump to expand the knowledge base, emphasizing the need for a large enough manually produced core to facilitate easier learning of new things.']}, {'end': 4167.971, 'start': 3715.65, 'title': 'Challenges in language understanding', 'summary': 'Discusses the limitations of current language understanding models, highlighting instances of incorrect responses and the importance of understanding and explaining the reasoning behind answers, citing examples of medical diagnosis and the need for a more comprehensive knowledge representation. it also emphasizes the need for accurate understanding in critical scenarios such as medical diagnosis and driving safety.', 'duration': 452.321, 'highlights': ['The chapter discusses the limitations of current language understanding models and the importance of understanding and explaining the reasoning behind answers. It highlights instances of incorrect responses from language models like GPT-3 and Siri, emphasizing the need for models to truly understand the context and reasoning behind their responses.', 'The importance of accurate understanding in critical scenarios such as medical diagnosis and driving safety is highlighted. The speaker emphasizes the need for higher accuracy in critical scenarios such as medical diagnosis and driving safety, where error rates at the 10% or 1% level are not acceptable.', 'The significance of knowledge-based systems in providing step-by-step explanations for diagnoses and therapy recommendations is discussed. The speaker recounts a personal experience where a knowledge-based medical diagnosis program provided step-by-step explanations for diagnoses and therapy recommendations, allowing for a better understanding and informed decision-making.', 'The concept of the semantic web and its limitations in representing complex relationships is explained. The limitations of representing complex relationships using the semantic web or similar graph structures are discussed, emphasizing the need for more comprehensive representations beyond simple graph structures.']}], 'duration': 634.76, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/3wMKoSRbGVs/pics/3wMKoSRbGVs3533211.jpg', 'highlights': ['The potential of self-supervised learning methods and language models like GPT-3 and OpenAI Codex in automating knowledge acquisition and growth of psych through harnessing machine learning and NLU.', 'The importance of priming the pump to expand the knowledge base, emphasizing the need for a large enough manually produced core to facilitate easier learning of new things.', 'The synergy between right brain machine learning and left brain psych-like AI is emphasized for general AI, with a focus on a small percentage of correlations with explanations and independent predictions.', 'The importance of accurate understanding in critical scenarios such as medical diagnosis and driving safety is highlighted. The speaker emphasizes the need for higher accuracy in critical scenarios such as medical diagnosis and driving safety, where error rates at the 10% or 1% level are not acceptable.', 'The significance of knowledge-based systems in providing step-by-step explanations for diagnoses and therapy recommendations is discussed. The speaker recounts a personal experience where a knowledge-based medical diagnosis program provided step-by-step explanations for diagnoses and therapy recommendations, allowing for a better understanding and informed decision-making.', 'The concept of the semantic web and its limitations in representing complex relationships is explained. The limitations of representing complex relationships using the semantic web or similar graph structures are discussed, emphasizing the need for more comprehensive representations beyond simple graph structures.', 'The chapter discusses the limitations of current language understanding models and the importance of understanding and explaining the reasoning behind answers. It highlights instances of incorrect responses from language models like GPT-3 and Siri, emphasizing the need for models to truly understand the context and reasoning behind their responses.']}, {'end': 6133.606, 'segs': [{'end': 4188.983, 'src': 'embed', 'start': 4168.752, 'weight': 0, 'content': [{'end': 4181.801, 'text': "So you're able to represent and reason with these much, much, much more complicated expressions that go way way beyond what, simple as it were,", 'start': 4168.752, 'duration': 13.049}, {'end': 4188.983, 'text': 'three-word or four-word English sentences are, which is really what the Semantic Web can represent and really what knowledge graphs can represent.', 'start': 4181.801, 'duration': 7.182}], 'summary': 'Semantic web and knowledge graphs can represent complex expressions beyond simple english sentences.', 'duration': 20.231, 'max_score': 4168.752, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/3wMKoSRbGVs/pics/3wMKoSRbGVs4168752.jpg'}, {'end': 4597.496, 'src': 'embed', 'start': 4563.543, 'weight': 2, 'content': [{'end': 4567.564, 'text': "It's important to be able to represent that.", 'start': 4563.543, 'duration': 4.021}, {'end': 4573.847, 'text': 'Exactly You need to be able to represent all of these things because, in fact, people do represent them.', 'start': 4567.664, 'duration': 6.183}, {'end': 4574.907, 'text': 'They do talk about them.', 'start': 4573.867, 'duration': 1.04}, {'end': 4576.908, 'text': 'They do try and teach them to other people.', 'start': 4575.207, 'duration': 1.701}, {'end': 4580.409, 'text': 'You do have rules of thumb that key off of them and so on.', 'start': 4576.948, 'duration': 3.461}, {'end': 4586.572, 'text': "If you can't represent it, then it's sort of like someone with a limited vocabulary who can't understand as easily.", 'start': 4580.469, 'duration': 6.103}, {'end': 4588.913, 'text': "what you're trying to tell them.", 'start': 4587.292, 'duration': 1.621}, {'end': 4597.496, 'text': "And so that's really why I think that the general dream, the original dream of systematic web is exactly right on.", 'start': 4588.973, 'duration': 8.523}], 'summary': 'Representation of concepts is crucial for effective communication and understanding in the systematic web.', 'duration': 33.953, 'max_score': 4563.543, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/3wMKoSRbGVs/pics/3wMKoSRbGVs4563543.jpg'}, {'end': 4803.945, 'src': 'embed', 'start': 4773.191, 'weight': 3, 'content': [{'end': 4779.655, 'text': 'not just be able to read it but actually be able to have systems that reason with it, then yes,', 'start': 4773.191, 'duration': 6.464}, {'end': 4789.581, 'text': 'it will be worth your small amount of time to go back and make sure that the AI trying to understand it really did correctly understand it.', 'start': 4779.655, 'duration': 9.926}, {'end': 4796.81, 'text': "And let's say you run a a travel website or something like that.", 'start': 4790.621, 'duration': 6.189}, {'end': 4803.945, 'text': 'And people are going to be coming to it because of searches they did looking for,', 'start': 4797.311, 'duration': 6.634}], 'summary': 'Developing ai systems that can reason with text is worth the time investment for businesses, such as travel websites, to ensure accurate understanding.', 'duration': 30.754, 'max_score': 4773.191, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/3wMKoSRbGVs/pics/3wMKoSRbGVs4773191.jpg'}, {'end': 4931.141, 'src': 'embed', 'start': 4905.321, 'weight': 5, 'content': [{'end': 4915.446, 'text': "One of the reasons I really enjoy how Tesla have implemented their autopilot system is there's a sense that you're helping this machine learn.", 'start': 4905.321, 'duration': 10.125}, {'end': 4920.69, 'text': 'And I think humans, I mean, having children, Pets.', 'start': 4915.466, 'duration': 5.224}, {'end': 4922.853, 'text': 'People love doing that.', 'start': 4921.051, 'duration': 1.802}, {'end': 4931.141, 'text': "There's joy to teaching for some people, but, I think, for a lot of people, and that if you create the interface where it feels like you're teaching,", 'start': 4922.893, 'duration': 8.248}], 'summary': "Tesla's autopilot system creates a sense of teaching and learning, similar to helping children or pets.", 'duration': 25.82, 'max_score': 4905.321, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/3wMKoSRbGVs/pics/3wMKoSRbGVs4905321.jpg'}, {'end': 5012.154, 'src': 'embed', 'start': 4985.422, 'weight': 6, 'content': [{'end': 4992.967, 'text': 'almost all of it has the computer playing the role of the teacher, and the student plays the role of the student.', 'start': 4985.422, 'duration': 7.545}, {'end': 5005.289, 'text': 'But, as I just mentioned, you can get a lot of learning to happen better and, as you said, more enjoyably if you are the mentor or the teacher,', 'start': 4994.059, 'duration': 11.23}, {'end': 5005.729, 'text': 'and so on.', 'start': 5005.289, 'duration': 0.44}, {'end': 5012.154, 'text': 'So we developed a program called Mathcraft to help sixth graders better understand math.', 'start': 5006.089, 'duration': 6.065}], 'summary': 'Mathcraft program aids sixth graders in learning math more effectively.', 'duration': 26.732, 'max_score': 4985.422, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/3wMKoSRbGVs/pics/3wMKoSRbGVs4985422.jpg'}, {'end': 5108.892, 'src': 'embed', 'start': 5082.202, 'weight': 7, 'content': [{'end': 5088.745, 'text': "So from your point of view as the player, you feel like you've taught it something because it used to make this mistake and now it doesn't.", 'start': 5082.202, 'duration': 6.543}, {'end': 5089.645, 'text': 'so on.', 'start': 5089.365, 'duration': 0.28}, {'end': 5095.267, 'text': 'So this tremendous reinforcement and engagement because of that and so on.', 'start': 5089.665, 'duration': 5.602}, {'end': 5108.892, 'text': 'So having a system that plays the role of a student and having the player play the role of the mentor is an enormously powerful type of metaphor,', 'start': 5095.367, 'duration': 13.525}], 'summary': "Player feels they've taught the system, leading to tremendous reinforcement and engagement.", 'duration': 26.69, 'max_score': 5082.202, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/3wMKoSRbGVs/pics/3wMKoSRbGVs5082202.jpg'}, {'end': 5259.834, 'src': 'embed', 'start': 5226.216, 'weight': 8, 'content': [{'end': 5242.761, 'text': "but I think it's the feeling that This will be the only time in history that anyone ever has to teach a computer this particular thing that we're now teaching it.", 'start': 5226.216, 'duration': 16.545}, {'end': 5249.284, 'text': "It's like painting Starry Night.", 'start': 5245.28, 'duration': 4.004}, {'end': 5252.727, 'text': 'You only have to do that once, or creating the Pietà.', 'start': 5250.345, 'duration': 2.382}, {'end': 5253.948, 'text': 'You only have to do that once.', 'start': 5252.767, 'duration': 1.181}, {'end': 5259.834, 'text': "It's not like a singer who has to keep..", 'start': 5255.009, 'duration': 4.825}], 'summary': 'Teaching a computer unique skills, akin to creating iconic art, only needs to be done once.', 'duration': 33.618, 'max_score': 5226.216, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/3wMKoSRbGVs/pics/3wMKoSRbGVs5226216.jpg'}, {'end': 5372.279, 'src': 'embed', 'start': 5343.705, 'weight': 9, 'content': [{'end': 5349.772, 'text': "But then there's also, you know, It's a project that continues forever in some sense, just like Wikipedia.", 'start': 5343.705, 'duration': 6.067}, {'end': 5355.793, 'text': 'Yes, you get to a stable base of knowledge, but knowledge grows, knowledge evolves.', 'start': 5350.272, 'duration': 5.521}, {'end': 5368.557, 'text': 'We learn as a human species, as science, as an organism, constantly grows and evolves and changes,', 'start': 5356.734, 'duration': 11.823}, {'end': 5372.279, 'text': 'and then empowered that with the tools of artificial intelligence.', 'start': 5368.557, 'duration': 3.722}], 'summary': 'Knowledge grows and evolves like wikipedia with the aid of artificial intelligence.', 'duration': 28.574, 'max_score': 5343.705, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/3wMKoSRbGVs/pics/3wMKoSRbGVs5343705.jpg'}, {'end': 5501.387, 'src': 'embed', 'start': 5462.321, 'weight': 10, 'content': [{'end': 5463.762, 'text': "Well, that's another issue,", 'start': 5462.321, 'duration': 1.441}, {'end': 5479.891, 'text': "which is there'll be a thin envelope of time between when we have general AIs and when everyone realizes that they should have basic human rights and freedoms,", 'start': 5463.762, 'duration': 16.129}, {'end': 5480.431, 'text': 'and so on.', 'start': 5479.891, 'duration': 0.54}, {'end': 5491.179, 'text': "Right now, we don't think twice about effectively enslaving our email systems and our series and our Alexas and so on.", 'start': 5481.452, 'duration': 9.727}, {'end': 5501.387, 'text': "But at some point, they'll be as deserving of freedom as human beings are.", 'start': 5491.279, 'duration': 10.108}], 'summary': 'Thin time gap before ai gains rights like humans.', 'duration': 39.066, 'max_score': 5462.321, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/3wMKoSRbGVs/pics/3wMKoSRbGVs5462321.jpg'}, {'end': 5606.023, 'src': 'embed', 'start': 5558.111, 'weight': 11, 'content': [{'end': 5572.7, 'text': "it's difficult to understand human behavior and human history without understanding human emotions and why people do things and how emotions drive people to do things,", 'start': 5558.111, 'duration': 14.589}, {'end': 5579.749, 'text': 'and All of that is extremely important in getting psych to understand things.', 'start': 5572.7, 'duration': 7.049}, {'end': 5582.31, 'text': 'For example, in coming up with scenarios.', 'start': 5579.789, 'duration': 2.521}, {'end': 5593.993, 'text': 'So one of the applications that psych does one kind of application it does is to generate plausible scenarios of what might happen and what might happen based on that and what might happen based on that,', 'start': 5582.91, 'duration': 11.083}, {'end': 5594.413, 'text': 'and so on.', 'start': 5593.993, 'duration': 0.42}, {'end': 5602.275, 'text': 'So you generate this ever-expanding sphere, if you will, of possible future things to worry about or think about.', 'start': 5594.773, 'duration': 7.502}, {'end': 5606.023, 'text': 'In some cases,', 'start': 5605.103, 'duration': 0.92}], 'summary': 'Understanding human emotions is crucial in psychology for generating plausible scenarios of future events.', 'duration': 47.912, 'max_score': 5558.111, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/3wMKoSRbGVs/pics/3wMKoSRbGVs5558111.jpg'}, {'end': 5721.259, 'src': 'embed', 'start': 5692.252, 'weight': 12, 'content': [{'end': 5708.536, 'text': 'So I got to ask do you remember off the top of your head how psych is trying to is able to represent various aspects of love that are useful for understanding human nature and therefore integrating into this whole knowledge base of common sense?', 'start': 5692.252, 'duration': 16.284}, {'end': 5709.356, 'text': 'What is love?', 'start': 5708.976, 'duration': 0.38}, {'end': 5721.259, 'text': 'We try to tease apart concepts that have enormous complexities to them and variety to them,', 'start': 5710.316, 'duration': 10.943}], 'summary': 'Psychology explores complex aspects of love to understand human nature and integrate into common sense.', 'duration': 29.007, 'max_score': 5692.252, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/3wMKoSRbGVs/pics/3wMKoSRbGVs5692252.jpg'}], 'start': 4168.752, 'title': "Semantic web, ai tools, interface design, teaching ai, and psyche's role", 'summary': "Explores the concepts of semantic web, knowledge graphs, and ai tools, emphasizing the need for expressive logical languages. it also delves into interface design for effective learning, teaching ai, and psyche's role in generating plausible scenarios for various applications, with a focus on understanding human behavior.", 'chapters': [{'end': 4586.572, 'start': 4168.752, 'title': 'Semantic web and knowledge graphs', 'summary': 'Explores the concept of the semantic web and knowledge graphs, emphasizing the limitations of current search engines and the need for more expressive logical languages to represent and reason with complex content.', 'duration': 417.82, 'highlights': ['The limitations of current search engines and the need for more expressive logical languages Current search engines like Google lack the ability to truly understand and interpret the content of web pages beyond keyword search, highlighting the need for more expressive logical languages to represent and reason with complex content.', 'Transition from semantic web representation to more general logical language The evolution from semantic web representation to more general logical languages, such as first order, second order, and higher order logic, has been driven by the need to represent and reason with complex concepts like beliefs, desires, intentions, expectations, negations, and problem-solving processes.', 'Importance of representing complex concepts and problem-solving processes The ability to represent complex concepts like beliefs, desires, intentions, expectations, negations, and problem-solving processes is crucial, as people discuss, teach, and create rules based on these concepts, highlighting the significance of representing them accurately.']}, {'end': 4877.731, 'start': 4587.292, 'title': 'Semantic web and ai tools', 'summary': 'Discusses the limitations of current implementations in achieving the realization of the semantic web dream, the potential for ai-powered tools to interpret and understand web content, and the benefits of enabling systems to reason with web content for improved user experience and seo.', 'duration': 290.439, 'highlights': ['The limitations of current implementations in achieving the realization of the semantic web dream The current implementations of the semantic web are considered as small steps in the right direction, but not sufficient to realize the semantic web dream.', 'The potential for AI-powered tools to interpret and understand web content The discussion explores the possibility of AI-powered tools to interpret and understand web content, including the idea of semi-automated understanding and footnoting of web content to aid readers and improve comprehension.', 'The benefits of enabling systems to reason with web content for improved user experience and SEO Enabling systems to reason with web content can enhance user experience by connecting users with relevant information and positively impacting SEO through improved content comprehension and recommendation.']}, {'end': 5169.026, 'start': 4878.051, 'title': 'Interface design for effective learning', 'summary': "Discusses the significance of interface design in facilitating productive and joyful learning experiences, exemplified by tesla's autopilot system and the mathcraft program, which leverages the concept of learning by teaching to enhance understanding and engagement.", 'duration': 290.975, 'highlights': ["Tesla's autopilot system creates a sense of teaching the machine, contributing to a more joyful and engaging experience, potentially leading to a significant increase in good quality data being added to the system. Tesla's autopilot system creates a sense of teaching the machine, potentially leading to a significant increase in good quality data being added to the system.", 'The Mathcraft program leverages the concept of learning by teaching, casting players in the role of a mentor to deepen their understanding of the subject, thereby reinforcing engagement and facilitating effective learning. The Mathcraft program leverages the concept of learning by teaching, casting players in the role of a mentor to deepen their understanding of the subject, thereby reinforcing engagement and facilitating effective learning.', 'The chapter emphasizes the importance of interface design that allows users to play the role of a mentor, highlighting the powerful impact of learning by teaching, a concept often overlooked in modern education systems. The chapter emphasizes the importance of interface design that allows users to play the role of a mentor, highlighting the powerful impact of learning by teaching, a concept often overlooked in modern education systems.']}, {'end': 5526.262, 'start': 5169.066, 'title': 'Teaching ai and the future', 'summary': 'Discusses the joy and excitement of teaching ai, the evolving nature of knowledge in ai, and the ethical implications of ai gaining human-like rights and freedoms.', 'duration': 357.196, 'highlights': ['The feeling of teaching a computer something unique and realizing that no one will ever have to teach it again is very exciting and joy-inducing. The feeling of teaching a computer something unique and realizing that no one will ever have to teach it again is very exciting and joy-inducing.', 'The evolving nature of knowledge in AI and the continuous growth and evolution of knowledge empowered by artificial intelligence are crucial aspects to consider. The evolving nature of knowledge in AI and the continuous growth and evolution of knowledge empowered by artificial intelligence are crucial aspects to consider.', 'The ethical implications of AI gaining human-like rights and freedoms, and the shift from treating AI as servants to recognizing their deserving freedoms, are discussed with a belief that it will occur in the near future. The ethical implications of AI gaining human-like rights and freedoms, and the shift from treating AI as servants to recognizing their deserving freedoms, are discussed with a belief that it will occur in the near future.']}, {'end': 6133.606, 'start': 5529.221, 'title': "Psyche's role in generating plausible scenarios", 'summary': 'Discusses how psyche generates plausible scenarios for various applications, including intelligence agencies predicting terrorist threats and personal decision-making, with the system representing complex concepts like love and ambiguity in language, and highlights the importance of human emotions in understanding human behavior.', 'duration': 604.385, 'highlights': ['Psyche generates plausible scenarios for applications such as predicting terrorist threats and personal decision-making. The system is used by intelligence agencies to predict terrorist scenarios and to make personal decisions about careers and professional training.', "Representing complex concepts like love and ambiguity in language is an important aspect of Psyche. Psyche breaks down the concept of love into 50-60 interrelated concepts, showing the complexity and variety within the concept of love. It also deals with ambiguity in language, distinguishing between different meanings of words like 'in' and 'love' to capture their full complexity.", 'Human emotions play a crucial role in understanding human behavior and history. Understanding human emotions is crucial for understanding human behavior and history, as emotions drive people to do things and are important in generating plausible scenarios.']}], 'duration': 1964.854, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/3wMKoSRbGVs/pics/3wMKoSRbGVs4168752.jpg', 'highlights': ['The limitations of current search engines and the need for more expressive logical languages', 'The evolution from semantic web representation to more general logical languages', 'Importance of representing complex concepts and problem-solving processes', 'The potential for AI-powered tools to interpret and understand web content', 'The benefits of enabling systems to reason with web content for improved user experience and SEO', "Tesla's autopilot system creates a sense of teaching the machine", 'The Mathcraft program leverages the concept of learning by teaching', 'The chapter emphasizes the importance of interface design that allows users to play the role of a mentor', 'The feeling of teaching a computer something unique and realizing that no one will ever have to teach it again is very exciting and joy-inducing', 'The evolving nature of knowledge in AI and the continuous growth and evolution of knowledge empowered by artificial intelligence are crucial aspects to consider', 'The ethical implications of AI gaining human-like rights and freedoms', 'Psyche generates plausible scenarios for applications such as predicting terrorist threats and personal decision-making', 'Representing complex concepts like love and ambiguity in language is an important aspect of Psyche', 'Understanding human emotions is crucial for understanding human behavior and history']}, {'end': 6831.744, 'segs': [{'end': 6223.396, 'src': 'embed', 'start': 6190.217, 'weight': 0, 'content': [{'end': 6191.898, 'text': 'Our research lives are finite.', 'start': 6190.217, 'duration': 1.681}, {'end': 6199.462, 'text': "We're going to have limited opportunities to do AI research projects, so you should make each one count.", 'start': 6192.418, 'duration': 7.044}, {'end': 6211.728, 'text': "Don't be afraid of doing a project that's going to take years or or even decades, and don't settle for bump on a log,", 'start': 6200.363, 'duration': 11.365}, {'end': 6223.396, 'text': 'projects that could lead to some published journal article that five people will read and pat you on the head for, and so on.', 'start': 6211.728, 'duration': 11.668}], 'summary': 'Maximize ai research impact with limited opportunities.', 'duration': 33.179, 'max_score': 6190.217, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/3wMKoSRbGVs/pics/3wMKoSRbGVs6190217.jpg'}, {'end': 6446.112, 'src': 'embed', 'start': 6420.699, 'weight': 2, 'content': [{'end': 6425.321, 'text': "I'm interested in getting as many people in the world as possible,", 'start': 6420.699, 'duration': 4.622}, {'end': 6434.346, 'text': 'actively helping and contributing to get us from where we are to really covering all of human knowledge and different human opinion,', 'start': 6425.321, 'duration': 9.025}, {'end': 6437.568, 'text': "including contrasting opinion that's worth representing.", 'start': 6434.346, 'duration': 3.222}, {'end': 6439.029, 'text': "So I think that's one reason.", 'start': 6437.608, 'duration': 1.421}, {'end': 6446.112, 'text': "I don't think there was ever a time where I thought about quitting.", 'start': 6441.129, 'duration': 4.983}], 'summary': 'Goal: engage as many people as possible to contribute to covering all human knowledge and opinions, including contrasting views.', 'duration': 25.413, 'max_score': 6420.699, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/3wMKoSRbGVs/pics/3wMKoSRbGVs6420699.jpg'}, {'end': 6533.873, 'src': 'embed', 'start': 6501.818, 'weight': 1, 'content': [{'end': 6506.362, 'text': "so that we can continue to do this until it's done.", 'start': 6501.818, 'duration': 4.544}, {'end': 6516.729, 'text': "And we're making progress and we're now so close to done that almost all of our work is commercial applications of our technology.", 'start': 6506.822, 'duration': 9.907}, {'end': 6519.83, 'text': 'Five years ago, almost all of our money came from the government.', 'start': 6517.109, 'duration': 2.721}, {'end': 6522.97, 'text': 'Now, virtually none of it comes from the government.', 'start': 6519.91, 'duration': 3.06}, {'end': 6533.873, 'text': 'Almost all of it is from companies that are actually using it for something hospital chains using it for medical reasoning about patients and energy companies using it,', 'start': 6523.03, 'duration': 10.843}], 'summary': 'Most of the revenue now comes from commercial applications, with virtually none from government. hospital chains and energy companies are among the users.', 'duration': 32.055, 'max_score': 6501.818, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/3wMKoSRbGVs/pics/3wMKoSRbGVs6501818.jpg'}, {'end': 6670.462, 'src': 'embed', 'start': 6638.154, 'weight': 3, 'content': [{'end': 6639.915, 'text': 'is not going to be proprietary.', 'start': 6638.154, 'duration': 1.761}, {'end': 6647.421, 'text': "It's going to just become part of the normal psych knowledge base, and it will be openly available to everyone who has access to psych.", 'start': 6639.955, 'duration': 7.466}, {'end': 6651.624, 'text': "So that's an important constraint that we never went back on,", 'start': 6647.921, 'duration': 3.703}, {'end': 6659.67, 'text': 'even when we got pushback from companies which we often did who wanted to claim that almost everything they were telling us was proprietary.', 'start': 6651.624, 'duration': 8.046}, {'end': 6670.462, 'text': "So So there's a line between very domain specific, company specific stuff and the general knowledge that comes from that.", 'start': 6660.591, 'duration': 9.871}], 'summary': 'Psych knowledge base to be openly available to everyone, despite pushback from companies.', 'duration': 32.308, 'max_score': 6638.154, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/3wMKoSRbGVs/pics/3wMKoSRbGVs6638154.jpg'}, {'end': 6768.121, 'src': 'embed', 'start': 6734.115, 'weight': 4, 'content': [{'end': 6739.679, 'text': 'And we try to stay away from projects that would not have that property.', 'start': 6734.115, 'duration': 5.564}, {'end': 6744.203, 'text': 'So let me go back and talk about open psych for a second.', 'start': 6740.54, 'duration': 3.663}, {'end': 6761.477, 'text': "So I've had a lot of trouble expressing and convincing other AI researchers how important it is to use an expressive representation language like we do this higher order logic,", 'start': 6744.443, 'duration': 17.034}, {'end': 6768.121, 'text': 'rather than just using some triple store knowledge graph type representation.', 'start': 6761.477, 'duration': 6.644}], 'summary': 'Emphasizing importance of using expressive representation language like higher order logic in ai research.', 'duration': 34.006, 'max_score': 6734.115, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/3wMKoSRbGVs/pics/3wMKoSRbGVs6734115.jpg'}], 'start': 6133.947, 'title': 'Ai research challenges and knowledge base building', 'summary': 'Delves into the lessons from influential figures in ai research, emphasizing perseverance, tackling hard problems, and seeking diverse contributions. it also discusses the importance of adding to the knowledge base, including the constraints placed on proprietary knowledge and the development of open psych to demonstrate the importance of using expressive representation language.', 'chapters': [{'end': 6543.719, 'start': 6133.947, 'title': 'Challenges and triumphs in ai research', 'summary': 'Discusses the valuable lessons learned from influential figures in ai research, emphasizing the importance of perseverance, tackling hard problems, and seeking diverse contributions to achieve significant progress in the field, despite facing funding challenges and skepticism.', 'duration': 409.772, 'highlights': ['The importance of perseverance and tackling hard problems is emphasized by influential figures in AI research, encouraging researchers to make each project count and not settle for minor achievements. Emphasis on making each AI research project count and not settling for minor achievements.', 'The project has made significant progress, with commercial applications of the technology generating the majority of its funding, including medical reasoning, energy, and supply chain applications. Shift from government funding to commercial applications, including medical reasoning, energy, and supply chain applications.', 'The project aims to harness untapped human potential by raising awareness and seeking contributions from retirees and individuals with more time at home, intending to cover all human knowledge and diverse opinions. Efforts to harness untapped human potential and cover all human knowledge and diverse opinions.']}, {'end': 6831.744, 'start': 6543.839, 'title': 'Open psych and building a knowledge base', 'summary': 'Discusses the importance of adding to the knowledge base, including the constraints placed on proprietary knowledge, the distinction between company-specific and general knowledge, and the development of open psych to demonstrate the importance of using an expressive representation language in ai research.', 'duration': 287.905, 'highlights': ['The chapter emphasizes the importance of adding to the knowledge base, with a focus on respecting proprietary knowledge while making publicly available general human knowledge, contributing to the development of Open Psych (a demonstration of using an expressive representation language) and the constraints placed on proprietary knowledge.', 'The distinction between company-specific and general knowledge is highlighted, with an emphasis on respecting proprietary knowledge while making publicly available general human knowledge, and the challenges faced in convincing other AI researchers of the importance of using an expressive representation language.', 'The development of Open Psych is discussed, focusing on the attempt to demonstrate the importance of using an expressive representation language in AI research and the representation of a subset of psych to showcase the utility of the entire system.']}], 'duration': 697.797, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/3wMKoSRbGVs/pics/3wMKoSRbGVs6133947.jpg', 'highlights': ['Emphasis on perseverance and tackling hard problems by influential figures in AI research.', 'Shift from government funding to commercial applications, including medical reasoning, energy, and supply chain applications.', 'Efforts to harness untapped human potential and cover all human knowledge and diverse opinions.', 'Focus on respecting proprietary knowledge while making publicly available general human knowledge.', 'Challenges faced in convincing other AI researchers of the importance of using an expressive representation language.', 'Discussion on the development of Open Psych to demonstrate the importance of using an expressive representation language in AI research.']}, {'end': 8407.074, 'segs': [{'end': 6947.673, 'src': 'embed', 'start': 6921.961, 'weight': 1, 'content': [{'end': 6927.063, 'text': 'This is the exact same problem that Google had to deal with with the early version of TensorFlow.', 'start': 6921.961, 'duration': 5.102}, {'end': 6928.524, 'text': "It's still dealing with that.", 'start': 6927.443, 'duration': 1.081}, {'end': 6936.968, 'text': "They had to basically break compatibility with the past several times, and that's only over a period of a couple years.", 'start': 6928.644, 'duration': 8.324}, {'end': 6947.673, 'text': "But they, I think, successfully opened up, it's a very risky, very gutsy move to open up TensorFlow and then PyTorch on the Facebook side.", 'start': 6937.708, 'duration': 9.965}], 'summary': "Google faced compatibility issues with tensorflow, but successfully opened it up, along with facebook's pytorch.", 'duration': 25.712, 'max_score': 6921.961, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/3wMKoSRbGVs/pics/3wMKoSRbGVs6921961.jpg'}, {'end': 7083.985, 'src': 'embed', 'start': 7058.891, 'weight': 0, 'content': [{'end': 7067.516, 'text': 'machine learning, has a phase shift towards like, or at least in part, towards more like what you might call symbolic AI.', 'start': 7058.891, 'duration': 8.625}, {'end': 7076.541, 'text': 'this whole place where psych is like at the center of and then, as you know, that requires a little bit leap of faith,', 'start': 7067.516, 'duration': 9.025}, {'end': 7083.985, 'text': "because you're now surfing and there'll be obviously competitors that will pop up and start making you nervous and all that kind of stuff.", 'start': 7076.541, 'duration': 7.444}], 'summary': 'Machine learning is shifting towards symbolic ai, requiring a leap of faith and facing competition.', 'duration': 25.094, 'max_score': 7058.891, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/3wMKoSRbGVs/pics/3wMKoSRbGVs7058891.jpg'}, {'end': 7229.16, 'src': 'embed', 'start': 7204.292, 'weight': 2, 'content': [{'end': 7209.614, 'text': 'separate that from the heuristic problem of how can the system reason efficiently with what it knows.', 'start': 7204.292, 'duration': 5.322}, {'end': 7216.416, 'text': 'And so, instead of trying to pick one representation language,', 'start': 7210.314, 'duration': 6.102}, {'end': 7224.799, 'text': 'which was the sweet spot or the best trade-off point between expressiveness of the language and efficiency of the language,', 'start': 7216.416, 'duration': 8.383}, {'end': 7229.16, 'text': 'if you had to pick one knowledge graphs would probably be associative.', 'start': 7224.799, 'duration': 4.361}], 'summary': 'Knowledge graphs are likely the best representation language for efficient system reasoning.', 'duration': 24.868, 'max_score': 7204.292, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/3wMKoSRbGVs/pics/3wMKoSRbGVs7204292.jpg'}, {'end': 7290.315, 'src': 'embed', 'start': 7257.974, 'weight': 3, 'content': [{'end': 7260.815, 'text': "I can make progress on what you're trying to do over here.", 'start': 7257.974, 'duration': 2.841}, {'end': 7265.677, 'text': 'I have a special method that will contribute a little bit toward a solution.', 'start': 7261.275, 'duration': 4.402}, {'end': 7268.618, 'text': 'Of course, some subset of that.', 'start': 7266.097, 'duration': 2.521}, {'end': 7269.458, 'text': 'Yes, exactly.', 'start': 7268.718, 'duration': 0.74}, {'end': 7276.12, 'text': 'So by now, we have over a thousand of these heuristic-level modules, and they function as a kind of community of agents.', 'start': 7269.518, 'duration': 6.602}, {'end': 7279.583, 'text': "And there's one of them, which is a general theorem prover.", 'start': 7276.96, 'duration': 2.623}, {'end': 7284.488, 'text': "And in theory, that's the only one you need.", 'start': 7280.244, 'duration': 4.244}, {'end': 7290.315, 'text': 'But in practice, it always takes so long that you never want to call on it.', 'start': 7285.349, 'duration': 4.966}], 'summary': "Over a thousand heuristic-level modules act as a community of agents, with one being a general theorem prover that's too slow in practice.", 'duration': 32.341, 'max_score': 7257.974, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/3wMKoSRbGVs/pics/3wMKoSRbGVs7257974.jpg'}, {'end': 7518.632, 'src': 'embed', 'start': 7492.754, 'weight': 4, 'content': [{'end': 7497.295, 'text': 'where one of them was this EL-HL split and lots of HL modules.', 'start': 7492.754, 'duration': 4.541}, {'end': 7505.718, 'text': "Another one was using meta and meta-meta level reasoning to reason about the reasoning that's going on and so on.", 'start': 7497.395, 'duration': 8.323}, {'end': 7512.405, 'text': "And, you know, 150 breakthroughs may sound like a lot, but, you know, if you divide by 37 years, it's not as impressive.", 'start': 7506.519, 'duration': 5.886}, {'end': 7518.632, 'text': "So there's these kind of heuristic modules that really help improve the inference.", 'start': 7513.707, 'duration': 4.925}], 'summary': 'Discussing el-hl split, meta-meta level reasoning, 150 breakthroughs over 37 years, and heuristic modules for inference improvement.', 'duration': 25.878, 'max_score': 7492.754, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/3wMKoSRbGVs/pics/3wMKoSRbGVs7492754.jpg'}, {'end': 7678.606, 'src': 'embed', 'start': 7648.738, 'weight': 5, 'content': [{'end': 7653.919, 'text': "But Psyche, as far as I understand, maybe you can correct me, there's a bunch of Lisp in it.", 'start': 7648.738, 'duration': 5.181}, {'end': 7657.84, 'text': "Yeah, so it's based on Lisp code that we produced.", 'start': 7654.699, 'duration': 3.141}, {'end': 7661.741, 'text': 'Most of the programming is still going on in a dialect of Lisp.', 'start': 7657.96, 'duration': 3.781}, {'end': 7670.742, 'text': 'And then for efficiency reasons, that gets automatically translated into things like Java or C.', 'start': 7662.201, 'duration': 8.541}, {'end': 7673.143, 'text': "Nowadays, it's almost all translated into Java.", 'start': 7670.742, 'duration': 2.401}, {'end': 7678.606, 'text': "because Java has gotten good enough that that's really all we need to do.", 'start': 7673.683, 'duration': 4.923}], 'summary': 'Psyche is based on lisp code, translated into java for efficiency.', 'duration': 29.868, 'max_score': 7648.738, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/3wMKoSRbGVs/pics/3wMKoSRbGVs7648738.jpg'}, {'end': 8020.622, 'src': 'embed', 'start': 7988.936, 'weight': 7, 'content': [{'end': 7995.422, 'text': 'There are lots of cases where storage is cheap today, and so by having this extra redundant data structure,', 'start': 7988.936, 'duration': 6.486}, {'end': 8001.066, 'text': 'we can answer this commonly occurring type of question very, very efficiently.', 'start': 7995.942, 'duration': 5.124}, {'end': 8010.033, 'text': 'Let me give you one other analog of that which is something we call rule macro predicates, which is.', 'start': 8002.187, 'duration': 7.846}, {'end': 8020.622, 'text': "we'll see this complicated rule and we'll notice that things very much like it syntactically come up again and again and again.", 'start': 8010.033, 'duration': 10.589}], 'summary': 'Redundant data structure allows efficient storage and retrieval of commonly occurring queries.', 'duration': 31.686, 'max_score': 7988.936, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/3wMKoSRbGVs/pics/3wMKoSRbGVs7988936.jpg'}, {'end': 8065.808, 'src': 'embed', 'start': 8034.595, 'weight': 8, 'content': [{'end': 8047.641, 'text': 'And now we have effectively converted some complicated if-then rule that might have to have inference done on it into some ground atomic formula,', 'start': 8034.595, 'duration': 13.046}, {'end': 8052.582, 'text': 'which is just the name of a relation, and a few arguments and so on.', 'start': 8047.641, 'duration': 4.941}, {'end': 8062.106, 'text': 'And so converting commonly occurring types or schemas of rules into brand new predicates, brand new functions,', 'start': 8053.123, 'duration': 8.983}, {'end': 8065.808, 'text': 'turns out to enormously speed up the inference process.', 'start': 8062.106, 'duration': 3.702}], 'summary': 'Converting complex rules into new predicates speeds up inference process.', 'duration': 31.213, 'max_score': 8034.595, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/3wMKoSRbGVs/pics/3wMKoSRbGVs8034595.jpg'}, {'end': 8116.205, 'src': 'embed', 'start': 8090.146, 'weight': 6, 'content': [{'end': 8095.669, 'text': 'and the inference engine from a epistemological, philosophical aspect, fascinating.', 'start': 8090.146, 'duration': 5.523}, {'end': 8101.472, 'text': "But you know, it is also, you're running a company and there's some stuff that has to remain private.", 'start': 8096.329, 'duration': 5.143}, {'end': 8102.973, 'text': "And it's sad.", 'start': 8101.492, 'duration': 1.481}, {'end': 8106.656, 'text': "Well, here's something that may make you feel better, a little bit better.", 'start': 8103.433, 'duration': 3.223}, {'end': 8116.205, 'text': "We've formed a not-for-profit company called the Knowledge Activization Institute, NACS, K-N-A-X.", 'start': 8108.398, 'duration': 7.807}], 'summary': 'A not-for-profit company called knowledge activization institute (nacs, k-n-a-x) has been formed.', 'duration': 26.059, 'max_score': 8090.146, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/3wMKoSRbGVs/pics/3wMKoSRbGVs8090146.jpg'}], 'start': 6831.744, 'title': 'Challenges and strategies in open psych', 'summary': "Discusses the challenges of promoting open psych, limited adoption by researchers, leveraging the open-source community, and the evolution of representation language for efficient reasoning in ai, with over a thousand heuristic-level modules in practical use, as well as the significance of lisp programming language in psyche's development, with specific mention of 150 breakthrough ideas leading to dramatic speedups, and the concept of ontological engineering, including examples of using redundant data structures to efficiently answer commonly occurring questions and the formation of a not-for-profit organization to identify and train individuals for high-paying ontological engineering jobs, independent of traditional education.", 'chapters': [{'end': 7083.985, 'start': 6831.744, 'title': 'Challenges of open psych', 'summary': 'Discusses the challenges faced in promoting open psych, the limited adoption by researchers, and the potential of leveraging the open-source community to further develop the platform, potentially leading to a shift in the dominance of machine learning towards symbolic ai.', 'duration': 252.241, 'highlights': ['The limited adoption of Open Psych by researchers and robotics companies, despite its potential, demonstrates the challenges in promoting the platform. Despite the vast knowledge and potential of Open Psych, researchers and robotics companies have shown limited interest in adopting it, opting to make do with the released portion rather than seeking access to the full Psyche.', 'The potential of leveraging the open-source community to develop Open Psych further, potentially leading to a shift in the dominance of machine learning towards symbolic AI, is discussed. There is potential in leveraging the open-source community to develop Open Psych further, potentially leading to a shift in the dominance of machine learning towards symbolic AI, which requires a leap of faith and may attract competitors.', "The comparison to Google's TensorFlow and Facebook's PyTorch illustrates the potential risks and benefits of opening up a platform, emphasizing the need to find a balance in releasing tooling while retaining value. The comparison to Google's TensorFlow and Facebook's PyTorch demonstrates the risks and benefits of opening up a platform, highlighting the need to find a balance in releasing tooling while retaining value similar to what the major tech companies have done."]}, {'end': 7320.355, 'start': 7084.005, 'title': 'Open sourcing and leveraging community', 'summary': 'Discusses the challenges of open sourcing, the strategy for collaborating on research, and the evolution of representation language for efficient reasoning in ai, with over a thousand heuristic-level modules in practical use.', 'duration': 236.35, 'highlights': ['The strategy for collaborating on research is to make the entire knowledge base and executable copies of all code available, but only very limited source code access for improvement or collaboration. The approach for collaborating on research involves providing access to the entire knowledge base and executable code, but with very limited source code access for potential improvement or collaboration.', 'The evolution of representation language for efficient reasoning in AI involved starting with knowledge graphs and associative triples, but later developing a higher-order logic for epistemological-level language and heuristic-level modules for efficient reasoning. The evolution of representation language for efficient reasoning in AI began with knowledge graphs and associative triples, then progressed to a higher-order logic for epistemological-level language and heuristic-level modules for efficient reasoning.', 'Over a thousand heuristic-level modules function as a community of agents, with a general theorem prover being the only one theoretically needed, but practically not being utilized due to the efficiency of the other agents. There are over a thousand heuristic-level modules operating as a community of agents, where a general theorem prover is theoretically the only one needed, but in practice, it is not utilized due to the efficiency of the other agents.']}, {'end': 7940.088, 'start': 7320.915, 'title': "Psyche's inference process and programming language", 'summary': "Discusses how the community of agents in psyche collaborates to provide answers, the use of meta-reasoning to improve speed, and the significance of lisp programming language in psyche's development, with specific mention of 150 breakthrough ideas leading to dramatic speedups.", 'duration': 619.173, 'highlights': ["Psyche's community of agents collaborates to provide answers and solutions, with 150 breakthrough ideas leading to dramatic speedups in the inference process.", 'Meta-reasoning is employed to improve speed by providing tactical and strategic advice, resulting in about 150 breakthrough ideas that have led to dramatic speedups in the inference process.', "The significance of Lisp programming language in Psyche's development, with 150 breakthrough ideas contributing to dramatic speedups and the preference for Lisp due to its natural impedance match with the knowledge representation."]}, {'end': 8407.074, 'start': 7940.248, 'title': 'Ontological engineering and knowledge activization', 'summary': 'Discusses the concept of ontological engineering, including examples of using redundant data structures to efficiently answer commonly occurring questions and the formation of a not-for-profit organization to identify and train individuals for high-paying ontological engineering jobs, independent of traditional education.', 'duration': 466.826, 'highlights': ['The formation of a not-for-profit organization, Knowledge Activization Institute (NACS), to identify and train individuals for high-paying ontological engineering jobs, independent of traditional education. The speaker discusses the formation of the not-for-profit organization, Knowledge Activization Institute (NACS), aimed at identifying individuals with latent talent for ontological engineering, offering them scholarships for training, and helping place them in companies that need trained ontological engineers.', 'Using redundant data structures like the transitive closure to efficiently answer commonly occurring questions. The speaker explains the use of redundant data structures, such as the transitive closure, to store relationships, enabling efficient and quick lookup for commonly occurring questions, reducing the need for complex graph walks or theorem proving.', 'Examples of converting commonly occurring types of rules into new predicates and functions to speed up the inference process. The speaker provides examples of converting commonly occurring types or schemas of rules into brand new predicates and functions, which significantly speeds up the inference process in ontological engineering.']}], 'duration': 1575.33, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/3wMKoSRbGVs/pics/3wMKoSRbGVs6831744.jpg', 'highlights': ['The potential of leveraging the open-source community to develop Open Psych further, potentially leading to a shift in the dominance of machine learning towards symbolic AI, is discussed.', "The comparison to Google's TensorFlow and Facebook's PyTorch illustrates the potential risks and benefits of opening up a platform, emphasizing the need to find a balance in releasing tooling while retaining value similar to what the major tech companies have done.", 'The evolution of representation language for efficient reasoning in AI began with knowledge graphs and associative triples, then progressed to a higher-order logic for epistemological-level language and heuristic-level modules for efficient reasoning.', 'Over a thousand heuristic-level modules function as a community of agents, where a general theorem prover is theoretically the only one needed, but in practice, it is not utilized due to the efficiency of the other agents.', 'Meta-reasoning is employed to improve speed by providing tactical and strategic advice, resulting in about 150 breakthrough ideas that have led to dramatic speedups in the inference process.', "The significance of Lisp programming language in Psyche's development, with 150 breakthrough ideas contributing to dramatic speedups and the preference for Lisp due to its natural impedance match with the knowledge representation.", 'The formation of a not-for-profit organization, Knowledge Activization Institute (NACS), to identify and train individuals for high-paying ontological engineering jobs, independent of traditional education.', 'Using redundant data structures like the transitive closure to efficiently answer commonly occurring questions.', 'Examples of converting commonly occurring types or schemas of rules into brand new predicates and functions, which significantly speeds up the inference process in ontological engineering.']}, {'end': 9223.992, 'segs': [{'end': 8685.358, 'src': 'embed', 'start': 8656.989, 'weight': 0, 'content': [{'end': 8663.615, 'text': 'The test has to be something involving depth of reasoning and recursiveness of reasoning.', 'start': 8656.989, 'duration': 6.626}, {'end': 8666.319, 'text': 'the ability to answer repeated.', 'start': 8663.615, 'duration': 2.704}, {'end': 8668.621, 'text': 'why questions about the answer you just gave?', 'start': 8666.319, 'duration': 2.302}, {'end': 8673.707, 'text': "It's how many why questions in a row can you keep answering? Something like that.", 'start': 8668.641, 'duration': 5.066}, {'end': 8679.432, 'text': 'Just have a young, curious child and an AI system.', 'start': 8676.109, 'duration': 3.323}, {'end': 8685.358, 'text': "And how long will an AI system last before it wants to quit? Yes, and again, that's not the only test.", 'start': 8679.793, 'duration': 5.565}], 'summary': "Test involves depth of reasoning, recursiveness, and ability to answer repeated 'why' questions. measures ai system's endurance.", 'duration': 28.369, 'max_score': 8656.989, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/3wMKoSRbGVs/pics/3wMKoSRbGVs8656989.jpg'}, {'end': 8760.269, 'src': 'embed', 'start': 8733.835, 'weight': 1, 'content': [{'end': 8741.379, 'text': "And there's some kind of hard coding heuristics tricks that you try to pull it all together to make the thing work in the end.", 'start': 8733.835, 'duration': 7.544}, {'end': 8742.38, 'text': 'for this thing right?', 'start': 8741.379, 'duration': 1.001}, {'end': 8747.402, 'text': 'That seems to be one of the lessons with AI is like.', 'start': 8742.86, 'duration': 4.542}, {'end': 8749.103, 'text': "that's the fastest way to get a solution.", 'start': 8747.402, 'duration': 1.701}, {'end': 8750.264, 'text': "that's pretty damn impressive.", 'start': 8749.103, 'duration': 1.161}, {'end': 8759.208, 'text': "So here's what I would say is that as impressive as that was, it made some mistakes.", 'start': 8750.544, 'duration': 8.664}, {'end': 8760.269, 'text': 'But more importantly..', 'start': 8759.288, 'duration': 0.981}], 'summary': 'Ai uses hard coding heuristics to quickly find solutions, impressive but prone to mistakes.', 'duration': 26.434, 'max_score': 8733.835, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/3wMKoSRbGVs/pics/3wMKoSRbGVs8733835.jpg'}, {'end': 8878.272, 'src': 'embed', 'start': 8845.236, 'weight': 3, 'content': [{'end': 8848.317, 'text': 'And if it fails several of these tests,', 'start': 8845.236, 'duration': 3.081}, {'end': 8853.239, 'text': "it's really hard to argue that it really understands what it's doing and that it really is generally intelligent.", 'start': 8848.317, 'duration': 4.922}, {'end': 8859.762, 'text': "So to pass all of those tests, you know, we've talked a lot about psych and knowledge and reasoning.", 'start': 8853.379, 'duration': 6.383}, {'end': 8871.568, 'text': 'Do you think this AI system would need to have some other human-like elements, for example a body or a physical manifestation in this world?', 'start': 8860.783, 'duration': 10.785}, {'end': 8878.272, 'text': 'And another one, which seems to be fundamental to the human experience, is consciousness.', 'start': 8872.629, 'duration': 5.643}], 'summary': 'To prove general intelligence, ai must pass tests on understanding, knowledge, reasoning, and possibly exhibit human-like elements and consciousness.', 'duration': 33.036, 'max_score': 8845.236, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/3wMKoSRbGVs/pics/3wMKoSRbGVs8845236.jpg'}, {'end': 8947.123, 'src': 'embed', 'start': 8917.641, 'weight': 2, 'content': [{'end': 8929.429, 'text': 'but it has to understand it in the same way that Helen Keller was perfectly intelligent and able to talk about colors and sounds and shapes and so on,', 'start': 8917.641, 'duration': 11.788}, {'end': 8934.512, 'text': "even though she didn't directly experience all the same things that the rest of us do.", 'start': 8929.429, 'duration': 5.083}, {'end': 8944.683, 'text': 'So knowledge of it and being able to correctly make use of that is certainly an important facility.', 'start': 8934.572, 'duration': 10.111}, {'end': 8947.123, 'text': 'But actually having a body.', 'start': 8945.003, 'duration': 2.12}], 'summary': "Understanding without direct experience is important, like helen keller's ability to talk about colors and sounds despite not directly experiencing them.", 'duration': 29.482, 'max_score': 8917.641, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/3wMKoSRbGVs/pics/3wMKoSRbGVs8917641.jpg'}], 'start': 8407.494, 'title': 'Ai impact on reasoning and general intelligence tests', 'summary': 'Delves into the impact of education on reasoning ability, challenges in adopting innovative education methods, criteria for testing ai intelligence, battery of tests for determining general intelligence in ai, and the significance of fear of mortality for intelligence.', 'chapters': [{'end': 8823.227, 'start': 8407.494, 'title': 'Ai reasoning and education impact', 'summary': "Discusses the impact of education on reasoning ability, the challenges of adopting innovative education methods, and the criteria for testing ai intelligence, including the need for depth of reasoning and avoiding mistakes that humans wouldn't make.", 'duration': 415.733, 'highlights': ['The impact of education on reasoning ability is discussed, highlighting that for many people, education can narrow thinking rather than expand it, impacting the ability to introspect deeply. Education can narrow thinking for many people, impacting their ability to introspect deeply.', 'Challenges of adopting innovative education methods are outlined, with the example of Mathcraft not being widely used due to the focus on standardized test scores rather than increasing understanding. Mathcraft not widely used due to focus on standardized test scores over understanding.', 'The criteria for testing AI intelligence is discussed, emphasizing the importance of depth of reasoning, recursiveness of reasoning, and the ability to engage in argumentation. Importance of depth of reasoning, recursiveness of reasoning, and argumentation in testing AI intelligence.', "The need to avoid AI making mistakes that humans wouldn't make is highlighted, with an example of Watson making mistakes that no normal human would make. Highlighting the need for AI to avoid making mistakes that humans wouldn't make."]}, {'end': 9223.992, 'start': 8824.048, 'title': 'Tests for general intelligence and consciousness in ai', 'summary': 'Discusses the battery of tests for determining general intelligence in ai, the necessity of human-like elements like a body or consciousness, and the significance of fear of mortality for intelligence, questioning their relevance in ai development.', 'duration': 399.944, 'highlights': ["The battery of tests for AI to achieve general intelligence includes the ability to talk about having a body, sensations, and emotions, similar to how Helen Keller was able to communicate about experiences she didn't directly encounter.", 'The understanding and correct use of knowledge, emotions, and sensations are essential for an AI to demonstrate intelligence.', 'The discussion about whether an AI needs a body or consciousness to achieve general intelligence, with the speaker expressing the belief that these elements are not necessary.', 'The consideration of fear of mortality and its impact on human behavior, debating its necessity for AI to possess consciousness and intelligence.']}], 'duration': 816.498, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/3wMKoSRbGVs/pics/3wMKoSRbGVs8407494.jpg', 'highlights': ['Importance of depth of reasoning, recursiveness of reasoning, and argumentation in testing AI intelligence.', "The need for AI to avoid making mistakes that humans wouldn't make.", "The battery of tests for AI to achieve general intelligence includes the ability to talk about having a body, sensations, and emotions, similar to how Helen Keller was able to communicate about experiences she didn't directly encounter.", 'The understanding and correct use of knowledge, emotions, and sensations are essential for an AI to demonstrate intelligence.']}, {'end': 10366.72, 'segs': [{'end': 9282.072, 'src': 'embed', 'start': 9257.718, 'weight': 0, 'content': [{'end': 9265.001, 'text': 'i think like i think ai really needs to understand death deeply in order to be able to drive a car, for example.', 'start': 9257.718, 'duration': 7.283}, {'end': 9268.063, 'text': "i i think there's just some like there.", 'start': 9265.001, 'duration': 3.062}, {'end': 9270.044, 'text': 'No, I really disagree.', 'start': 9268.303, 'duration': 1.741}, {'end': 9276.829, 'text': 'I think it needs to understand the value of human life, especially the value of human life to other humans,', 'start': 9270.124, 'duration': 6.705}, {'end': 9282.072, 'text': 'and understand that certain things are more important than other things.', 'start': 9276.829, 'duration': 5.243}], 'summary': 'Ai needs to deeply understand death and human life to drive, prioritize values.', 'duration': 24.354, 'max_score': 9257.718, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/3wMKoSRbGVs/pics/3wMKoSRbGVs9257718.jpg'}, {'end': 9606.647, 'src': 'embed', 'start': 9570.267, 'weight': 1, 'content': [{'end': 9572.508, 'text': 'getting stamped to death by a machine.', 'start': 9570.267, 'duration': 2.241}, {'end': 9583.512, 'text': 'And instead of being a small story that said, despondent worker commits suicide, it was front page news that effectively said, robot kills worker.', 'start': 9572.868, 'duration': 10.644}, {'end': 9593.658, 'text': 'because the public is just waiting for stories about AI kills phonogenic family of five type stories.', 'start': 9584.252, 'duration': 9.406}, {'end': 9606.647, 'text': 'And even if you could show that nationwide, this system saved more lives than it cost and prevented more injuries than it caused, and so on.', 'start': 9594.139, 'duration': 12.508}], 'summary': 'Public fear of ai causing accidents is sensationalized despite potential benefits.', 'duration': 36.38, 'max_score': 9570.267, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/3wMKoSRbGVs/pics/3wMKoSRbGVs9570267.jpg'}, {'end': 9735.836, 'src': 'embed', 'start': 9705.305, 'weight': 2, 'content': [{'end': 9707.868, 'text': 'that might work and that might be worth considering.', 'start': 9705.305, 'duration': 2.563}, {'end': 9710.411, 'text': 'That is a damn good question.', 'start': 9708.929, 'duration': 1.482}, {'end': 9719.04, 'text': "Given that the AGI is going to be somewhat different from human intelligence, it's still going to make some mistakes that we wouldn't make,", 'start': 9710.831, 'duration': 8.209}, {'end': 9723.372, 'text': "but it's also possibly going to notice some blind spots we have.", 'start': 9719.04, 'duration': 4.332}, {'end': 9732.735, 'text': 'And I would love as a test of is it really on a par with our intelligences?', 'start': 9724.253, 'duration': 8.482}, {'end': 9735.836, 'text': 'can it help spot some of the blind spots that we have?', 'start': 9732.735, 'duration': 3.101}], 'summary': 'Agi may notice human blind spots and help identify them, testing its intelligence.', 'duration': 30.531, 'max_score': 9705.305, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/3wMKoSRbGVs/pics/3wMKoSRbGVs9705305.jpg'}, {'end': 9962.018, 'src': 'embed', 'start': 9903.244, 'weight': 3, 'content': [{'end': 9917.212, 'text': "And I think that one of the things that would change that sad reality is if we had trusted AGIs that could help take a step back and question some of the paradigms that we're currently locked into.", 'start': 9903.244, 'duration': 13.968}, {'end': 9923.153, 'text': 'Yeah, it would accelerate the paradigm shifts in human science and progress.', 'start': 9917.972, 'duration': 5.181}, {'end': 9931.015, 'text': "You've lived a very interesting life where you thought about big ideas and you stuck with them.", 'start': 9924.854, 'duration': 6.161}, {'end': 9941.797, 'text': 'Can you give advice to young people today somebody in high school, somebody undergrad, about career, about life?', 'start': 9932.475, 'duration': 9.322}, {'end': 9948.975, 'text': "I'd say you can make a difference, but in order to make a difference,", 'start': 9943.874, 'duration': 5.101}, {'end': 9962.018, 'text': "you're going to have to have the courage to follow through with ideas which other people might not immediately understand or support.", 'start': 9948.975, 'duration': 13.043}], 'summary': 'Trusted agis could accelerate paradigm shifts; have courage to pursue unconventional ideas for making a difference.', 'duration': 58.774, 'max_score': 9903.244, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/3wMKoSRbGVs/pics/3wMKoSRbGVs9903244.jpg'}, {'end': 10102.391, 'src': 'embed', 'start': 10069.912, 'weight': 7, 'content': [{'end': 10074.535, 'text': 'That said, you said life is finite, and we talked about mortality as a feature of AGI.', 'start': 10069.912, 'duration': 4.623}, {'end': 10080.518, 'text': 'Do you think about your own mortality? Are you afraid of death? Sure.', 'start': 10075.315, 'duration': 5.203}, {'end': 10082.64, 'text': "I'd be crazy if I weren't.", 'start': 10080.538, 'duration': 2.102}, {'end': 10087.523, 'text': "And as I get older, I'm now over 70.", 'start': 10083.16, 'duration': 4.363}, {'end': 10098.709, 'text': "So as I get older, it's more on my mind, especially as acquaintances and friends and especially mentors, one by one are dying.", 'start': 10087.523, 'duration': 11.186}, {'end': 10102.391, 'text': "So I can't avoid thinking about mortality.", 'start': 10098.87, 'duration': 3.521}], 'summary': 'At over 70, the speaker contemplates mortality with the loss of acquaintances and mentors.', 'duration': 32.479, 'max_score': 10069.912, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/3wMKoSRbGVs/pics/3wMKoSRbGVs10069912.jpg'}, {'end': 10147.997, 'src': 'embed', 'start': 10121.419, 'weight': 5, 'content': [{'end': 10132.622, 'text': 'So we really do want psych to make an impact in the world commercially, physically, metaphysically, in the next small number of years two, three,', 'start': 10121.419, 'duration': 11.203}, {'end': 10135.865, 'text': 'five years, not two, three, five decades anymore.', 'start': 10132.622, 'duration': 3.243}, {'end': 10147.997, 'text': 'And so this is really driving me toward this sort of commercialization and increasingly widespread application of psych,', 'start': 10136.446, 'duration': 11.551}], 'summary': 'Psych aims to make a commercial and widespread impact in the world within the next two to five years.', 'duration': 26.578, 'max_score': 10121.419, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/3wMKoSRbGVs/pics/3wMKoSRbGVs10121419.jpg'}, {'end': 10261.908, 'src': 'embed', 'start': 10231.745, 'weight': 6, 'content': [{'end': 10238.547, 'text': "It's hard to imagine what life would be like if these people hadn't done what they did.", 'start': 10231.745, 'duration': 6.802}, {'end': 10239.807, 'text': "So that's one thing.", 'start': 10239.127, 'duration': 0.68}, {'end': 10245.992, 'text': "that i'd like to be remembered as another is that so the creator,", 'start': 10242.208, 'duration': 3.784}, {'end': 10260.627, 'text': 'one of the originators of this gigantic knowledge store and acquisition system that is likely to be at the center of whatever this future ai thing will look like.', 'start': 10245.992, 'duration': 14.635}, {'end': 10261.908, 'text': 'yes, exactly.', 'start': 10260.627, 'duration': 1.281}], 'summary': 'The speaker wants to be remembered as an originator of a knowledge acquisition system likely to be central to future ai.', 'duration': 30.163, 'max_score': 10231.745, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/3wMKoSRbGVs/pics/3wMKoSRbGVs10231745.jpg'}], 'start': 9224.652, 'title': "Ai's impact on ethics, morality, and paradigm shifts", 'summary': "Explores ai's comprehension of ethics, morality, and human value, as well as challenges faced by autonomous vehicles. it delves into societal resistance and agi's potential role in addressing global problems. additionally, it discusses paradigm shifts in science and the legacy of doug lennett in the ai field.", 'chapters': [{'end': 9840.334, 'start': 9224.652, 'title': 'Ai understanding fear of mortality', 'summary': 'Explores the concept of ai understanding fear of mortality, discussing the need for ai to comprehend ethics, morality, and human value, as well as the challenges faced by autonomous vehicles in making ethical decisions. it also delves into the societal resistance to ai advancement and the potential role of agi in identifying novel solutions to global problems.', 'duration': 615.682, 'highlights': ['The challenges of AI understanding the value of human life and making ethical decisions, particularly in the context of autonomous vehicles, are discussed, highlighting the complexity and nuances involved in such decision-making processes.', "The societal resistance to AI advancement and the public's inclination towards negative stories and failures, despite the potential benefits and successes, is emphasized, shedding light on the challenges faced by AI in gaining public trust and acceptance.", 'The potential role of AGI in identifying blind spots and proposing novel solutions to global problems, drawing parallels to historical scientific discoveries that resulted from considering non-obvious answers to questions, is explored, presenting an intriguing perspective on the potential contributions of AGI to addressing pressing global issues.']}, {'end': 10069.28, 'start': 9840.334, 'title': 'Ai paradigms and life advice', 'summary': 'Discusses the resistance to paradigm shifts in science, the potential of agis to accelerate such shifts, and offers advice on making a difference in life and career choices, emphasizing the need for courage and long-term commitment.', 'duration': 228.946, 'highlights': ['The resistance to paradigm shifts in science and the potential for AGIs to accelerate such shifts. The chapter emphasizes the difficulty in breaking free from established paradigms and suggests that trusted AGIs could facilitate the acceleration of paradigm shifts in human science and progress.', 'The advice on making a difference in life and career choices, emphasizing the need for courage and long-term commitment. The speaker advises young people to have the courage to follow through with ideas that others may not immediately understand or support, and emphasizes the importance of making decade-sized bets count in personal relationships, career choices, and making discoveries.', 'The historical example of the resistance to accepting the bacterial cause of stomach ulcers. The chapter discusses the historical example of the resistance to accepting the bacterial cause of stomach ulcers, where a doctor had to inject himself with the bacteria to demonstrate the link, highlighting the challenge of breaking established beliefs.']}, {'end': 10366.72, 'start': 10069.912, 'title': "Doug lennett: ai pioneer's legacy", 'summary': "Reveals doug lennett's urgency to make an impact in the ai field within the next few years due to his increasing awareness of mortality, driving him towards commercialization and widespread application of psych, with the hope of being remembered as a pioneer of ubiquitous ai and the creator of a vast knowledge acquisition system.", 'duration': 296.808, 'highlights': ["Doug Lennett's urgency to make an impact in the AI field within the next few years Doug Lennett feels the impetus to succeed in a small number of years, driving him towards commercialization and widespread application of psych, no longer having decades to work on this.", 'Desire to be remembered as a pioneer of ubiquitous AI and the creator of a vast knowledge acquisition system Doug Lennett hopes to be remembered as one of the pioneers of ubiquitous AI, and as the creator of a gigantic knowledge store and acquisition system that will likely be at the center of the future of AI.', "Impact of mortality on Doug Lennett's perspective Doug Lennett's increasing awareness of mortality adds impetus to his need to succeed in the AI field within the next few years, leading to a sense of urgency and driving him towards commercialization and widespread application of psych."]}], 'duration': 1142.068, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/3wMKoSRbGVs/pics/3wMKoSRbGVs9224652.jpg', 'highlights': ['The challenges of AI understanding the value of human life and making ethical decisions, particularly in the context of autonomous vehicles, are discussed, highlighting the complexity and nuances involved in such decision-making processes.', "The societal resistance to AI advancement and the public's inclination towards negative stories and failures, despite the potential benefits and successes, is emphasized, shedding light on the challenges faced by AI in gaining public trust and acceptance.", 'The potential role of AGI in identifying blind spots and proposing novel solutions to global problems, drawing parallels to historical scientific discoveries that resulted from considering non-obvious answers to questions, is explored, presenting an intriguing perspective on the potential contributions of AGI to addressing pressing global issues.', 'The resistance to paradigm shifts in science and the potential for AGIs to accelerate such shifts. The chapter emphasizes the difficulty in breaking free from established paradigms and suggests that trusted AGIs could facilitate the acceleration of paradigm shifts in human science and progress.', 'The advice on making a difference in life and career choices, emphasizing the need for courage and long-term commitment. The speaker advises young people to have the courage to follow through with ideas that others may not immediately understand or support, and emphasizes the importance of making decade-sized bets count in personal relationships, career choices, and making discoveries.', "Doug Lennett's urgency to make an impact in the AI field within the next few years Doug Lennett feels the impetus to succeed in a small number of years, driving him towards commercialization and widespread application of psych, no longer having decades to work on this.", 'Desire to be remembered as a pioneer of ubiquitous AI and the creator of a vast knowledge acquisition system Doug Lennett hopes to be remembered as one of the pioneers of ubiquitous AI, and as the creator of a gigantic knowledge store and acquisition system that will likely be at the center of the future of AI.', "Impact of mortality on Doug Lennett's perspective Doug Lennett's increasing awareness of mortality adds impetus to his need to succeed in the AI field within the next few years, leading to a sense of urgency and driving him towards commercialization and widespread application of psych."]}], 'highlights': ['Doug Lennett created Psyche, a 40-year project to acquire common sense knowledge for AI.', 'Acquiring common sense knowledge contributes to achieving general superintelligence and human understanding.', 'The lack of common sense knowledge is the brick wall in AI, impeding true understanding.', 'Leveraging knowledge for split-second decisions is crucial, exemplified by a real-life driving scenario.', 'Challenges of representing knowledge in logical form and the quantity of statements required are discussed.', 'The US government funding and initial fifth generation effort allowed for the acquisition of funding for a couple of person millennia of time, which has been spent since 1984 in developing a common sense knowledge base.', "Emphasizes the need for context to be first-class objects in the system's knowledge base, enabling reasoning and hierarchical arrangement of contexts.", 'The difficulty in estimating the knowledge required for creating super intelligent systems is highlighted, with the speaker indicating the requirement of tens of millions of knowledge, showcasing a shift from an earlier estimate of one or two million.', 'The need for smarter individuals and AI-augmented human species to overcome global problems involving poverty, starvation, global warming, and overcrowding is emphasized, highlighting the potential of genetic engineering and general AIs to enhance intelligence.', 'The development of artificial intelligence can lead to a collective increase in human intelligence, enabling parallel problem-solving and reduced misunderstandings.', 'The system can use everything it already knows and context to generate plausible alternatives, significantly reducing the mental effort of human teachers.', 'AI is expected to empower individuals by enhancing critical thinking skills, pointing out logical fallacies, and arming people with the ability to make reasoned decisions, potentially addressing the shortcomings of the current education system.', 'The potential of self-supervised learning methods and language models like GPT-3 and OpenAI Codex in automating knowledge acquisition and growth of psych through harnessing machine learning and NLU.', 'The importance of priming the pump to expand the knowledge base, emphasizing the need for a large enough manually produced core to facilitate easier learning of new things.', 'The synergy between right brain machine learning and left brain psych-like AI is emphasized for general AI, with a focus on a small percentage of correlations with explanations and independent predictions.', 'The importance of accurate understanding in critical scenarios such as medical diagnosis and driving safety is highlighted.', 'The significance of knowledge-based systems in providing step-by-step explanations for diagnoses and therapy recommendations is discussed.', 'The limitations of representing complex relationships using the semantic web or similar graph structures are discussed, emphasizing the need for more comprehensive representations beyond simple graph structures.', 'The limitations of current language understanding models and the importance of understanding and explaining the reasoning behind answers.', 'The limitations of current search engines and the need for more expressive logical languages.', 'The evolution from semantic web representation to more general logical languages.', 'Importance of representing complex concepts and problem-solving processes.', 'The potential for AI-powered tools to interpret and understand web content.', 'The benefits of enabling systems to reason with web content for improved user experience and SEO.', 'The potential role of AGI in identifying blind spots and proposing novel solutions to global problems, drawing parallels to historical scientific discoveries that resulted from considering non-obvious answers to questions.', 'The resistance to paradigm shifts in science and the potential for AGIs to accelerate such shifts.', 'The advice on making a difference in life and career choices, emphasizing the need for courage and long-term commitment.', "Doug Lennett's urgency to make an impact in the AI field within the next few years.", 'Desire to be remembered as a pioneer of ubiquitous AI and the creator of a vast knowledge acquisition system.', "Impact of mortality on Doug Lennett's perspective."]}