title
NOAM CHOMSKY - THE GHOST IN THE MACHINE

description
Patreon: https://www.patreon.com/mlst Discord: https://discord.gg/ESrGqhf5CB In this special edition episode, we're elated to unveil the Professor Noam Chomsky video! In this select episode, we had the opportunity to converse with one of the most significant thinkers of our generation. We proffered a variety of topics to Professor Chomsky, encompassing his sentiments on the advancement of linguistics and cognitive science, to the convoluted mysteries of science and philosophy. This was a priceless chance to procure perspicacity from one of the most preeminent intellectuals of the 20th century, and we're appreciative to have had the opening to talk with him. We have crafted a comprehensive opening section in which we explore Yann LeCun's recent stance on AGI, a new paper on emergence in LLMs, empiricism pertaining to cognitive science, cognitive templates, "the ghost in the machine" and language. Panel: Dr. Tim Scarfe Dr. Keith Duggar Dr. Walid Saba Pod version: https://anchor.fm/machinelearningstreettalk/episodes/MLST-78---Prof--NOAM-CHOMSKY-Special-Edition-e1l0760 Transcript of Chomsky interview; https://whimsical.com/chomsky-transcript-WgFJLguL7JhzyNhsdgwATy Original corrupt recording: https://share.descript.com/view/N9KNaZTav27 00:00:00 Kick off 00:02:24 C1: LeCun's recent position paper on AI, JEPA, Schmidhuber, EBMs 00:48:38 C2: Emergent abilities in LLMs paper 00:51:32 C3: Empiricism 01:25:33 C4: Cognitive Templates 01:35:47 C5: The Ghost in the Machine 02:00:08 C6: Connectionism and Cognitive Architecture: A Critical Analysis by Fodor and Pylyshyn 02:20:12 C7: We deep-faked Chomsky 02:29:58 C8: Language 02:34:34 C9: Chomsky interview kick-off! 02:35:32 Q1: Large Language Models such as GPT-3 02:39:07 Q2: Connectionism and radical empiricism 02:44:37 Q3: Hybrid systems such as neurosymbolic 02:48:40 Q4: Computationalism silicon vs biological 02:53:21 Q5: Limits of human understanding 03:00:39 Q6: Semantics state-of-the-art 03:06:36 Q7: Universal grammar, I-Language, and language of thought 03:16:20 Q8: Profound and enduring misunderstandings 03:25:34 Q9: Greatest remaining mysteries science and philosophy 03:33:04 Debrief and 'Chuckles' from Chomsky References; (Currently incomplete, we will add to this) LeCun Path to Autonomous AI paper https://openreview.net/forum?id=BZ5a1r-kVsf Tim’s marked up version: https://acrobat.adobe.com/link/review?uri=urn:aaid:scds:US:8c5260f5-8959-3f11-bb3b-befb3bc65f13 Emergent Abilities of Large Language Models [Wei et al] 2022 https://arxiv.org/abs/2206.07682 Connectionism and Cognitive Architecture: A Critical Analysis [Fodor, Pylyshyn] 1988 http://ruccs.rutgers.edu/images/personal-zenon-pylyshyn/docs/jaf.pdf Ghost in the machine https://psychology.fandom.com/wiki/Ghost_in_the_machine https://forum.wordreference.com/threads/in-an-aristotelian-sense.3350478/ https://news.ycombinator.com/item?id=26448901 (thanks to user tikwidd for your analysis) Noam Chomsky in Greece: Philosophies of Democracy (1994) [Language chapter] https://www.youtube.com/watch?v=-kL0UNWcWFc Richard Feynman clip https://vimeo.com/340695809 Chomsky Bryan Magee BBC interview: https://www.youtube.com/watch?v=PxvSQnmcYLo Randy Gallistel's work (question 3) Helmholtz “NNs : they’ve damn slow” Purkinje cells Barbara Partee https://www.youtube.com/watch?v=aA_T20HAzyY Iris Berent https://cos.northeastern.edu/people/iris-berent/ Penrose Orch OR https://en.wikipedia.org/wiki/Orchestrated_objective_reduction https://en.wikipedia.org/wiki/Shadows_of_the_Mind Alan Turing “Systems of Logic Based on Ordinals” https://rauterberg.employee.id.tue.nl/lecturenotes/DDM110%20CAS/Turing/Turing-1939%20Sysyems%20of%20logic%20based%20on%20ordinals.pdf https://www.amazon.com/Events-Semantic-Architecture-Paul-Pietroski/dp/0199244316https://www.amazon.com/Conjunction-Reduction-Redux-MIT-Press/dp/0262035634 Fodor “The Language of Thought” https://www.amazon.com/Language-Thought/dp/0674510305 Least Effort http://materias.df.uba.ar/dnla2019c1/files/2019/03/scaling_in_language.pdf structure dependence in grammar formation https://www.jstor.org/stable/415004 https://www.amazon.com/Minimalist-Program-MIT-Press/dp/0262527340 three models https://chomsky.info/wp-content/uploads/195609-.pdf https://en.wikipedia.org/wiki/Transformational_grammar https://www.amazon.com/Aspects-Theory-Syntax-Noam-Chomsky/dp/0262530074 Darwin's problem https://chomsky.info/20140826/ Descartes's problem https://en.wikipedia.org/wiki/Mind%E2%80%93body_problem Control Theory https://en.wikipedia.org/wiki/Control_(linguistics) Thanks; Thanks to Pikachu for helping us get the Chomsky NVidia Tacotron 2 model working 😍.

detail
{'title': 'NOAM CHOMSKY - THE GHOST IN THE MACHINE', 'heatmap': [{'end': 11719.318, 'start': 11581.299, 'weight': 0.706}], 'summary': "Titled 'noam chomsky - the ghost in the machine' features interviews with noam chomsky at lake como, italy, discussing ai and cognition, lacoon's modern approach, challenges in bayesian logic, adapting to new rules, empiricism, beauty in science, inertia, language alignment, evolving science, natural language semantics, and understanding philosophy and science, offering diverse perspectives and thought-provoking questions on these topics.", 'chapters': [{'end': 92.353, 'segs': [{'end': 50.308, 'src': 'embed', 'start': 26.655, 'weight': 0, 'content': [{'end': 34.299, 'text': "we've been through quite a tumultuous journey to to create this episode, for reasons which will become abundantly clear pretty soon,", 'start': 26.655, 'duration': 7.644}, {'end': 37.121, 'text': 'but noam chomsky is an intellectual heavyweight.', 'start': 34.299, 'duration': 2.822}, {'end': 44.805, 'text': "he's one of my personal heroes and it truly is a dream come true for us to have this experience of interviewing him.", 'start': 37.121, 'duration': 7.684}, {'end': 47.046, 'text': 'so anyway, i really hope you enjoy the show today.', 'start': 44.805, 'duration': 2.241}, {'end': 50.308, 'text': 'we are coming live from lake como in italy.', 'start': 47.046, 'duration': 3.262}], 'summary': 'Interviewing noam chomsky, a dream come true for us.', 'duration': 23.653, 'max_score': 26.655, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q26655.jpg'}, {'end': 100.604, 'src': 'embed', 'start': 72.52, 'weight': 1, 'content': [{'end': 76.622, 'text': 'Now, psychologically, if nothing else, it helps us so much having your support.', 'start': 72.52, 'duration': 4.102}, {'end': 79.444, 'text': "And we're also interested in finding sponsors for the show.", 'start': 77.143, 'duration': 2.301}, {'end': 83.326, 'text': "And of course, don't forget to join our amazing Discord community.", 'start': 79.524, 'duration': 3.802}, {'end': 87.108, 'text': 'But no, we really appreciate our community and our support.', 'start': 83.766, 'duration': 3.342}, {'end': 88.209, 'text': 'And thank you so much.', 'start': 87.488, 'duration': 0.721}, {'end': 89.79, 'text': 'Welcome back to Street Talk.', 'start': 88.849, 'duration': 0.941}, {'end': 92.353, 'text': 'Today is absolutely incredible.', 'start': 89.991, 'duration': 2.362}, {'end': 97.48, 'text': 'Our dream came true and we got to interview our hero, Professor Noam Chomsky.', 'start': 92.594, 'duration': 4.886}, {'end': 100.604, 'text': "Waleed couldn't even keep his shit together.", 'start': 98.161, 'duration': 2.443}], 'summary': 'Show seeks sponsors, values community support, interviews professor noam chomsky.', 'duration': 28.084, 'max_score': 72.52, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q72520.jpg'}], 'start': 2.519, 'title': 'Interviewing noam chomsky', 'summary': 'Delves into the emotional significance of interviewing noam chomsky at lake como, italy, and emphasizes the role of community support and sponsors for the show.', 'chapters': [{'end': 92.353, 'start': 2.519, 'title': 'Noam chomsky interview at lake como', 'summary': 'Highlights the emotional significance of interviewing noam chomsky at lake como, italy, and the importance of community support and sponsors for the show.', 'duration': 89.834, 'highlights': ['The emotional significance of interviewing Noam Chomsky at Lake Como in Italy is highlighted as a dream come true, adding a personal touch to the episode.', 'The importance of community support and sponsors for the show is emphasized, indicating the psychological and practical benefits of such support.']}], 'duration': 89.834, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q2519.jpg', 'highlights': ['The emotional significance of interviewing Noam Chomsky at Lake Como in Italy is highlighted as a dream come true, adding a personal touch to the episode.', 'The importance of community support and sponsors for the show is emphasized, indicating the psychological and practical benefits of such support.']}, {'end': 1904.668, 'segs': [{'end': 120.517, 'src': 'embed', 'start': 92.594, 'weight': 0, 'content': [{'end': 97.48, 'text': 'Our dream came true and we got to interview our hero, Professor Noam Chomsky.', 'start': 92.594, 'duration': 4.886}, {'end': 100.604, 'text': "Waleed couldn't even keep his shit together.", 'start': 98.161, 'duration': 2.443}, {'end': 104.308, 'text': 'He had a jizz face for pretty much the entire interview.', 'start': 100.684, 'duration': 3.624}, {'end': 106.59, 'text': 'Chomsky is an intellectual heavyweight.', 'start': 104.969, 'duration': 1.621}, {'end': 109.872, 'text': "He's probably the most important intellectual of the 20th century.", 'start': 106.61, 'duration': 3.262}, {'end': 112.193, 'text': "He's been cited so many times.", 'start': 110.512, 'duration': 1.681}, {'end': 120.517, 'text': "I mean, we're quite spoilt here on MLST because Friston and Bengio also are in the top 100 of scientists based on their age score.", 'start': 112.213, 'duration': 8.304}], 'summary': 'Interviewed noam chomsky, an intellectual heavyweight, cited many times.', 'duration': 27.923, 'max_score': 92.594, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q92594.jpg'}, {'end': 172.635, 'src': 'embed', 'start': 135.868, 'weight': 1, 'content': [{'end': 142.934, 'text': "But yeah, he's been a towering figure, I think, in so many people's lives throughout the entire 20th century and beyond.", 'start': 135.868, 'duration': 7.066}, {'end': 145.953, 'text': 'This is going to be a long show.', 'start': 144.892, 'duration': 1.061}, {'end': 147.554, 'text': "It's going to be over three and a half hours long.", 'start': 146.033, 'duration': 1.521}, {'end': 149.376, 'text': 'So use the table of contents to skip around.', 'start': 147.594, 'duration': 1.782}, {'end': 152.498, 'text': 'This first chapter is mostly about Yann LeCun.', 'start': 149.856, 'duration': 2.642}, {'end': 160.705, 'text': "And the reason we wanted to talk about him and his recent paper is because as a radical empiricist, he's the antithesis of Chomsky.", 'start': 152.839, 'duration': 7.866}, {'end': 164.408, 'text': 'And I think it frames some of the subsequent discussion nicely.', 'start': 161.025, 'duration': 3.383}, {'end': 170.353, 'text': "But if you want to skip ahead and, you know, just only do Chomsky stuff, then it's about 15 minutes to skip.", 'start': 164.768, 'duration': 5.585}, {'end': 171.474, 'text': 'Look at the table of contents.', 'start': 170.473, 'duration': 1.001}, {'end': 172.635, 'text': 'So chapter one.', 'start': 171.834, 'duration': 0.801}], 'summary': 'The show is over 3.5 hours long, with first chapter mainly about yann lecun, taking approximately 15 minutes to skip.', 'duration': 36.767, 'max_score': 135.868, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q135868.jpg'}, {'end': 224.387, 'src': 'embed', 'start': 197.06, 'weight': 3, 'content': [{'end': 204.491, 'text': "I've always thought that it would be very difficult to just design from scratch an intelligent system.", 'start': 197.06, 'duration': 7.431}, {'end': 210.463, 'text': 'An intelligent system has to basically design itself more or less through learning.', 'start': 205.172, 'duration': 5.291}, {'end': 217.442, 'text': 'So, Lacoon just released a position paper called A Path Towards Autonomous Machine Intelligence.', 'start': 211.317, 'duration': 6.125}, {'end': 219.463, 'text': 'Now, I admire Lacoon a lot.', 'start': 218.042, 'duration': 1.421}, {'end': 224.387, 'text': 'I want to give him credit personally for explaining things in very simple terms.', 'start': 220.324, 'duration': 4.063}], 'summary': 'Lacoon released a paper on autonomous machine intelligence, simplifying complex concepts.', 'duration': 27.327, 'max_score': 197.06, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q197060.jpg'}, {'end': 845.946, 'src': 'embed', 'start': 755.602, 'weight': 5, 'content': [{'end': 762.646, 'text': 'We have another huge part of cognition that has nothing to do with perception.', 'start': 755.602, 'duration': 7.044}, {'end': 774.126, 'text': 'Okay?. The most remarkable quality of human cognition, the very core of our cognition,', 'start': 763.787, 'duration': 10.339}, {'end': 780.748, 'text': 'is the ability to take any two objects and select from an infinite set of possible abstractions.', 'start': 774.126, 'duration': 6.622}, {'end': 784.41, 'text': 'Abstractions which are not deducible from percepts.', 'start': 781.349, 'duration': 3.061}, {'end': 794.08, 'text': 'Why does Lacoon think that the only abstractions which are needed are directly deducible from perceptual information?', 'start': 787.096, 'duration': 6.984}, {'end': 802.785, 'text': 'All of this fails when you consider the fact that perception-derived data cannot deduce most rules about the world,', 'start': 794.94, 'duration': 7.845}, {'end': 805.186, 'text': 'certainly not in limited time and space.', 'start': 802.785, 'duration': 2.401}, {'end': 810.289, 'text': "I'm talking about world models and abstractions and hierarchies.", 'start': 806.227, 'duration': 4.062}, {'end': 813.651, 'text': 'Surely they cannot just be probabilities.', 'start': 811.009, 'duration': 2.642}, {'end': 819.815, 'text': "Now Lacoon's architecture can produce a tiny sliver of abstractions,", 'start': 815.051, 'duration': 4.764}, {'end': 828.061, 'text': 'a minimum-spanning tree of which directly deducible from the hand-crafted priors of the encoders and the prediction models.', 'start': 819.815, 'duration': 8.246}, {'end': 837.468, 'text': 'I agree that these are abstractions, but they are essentially human-crafted or at least human-seeded.', 'start': 829.122, 'duration': 8.346}, {'end': 843.765, 'text': 'the space of possible abstractions between two objects is infinite.', 'start': 839.244, 'duration': 4.521}, {'end': 845.946, 'text': 'Yes, infinite.', 'start': 844.426, 'duration': 1.52}], 'summary': 'Human cognition can create infinite abstractions beyond perception, challenging the notion of perceptual abstractions.', 'duration': 90.344, 'max_score': 755.602, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q755602.jpg'}, {'end': 921.094, 'src': 'embed', 'start': 896.378, 'weight': 8, 'content': [{'end': 904.704, 'text': 'He said that the joint embedding prediction architecture finds a trade-off between the completeness and the predictability of the representations.', 'start': 896.378, 'duration': 8.326}, {'end': 913.991, 'text': 'What is predictable and what does not get represented is determined implicitly by the architectures of the encoders and the predictors.', 'start': 905.425, 'duration': 8.566}, {'end': 921.094, 'text': 'They determine an inductive bias that defines what information is predictable or not.', 'start': 914.972, 'duration': 6.122}], 'summary': 'Joint embedding prediction architecture balances completeness and predictability of representations.', 'duration': 24.716, 'max_score': 896.378, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q896378.jpg'}, {'end': 1047.38, 'src': 'embed', 'start': 1015.191, 'weight': 13, 'content': [{'end': 1018.535, 'text': 'This is a very low resolution view of a complex topic.', 'start': 1015.191, 'duration': 3.344}, {'end': 1023.66, 'text': "It's the equivalent of predicting the weather tomorrow using the average temperature of this month.", 'start': 1018.995, 'duration': 4.665}, {'end': 1029.626, 'text': 'Deep learning folks tend to lean into the parlor tricks and lean away from any mechanistic understanding.', 'start': 1024.241, 'duration': 5.385}, {'end': 1040.194, 'text': 'Lacoon admits there is an exponential blow-up traversing state-action spaces in his hierarchical joint-embedding architecture and suggests using a discrete,', 'start': 1030.387, 'duration': 9.807}, {'end': 1047.38, 'text': 'approximate dynamic programming algorithm like Monte Carlo tree search to find good trajectories in tractable time.', 'start': 1040.194, 'duration': 7.186}], 'summary': 'Deep learning lacks mechanistic understanding, suggests using monte carlo tree search for tractable trajectories.', 'duration': 32.189, 'max_score': 1015.191, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q1015191.jpg'}, {'end': 1078.373, 'src': 'embed', 'start': 1053.524, 'weight': 9, 'content': [{'end': 1062.627, 'text': 'Compositionality is that the meaning of a complex expression is fully determined by its structure and the meanings of its constituents.', 'start': 1053.524, 'duration': 9.103}, {'end': 1069.59, 'text': "Once we fix what the parts mean and how they're put together, we have no more leeway regarding the meaning of the whole.", 'start': 1063.228, 'duration': 6.362}, {'end': 1078.373, 'text': 'This is the principle of compositionality, a fundamental presupposition of most contemporary work in semantics or the study of meaning.', 'start': 1070.15, 'duration': 8.223}], 'summary': 'Compositionality: complex expression meaning determined by structure and constituents.', 'duration': 24.849, 'max_score': 1053.524, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q1053524.jpg'}, {'end': 1233.843, 'src': 'embed', 'start': 1207.587, 'weight': 10, 'content': [{'end': 1212.791, 'text': "And it's been a long, hard slog to get people to realize how important abstraction is.", 'start': 1207.587, 'duration': 5.204}, {'end': 1217.435, 'text': "And I think that there's been a real sea change in the last couple of years.", 'start': 1212.871, 'duration': 4.564}, {'end': 1229.521, 'text': 'The important thing to realize is the only way to represent infinite objects in a finite way is using quantification or logic over typed symbolic structures.', 'start': 1218.095, 'duration': 11.426}, {'end': 1233.843, 'text': "That's why neural networks cannot do basic arithmetic.", 'start': 1230.181, 'duration': 3.662}], 'summary': "Abstraction is crucial; recent shift in understanding; quantification/logic needed for representing infinite objects; neural networks can't do basic arithmetic.", 'duration': 26.256, 'max_score': 1207.587, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q1207587.jpg'}, {'end': 1559.086, 'src': 'embed', 'start': 1535.217, 'weight': 12, 'content': [{'end': 1547.126, 'text': "He said that Jan LeCun's 2022 paper on autonomous machine intelligence rehashes but does not cite essential work of his lab from 1990 to 2015.", 'start': 1535.217, 'duration': 11.909}, {'end': 1559.086, 'text': 'Now, Juergen Schmidhuber has established a bit of a reputation for constantly saying that All of the current ideas that are being published in Deep Learning Today were already done previously in his lab,', 'start': 1547.126, 'duration': 11.96}], 'summary': "Schmidhuber criticizes lecun's paper for not citing his lab's essential work from 1990-2015.", 'duration': 23.869, 'max_score': 1535.217, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q1535217.jpg'}, {'end': 1737.737, 'src': 'embed', 'start': 1718.484, 'weight': 11, 'content': [{'end': 1734.335, 'text': 'He said he released a 2020 position paper which offers a conceptual framework for addressing this problem and provides an in-depth analysis of the challenges and requirements and corresponding inductive biases required for symbolic manipulation to emerge naturally in neural networks.', 'start': 1718.484, 'duration': 15.851}, {'end': 1737.737, 'text': "I must admit I've not read that paper, I'm interested to check it out now.", 'start': 1734.695, 'duration': 3.042}], 'summary': 'A 2020 position paper offers a conceptual framework for addressing challenges in neural networks.', 'duration': 19.253, 'max_score': 1718.484, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q1718484.jpg'}], 'start': 92.594, 'title': 'Ai and cognition', 'summary': 'Discusses the significance of interviewing noam chomsky, the evolution of ai, the limitations of deriving abstractions from perceptual information, the importance of compositionality in semantics and planning, and the dispute between schmidhuber and lecun regarding deep learning limitations, emphasizing key concepts and viewpoints from notable figures in the field.', 'chapters': [{'end': 172.635, 'start': 92.594, 'title': 'Interview with noam chomsky', 'summary': "Discusses the honor of interviewing professor noam chomsky, highlighting his intellectual significance in the 20th century and the anticipation for a long show featuring yann lecun and the contrast with chomsky's ideology.", 'duration': 80.041, 'highlights': ["Interviewing Professor Noam Chomsky, the most important intellectual of the 20th century, is an incredible honor and a significant event in the podcast's history.", "Yann LeCun's radical empiricist ideology is contrasted with Chomsky's, setting the stage for subsequent discussions.", 'The show is anticipated to be over three and a half hours long, with the first chapter focusing on Yann LeCun and providing an option to skip ahead to the Chomsky content.']}, {'end': 726.888, 'start': 173.226, 'title': 'Ai evolution and learning', 'summary': "Discusses the evolution of ai, emphasizing the importance of learning in creating intelligent systems, while highlighting lacoon's position paper on autonomous machine intelligence and his critiques of current ai approaches. the paper delves into the challenges of current ai research, the need for passive learning, efficient learning, and abstractions in ai, and introduces lacoon's perspective on self-supervised learning and contrastive models.", 'duration': 553.662, 'highlights': ['Lacoon emphasizes the importance of learning in creating intelligent systems and critiques current AI approaches, stating that our best systems still fall short of matching human reliability in real-world tasks, despite being fed with large amounts of data and having behaviors hardwired into the model. Importance of learning in creating intelligent systems, critique of current AI approaches, limitations in matching human reliability in real-world tasks despite large amounts of data and hardwired behaviors.', 'Lacoon discusses the challenges of AI research, focusing on the need for passive learning, efficient learning, and introducing type two models, as well as the ability to learn abstractions and compositional semantics, while recognizing the current limitations in achieving compositional semantics with the current paradigm. Challenges of AI research, need for passive learning and efficient learning, introduction of type two models, limitations in achieving compositional semantics with the current paradigm.', 'Lacoon introduces the concept of self-supervised learning and contrastive models, highlighting the use of latent variables of unnormalized energy to represent possible futures, and the regularization of latent variables to prevent information leakage and overcome the curse of dimensionality. Introduction of self-supervised learning and contrastive models, use of latent variables of unnormalized energy, regularization of latent variables to prevent information leakage and overcome the curse of dimensionality.']}, {'end': 1029.626, 'start': 729.589, 'title': 'Cognition and abstraction in ai', 'summary': "Discusses the limitations of deriving abstractions solely from perceptual information, emphasizing the infinite nature of possible abstractions and the essential role of non-perceptual cognition in intelligent behavior, with a critique of yann lecun's approach to reasoning in ai.", 'duration': 300.037, 'highlights': ["The ability to take any two objects and select from an infinite set of possible abstractions is the core of human cognition. Human cognition's remarkable quality lies in its capacity to choose from an infinite array of abstractions, independent of perceptual input.", 'Perception-derived data cannot deduce most rules about the world, including world models and abstractions, which are not limited to probabilities. Perceptual information fails to deduce most rules about the world, including world models and abstractions, which are not confined to probabilities.', "The architecture of Lacoon produces a limited subset of abstractions directly deducible from hand-crafted priors, though the space of possible abstractions is infinite. Lacoon's architecture generates a small subset of abstractions deducible from hand-crafted priors, while the space of possible abstractions remains infinite.", 'The joint embedding prediction architecture determines what information is predictable or not based on inductive biases defined by the encoders and predictors. The joint embedding prediction architecture establishes what information is predictable based on inductive biases defined by the encoders and predictors.', "Yann LeCun's approach to reasoning in AI is critiqued for its low-resolution view of a complex topic, leaning into parlor tricks rather than seeking mechanistic understanding. Yann LeCun's approach to AI reasoning is criticized for its low-resolution perspective, favoring parlor tricks over seeking mechanistic understanding."]}, {'end': 1430.438, 'start': 1030.387, 'title': 'Compositionality in semantics and planning', 'summary': 'Explores the importance of compositionality in semantics and planning, emphasizing the fundamental role of symbolic logic and algebraic approach, and highlights the significance of abstraction and quantification in representing infinite objects, with a focus on the mathematical work of richard montague.', 'duration': 400.051, 'highlights': ['The importance of compositionality in semantics and planning is emphasized, with a focus on the fundamental role of symbolic logic and algebraic approach. Emphasizes the importance of compositionality in semantics and planning; highlights the role of symbolic logic and algebraic approach.', 'The significance of abstraction and quantification in representing infinite objects is highlighted, with a focus on the mathematical work of Richard Montague. Emphasizes the significance of abstraction and quantification in representing infinite objects; highlights the work of Richard Montague.', 'The exponential blow-up traversing state-action spaces in hierarchical joint-embedding architecture is discussed, along with the suggestion of using a discrete, approximate dynamic programming algorithm like Monte Carlo tree search to find good trajectories in tractable time. Discusses the exponential blow-up in state-action spaces and suggests using Monte Carlo tree search for finding good trajectories.']}, {'end': 1904.668, 'start': 1432.479, 'title': "Montague's algebra and deep learning limitations", 'summary': "Discusses the essence of reasoning, the potential limitations of deep learning, and the dispute between juergen schmidhuber and jan lecun regarding the novelty of ideas in lecun's 2022 paper on autonomous machine intelligence, highlighting schmidhuber's resentful claim of lack of recognition and the need to incorporate inductive biases into neural networks for symbolic manipulation.", 'duration': 472.189, 'highlights': ["Juergen Schmidhuber's claim of lack of recognition and the need to incorporate inductive biases into neural networks for symbolic manipulation Schmidhuber asserts that much of the closely related work in LeCun's 2022 paper was done in his lab and he wishes to be acknowledged and recognized, emphasizing the importance of incorporating inductive biases into neural networks to efficiently learn about symbols.", "Dispute between Juergen Schmidhuber and Jan LeCun regarding the novelty of ideas in LeCun's 2022 paper Schmidhuber expresses resentment that he hasn't had proper attribution for the ideas in LeCun's paper and claims that many of the ideas described in the paper read like deja vu of the papers from his lab, going back to 1990 without any citations.", 'Essence of reasoning and potential limitations of deep learning The discussion delves into the definition of reasoning and whether there is an essential qualitative difference between models that merely perform curve fitting and those that establish a causal model of the observed data, questioning the need for fundamentally different things from deep learning, such as discrete symbolic reasoning.']}], 'duration': 1812.074, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q92594.jpg', 'highlights': ["Interviewing Professor Noam Chomsky, the most important intellectual of the 20th century, is an incredible honor and a significant event in the podcast's history.", "Yann LeCun's radical empiricist ideology is contrasted with Chomsky's, setting the stage for subsequent discussions.", 'The show is anticipated to be over three and a half hours long, with the first chapter focusing on Yann LeCun and providing an option to skip ahead to the Chomsky content.', 'Lacoon emphasizes the importance of learning in creating intelligent systems and critiques current AI approaches, stating that our best systems still fall short of matching human reliability in real-world tasks, despite being fed with large amounts of data and having behaviors hardwired into the model.', 'Lacoon discusses the challenges of AI research, focusing on the need for passive learning, efficient learning, and introducing type two models, as well as the ability to learn abstractions and compositional semantics, while recognizing the current limitations in achieving compositional semantics with the current paradigm.', 'The ability to take any two objects and select from an infinite set of possible abstractions is the core of human cognition.', 'Perception-derived data cannot deduce most rules about the world, including world models and abstractions, which are not limited to probabilities.', 'The architecture of Lacoon produces a limited subset of abstractions directly deducible from hand-crafted priors, though the space of possible abstractions is infinite.', 'The joint embedding prediction architecture determines what information is predictable or not based on inductive biases defined by the encoders and predictors.', 'The importance of compositionality in semantics and planning is emphasized, with a focus on the fundamental role of symbolic logic and algebraic approach.', 'The significance of abstraction and quantification in representing infinite objects is highlighted, with a focus on the mathematical work of Richard Montague.', "Juergen Schmidhuber's claim of lack of recognition and the need to incorporate inductive biases into neural networks for symbolic manipulation.", "Dispute between Juergen Schmidhuber and Jan LeCun regarding the novelty of ideas in LeCun's 2022 paper.", 'Essence of reasoning and potential limitations of deep learning.']}, {'end': 2571.952, 'segs': [{'end': 1983.705, 'src': 'embed', 'start': 1923.577, 'weight': 0, 'content': [{'end': 1925.098, 'text': "It's using modern methods.", 'start': 1923.577, 'duration': 1.521}, {'end': 1926.938, 'text': "It's not using RNNs, for example.", 'start': 1925.358, 'duration': 1.58}, {'end': 1929.139, 'text': "Yes, it's using the same abstract ideas.", 'start': 1927.279, 'duration': 1.86}, {'end': 1932.74, 'text': 'In fact, what I like about Lacoon is that he presents all of his work in an abstract way.', 'start': 1929.199, 'duration': 3.541}, {'end': 1936.681, 'text': 'The pictorial formalism of energy-based models is very abstract.', 'start': 1933.24, 'duration': 3.441}, {'end': 1938.022, 'text': "That's what makes it understandable.", 'start': 1936.702, 'duration': 1.32}, {'end': 1942.225, 'text': 'The way he describes contrastive and self-supervised learning is very abstract.', 'start': 1938.522, 'duration': 3.703}, {'end': 1949.932, 'text': "He's talking about predicting unobserved information from observed information, and he uses a language which is very accessible to lots of people.", 'start': 1942.526, 'duration': 7.406}, {'end': 1954.836, 'text': "So I can understand why other researchers would look at it and say, oh, that's basically the same as what I've done.", 'start': 1950.252, 'duration': 4.584}, {'end': 1956.978, 'text': "But that's because he's talking in the abstract.", 'start': 1955.176, 'duration': 1.802}, {'end': 1960.501, 'text': 'If you look at his physical models, they are different, in my opinion.', 'start': 1957.498, 'duration': 3.003}, {'end': 1968.384, 'text': 'Lacoon has been a huge advocate of passive or so-called self-supervised learning.', 'start': 1963.294, 'duration': 5.09}, {'end': 1971.281, 'text': 'supervised learning sucks.', 'start': 1969.58, 'duration': 1.701}, {'end': 1979.984, 'text': "I mean it's very limited in the sense that you can train machines to do very specific tasks and because they're trying to do very specific tasks,", 'start': 1971.781, 'duration': 8.203}, {'end': 1983.705, 'text': "they're going to use all the biases that are in the data to do that task.", 'start': 1979.984, 'duration': 3.721}], 'summary': 'Lacoon presents work in abstract, accessible manner, advocating for self-supervised learning over supervised learning.', 'duration': 60.128, 'max_score': 1923.577, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q1923577.jpg'}, {'end': 2033.01, 'src': 'embed', 'start': 2004.313, 'weight': 1, 'content': [{'end': 2009.156, 'text': "Okay They've always been essentially maximum likelihood estimators.", 'start': 2004.313, 'duration': 4.843}, {'end': 2010.477, 'text': "It's like okay.", 'start': 2009.456, 'duration': 1.021}, {'end': 2016.3, 'text': 'I train my neural network to take in a whole bunch of inputs and to give me the one true value, you know,', 'start': 2010.477, 'duration': 5.823}, {'end': 2021.363, 'text': 'which is really just the maximum kind of likelihood computation of of my inputs.', 'start': 2016.3, 'duration': 5.063}, {'end': 2028.348, 'text': "And, you know, as a Bayesian, it's like, I mean, guys, that's just an estimate, you know, like the real truth is it's a distribution.", 'start': 2021.624, 'duration': 6.724}, {'end': 2033.01, 'text': 'Like there are many possible, you know, output values that this input should have given me.', 'start': 2029.048, 'duration': 3.962}], 'summary': 'Neural networks are maximum likelihood estimators, but bayesian view it as a distribution.', 'duration': 28.697, 'max_score': 2004.313, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q2004313.jpg'}, {'end': 2538.009, 'src': 'embed', 'start': 2510.446, 'weight': 2, 'content': [{'end': 2514.649, 'text': "I was just thinking, you know this idea that, Hey, we'll work on the probabilities.", 'start': 2510.446, 'duration': 4.203}, {'end': 2517.212, 'text': 'We can just ignore the normalization.', 'start': 2514.67, 'duration': 2.542}, {'end': 2525.039, 'text': "Right And just work with the energies is almost like somebody coming along and saying you know, whenever I'm doing arithmetic like adding fractions,", 'start': 2517.272, 'duration': 7.767}, {'end': 2530.544, 'text': "it's such a pain because I have to have the same denominator every time I add the fractions together.", 'start': 2525.039, 'duration': 5.505}, {'end': 2538.009, 'text': 'know, and when the denominators get to be these really big integers, i have to do this massive multi-digit multiplication.', 'start': 2531.264, 'duration': 6.745}], 'summary': 'Discussing the idea of working with energies instead of probabilities in a scenario relating to arithmetic with fractions.', 'duration': 27.563, 'max_score': 2510.446, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q2510446.jpg'}], 'start': 1904.768, 'title': "Lacoon's modern approach and critique", 'summary': "Delves into lacoon's modern approach to abstract models, highlighting architectural differences, accessibility through abstract language, and pictorial formalism, as well as discussing lacoon's critique of deep learning, emphasizing the limitations of supervised learning and the need for embracing uncertainty and probability theory in machine learning.", 'chapters': [{'end': 1960.501, 'start': 1904.768, 'title': "Lacoon's modern approach to abstract models", 'summary': "Discusses lacoon's modern approach to abstract models, emphasizing the differences in architecture and approach compared to existing concepts, and highlighting the accessibility and understandability of his work through abstract language and pictorial formalism.", 'duration': 55.733, 'highlights': ['Lacoon presents his work in an abstract way, using modern methods and pictorial formalism, making it understandable and accessible to a wider audience.', 'His approach involves predicting unobserved information from observed information, using language that is accessible to many people.', "Although conceptually similar to other works, Lacoon's architecture and physical models are different, showcasing the uniqueness of his approach."]}, {'end': 2571.952, 'start': 1963.294, 'title': "Lacoon's critique of deep learning", 'summary': "Discusses lacoon's critique of deep learning, emphasizing the limitations of supervised learning and the need for embracing uncertainty and probability theory in machine learning, with a focus on the transition from maximum likelihood estimation to probability-based modeling.", 'duration': 608.658, 'highlights': ['Lacoon criticizes supervised learning for its limitations, advocating for the need to embrace uncertainty in machine learning. Critique of supervised learning; Embracing uncertainty in machine learning.', 'The chapter emphasizes the importance of transitioning from maximum likelihood estimation to probability-based modeling in machine learning. Transition from maximum likelihood estimation to probability-based modeling.', "The discussion highlights Lacoon's pragmatic approach and his efforts to guide the machine learning community towards embracing theoretical results and addressing challenging problems. Lacoon's pragmatic approach; Guiding the machine learning community.", 'The critique of energy-based models and the comparison to arithmetic with fractions underscores the importance of addressing normalization and embracing fundamental principles in modeling. Critique of energy-based models; Importance of addressing normalization and fundamental principles in modeling.']}], 'duration': 667.184, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q1904768.jpg', 'highlights': ['Lacoon criticizes supervised learning for its limitations, advocating for the need to embrace uncertainty in machine learning.', 'The chapter emphasizes the importance of transitioning from maximum likelihood estimation to probability-based modeling in machine learning.', 'The critique of energy-based models and the comparison to arithmetic with fractions underscores the importance of addressing normalization and embracing fundamental principles in modeling.', "Lacoon's pragmatic approach and his efforts to guide the machine learning community towards embracing theoretical results and addressing challenging problems.", 'Lacoon presents his work in an abstract way, using modern methods and pictorial formalism, making it understandable and accessible to a wider audience.', 'His approach involves predicting unobserved information from observed information, using language that is accessible to many people.', "Although conceptually similar to other works, Lacoon's architecture and physical models are different, showcasing the uniqueness of his approach."]}, {'end': 4195.15, 'segs': [{'end': 2614.143, 'src': 'embed', 'start': 2573.113, 'weight': 0, 'content': [{'end': 2574.994, 'text': 'What do you think Lacoon would say, though, to that??', 'start': 2573.113, 'duration': 1.881}, {'end': 2586.166, 'text': "Well, I think he says, I mean you know, in the paper itself, like there are many scenarios where you can't ignore the normalization.", 'start': 2576.963, 'duration': 9.203}, {'end': 2591.267, 'text': 'And, like, he kind of admits that and uses it in certain places.', 'start': 2586.246, 'duration': 5.021}, {'end': 2592.868, 'text': "But that's just Bayesian logic.", 'start': 2591.667, 'duration': 1.201}, {'end': 2595.489, 'text': 'Bayesian probability at that point.', 'start': 2593.928, 'duration': 1.561}, {'end': 2600.673, 'text': "It's like Bayesians don't have any problem, you know, understanding that there is a normalization constant.", 'start': 2595.509, 'duration': 5.164}, {'end': 2602.234, 'text': 'Sometimes you can work without it.', 'start': 2600.793, 'duration': 1.441}, {'end': 2603.876, 'text': "Sometimes you can't.", 'start': 2602.915, 'duration': 0.961}, {'end': 2607.378, 'text': "And so they'll defer normalization until necessary.", 'start': 2604.076, 'duration': 3.302}, {'end': 2612.822, 'text': "But the problem is there's so many scenarios where normalization is required.", 'start': 2607.418, 'duration': 5.404}, {'end': 2614.143, 'text': 'So many useful scenarios.', 'start': 2612.902, 'duration': 1.241}], 'summary': 'Lacoon acknowledges the need for normalization in bayesian logic, deferring it until necessary, but recognizing its requirement in many useful scenarios.', 'duration': 41.03, 'max_score': 2573.113, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q2573113.jpg'}, {'end': 2730.206, 'src': 'embed', 'start': 2700.399, 'weight': 3, 'content': [{'end': 2703.601, 'text': 'Well, your world model is going to need to be able to model those things.', 'start': 2700.399, 'duration': 3.202}, {'end': 2705.962, 'text': "And I'm really skeptical.", 'start': 2704.641, 'duration': 1.321}, {'end': 2716.831, 'text': "that I mean if we're trying to model symbolic systems, Are we really hopeful that we can model them well enough with non-symbolic,", 'start': 2705.962, 'duration': 10.869}, {'end': 2718.032, 'text': 'differentiable systems?', 'start': 2716.831, 'duration': 1.201}, {'end': 2719.975, 'text': "You know, I'm pretty skeptical.", 'start': 2718.353, 'duration': 1.622}, {'end': 2726.362, 'text': "I think we have to just grab the bull by the horns, as they would say, just embrace the fact that we've got to figure out.", 'start': 2720.035, 'duration': 6.327}, {'end': 2730.206, 'text': 'how to do search over these discrete spaces.', 'start': 2727.003, 'duration': 3.203}], 'summary': 'Skeptical about modeling symbolic systems with non-symbolic, differentiable systems.', 'duration': 29.807, 'max_score': 2700.399, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q2700399.jpg'}, {'end': 2761.033, 'src': 'embed', 'start': 2738.372, 'weight': 4, 'content': [{'end': 2748.219, 'text': "if we can crack the nut of learning how to do searches over the space of all possible programs outside of the space, that's differential,", 'start': 2738.372, 'duration': 9.847}, {'end': 2750.461, 'text': 'differentially accessible.', 'start': 2748.219, 'duration': 2.242}, {'end': 2751.202, 'text': "that's you know.", 'start': 2750.461, 'duration': 0.741}, {'end': 2754.204, 'text': "we're going to make, like huge project there, progress there right?", 'start': 2751.202, 'duration': 3.002}, {'end': 2758.068, 'text': "Whether it's DreamCoder or something else, You know, I don't know neat.", 'start': 2754.244, 'duration': 3.824}, {'end': 2761.033, 'text': 'you know, neuroevolutionary topologies, whatever it is.', 'start': 2758.068, 'duration': 2.965}], 'summary': 'Cracking the nut of learning to search over all possible programs will lead to significant progress in projects like dreamcoder or neuroevolutionary topologies.', 'duration': 22.661, 'max_score': 2738.372, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q2738372.jpg'}, {'end': 2903.335, 'src': 'embed', 'start': 2877.435, 'weight': 5, 'content': [{'end': 2886.04, 'text': "But I think it's very unfair towards Gary Marcus because, unlike open ai that's been spouting all of this utter nonsense,", 'start': 2877.435, 'duration': 8.605}, {'end': 2888.422, 'text': 'gary marcus just has a different perspective.', 'start': 2886.04, 'duration': 2.382}, {'end': 2894.447, 'text': 'there are so many different perspectives in artificial general intelligence and gary marcus has the same perspective as noam chomsky,', 'start': 2888.422, 'duration': 6.025}, {'end': 2898.731, 'text': 'which is that psychology has a lot to say about artificial general intelligence.', 'start': 2894.447, 'duration': 4.284}, {'end': 2903.335, 'text': 'And yes, maybe he has a philosophical agenda, maybe even has a monetary agenda,', 'start': 2899.111, 'duration': 4.224}], 'summary': "Gary marcus offers a different perspective on agi, aligning with noam chomsky's view on psychology's role.", 'duration': 25.9, 'max_score': 2877.435, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q2877435.jpg'}, {'end': 2964.607, 'src': 'embed', 'start': 2941.351, 'weight': 8, 'content': [{'end': 2949.196, 'text': 'This paper, they say, discusses the unpredictable phenomena which they refer to as the emergent ability of large language models.', 'start': 2941.351, 'duration': 7.845}, {'end': 2955.801, 'text': "And they consider an ability to be emergent if it's not present in smaller models, but is present in larger models.", 'start': 2949.597, 'duration': 6.204}, {'end': 2961.325, 'text': 'Thus, emergent abilities cannot be predicted simply by extrapolating the performance of smaller models.', 'start': 2956.161, 'duration': 5.164}, {'end': 2964.607, 'text': "I don't really like their definition.", 'start': 2963.486, 'duration': 1.121}], 'summary': 'Paper discusses emergent abilities in large language models.', 'duration': 23.256, 'max_score': 2941.351, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q2941351.jpg'}, {'end': 3129.118, 'src': 'embed', 'start': 3083.121, 'weight': 6, 'content': [{'end': 3088.484, 'text': "then suddenly we'll get this snap and we'll get this recursive, self-improving, artificial general intelligence.", 'start': 3083.121, 'duration': 5.363}, {'end': 3090.484, 'text': "That just doesn't make any sense to me.", 'start': 3089.084, 'duration': 1.4}, {'end': 3091.745, 'text': "I genuinely don't believe that.", 'start': 3090.564, 'duration': 1.181}, {'end': 3103.125, 'text': 'Chapter 3 On Empiricism, Chomsky argued that the way we actually acquire the faculty of language and therefore its relationship to experience,', 'start': 3092.776, 'duration': 10.349}, {'end': 3107.948, 'text': 'and indeed the physical world, are radically different from the empiricist tradition.', 'start': 3103.125, 'duration': 4.823}, {'end': 3114.091, 'text': 'Chomsky became well known for his famous 1950s critique of behavioral psychology.', 'start': 3108.729, 'duration': 5.362}, {'end': 3122.635, 'text': 'Now, behavioral psychologists tended to believe that humans were an unadulterated blank slate made of putty,', 'start': 3114.692, 'duration': 7.943}, {'end': 3129.118, 'text': 'which was then molded and shaped by our environment, you know, through a process of empirical stimulus and response.', 'start': 3122.635, 'duration': 6.483}], 'summary': 'Chomsky critiques behavior psychology, arguing for language acquisition as distinct from empiricism.', 'duration': 45.997, 'max_score': 3083.121, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q3083121.jpg'}, {'end': 3757.258, 'src': 'embed', 'start': 3720.385, 'weight': 10, 'content': [{'end': 3723.707, 'text': 'he he admits that there is prior knowledge.', 'start': 3720.385, 'duration': 3.322}, {'end': 3727.629, 'text': 'You know, you need these inductive priors to be quite useful.', 'start': 3723.907, 'duration': 3.722}, {'end': 3736.192, 'text': 'After all, one of his huge achievements was CNN, i.e. structurally encoding a certain kind of, of prior, this translation and variance.', 'start': 3727.929, 'duration': 8.263}, {'end': 3739.553, 'text': 'into neural networks.', 'start': 3737.933, 'duration': 1.62}, {'end': 3743.074, 'text': 'but he thinks it should be as close to zero as possible,', 'start': 3739.553, 'duration': 3.521}, {'end': 3757.258, 'text': "that it should be so minimal that it's like just enough to kind of jump start or bootstrap a learning system and then from then on it can kind of just learn everything on its own through observation of of data.", 'start': 3743.074, 'duration': 14.184}], 'summary': 'Inductive priors are crucial for cnn, but should be minimal for independent learning.', 'duration': 36.873, 'max_score': 3720.385, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q3720385.jpg'}, {'end': 3788.651, 'src': 'embed', 'start': 3763.385, 'weight': 11, 'content': [{'end': 3775.299, 'text': "and I think Wally too is that we believe that there's just not enough data, not enough data, not enough computational resources, not enough time.", 'start': 3763.385, 'duration': 11.914}, {'end': 3785.068, 'text': 'uh, for that to make sense, that sure, there are a lot of concepts, a lot of knowledge that you can learn by by observation, and then,', 'start': 3776.68, 'duration': 8.388}, {'end': 3788.651, 'text': 'and then you know, reasoning, kind of on top of that.', 'start': 3785.068, 'duration': 3.583}], 'summary': 'Insufficient data, computational resources, and time hinder sense-making and learning processes.', 'duration': 25.266, 'max_score': 3763.385, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q3763385.jpg'}, {'end': 4128.709, 'src': 'embed', 'start': 4103.505, 'weight': 12, 'content': [{'end': 4112.848, 'text': 'So I could quickly come up with a cognitive template in my mind that in this computer world you can have two blue objects simultaneously existing in two different rooms.', 'start': 4103.505, 'duration': 9.343}, {'end': 4116.589, 'text': 'And I would very quickly learn that cognitive template and I could reason over it.', 'start': 4113.167, 'duration': 3.422}, {'end': 4120.25, 'text': 'So I have just empirically come up with a new cognitive template.', 'start': 4116.868, 'duration': 3.382}, {'end': 4128.709, 'text': "I mean, I agree because you don't even have to imagine simulations for very strange.", 'start': 4121.787, 'duration': 6.922}], 'summary': 'New cognitive template formed for reasoning over simultaneous existence of two blue objects.', 'duration': 25.204, 'max_score': 4103.505, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q4103505.jpg'}], 'start': 2573.113, 'title': 'Challenges in bayesian logic, modeling symbolic systems, and large language models', 'summary': 'Explores the concept of normalization in bayesian logic, challenges in modeling symbolic systems with differentiable learning methods, and emergent abilities of large language models, emphasizing the need for normalization in bayesian scenarios, skepticism towards modeling symbolic systems with non-symbolic, differentiable systems, and the limitations and implications of knowledge acquisition.', 'chapters': [{'end': 2614.143, 'start': 2573.113, 'title': 'Bayesian logic and normalization', 'summary': 'Discusses the concept of normalization in bayesian logic, highlighting that while bayesians can work without it in some scenarios, there are many useful scenarios where normalization is required.', 'duration': 41.03, 'highlights': ['Bayesians understand the presence of a normalization constant and can work without it in certain scenarios, but there are many useful scenarios where normalization is required.', 'Lacoon acknowledges the necessity of normalization in various scenarios and utilizes it in certain places within the paper.', 'The presence of normalization is an integral aspect of Bayesian probability and is acknowledged by practitioners.']}, {'end': 2919.632, 'start': 2614.263, 'title': 'Challenges in modeling symbolic systems', 'summary': 'Discusses the challenges of using differentiable learning methods to model symbolic computational systems, highlighting the skepticism towards modeling symbolic systems with non-symbolic, differentiable systems and the need to put more effort into learning how to search the space of all possible programs outside of the differentially accessible space.', 'duration': 305.369, 'highlights': ['The challenge of modeling symbolic systems with non-symbolic, differentiable systems The speaker expresses skepticism about the ability of non-symbolic, differentiable systems to effectively model symbolic computational systems, emphasizing the need to embrace the challenge of searching over discrete spaces outside of the differentially accessible space.', 'The need to focus on learning how to search the space of all possible programs outside of the differentially accessible space The chapter emphasizes the necessity of dedicating more effort to learning how to conduct searches over the space of all potential programs that are not differentially accessible, highlighting the potential for significant progress in this area.', 'Critique of AI hype and different perspectives on artificial general intelligence The speaker criticizes the prevalent hype in the AI community and discusses the various perspectives on artificial general intelligence, highlighting the unfairness in criticizing certain perspectives and the need for diverse viewpoints in the field.']}, {'end': 3589.23, 'start': 2919.792, 'title': 'Emergent abilities of large language models', 'summary': "Discusses the emergent abilities of large language models, challenging the concept of emergence and extrapolation, and delves into chomsky's critique of behavioral psychology and the implications of empiricism, highlighting the limitations and implications of knowledge acquisition.", 'duration': 669.438, 'highlights': ["Chomsky's critique of behavioral psychology and the implications of empiricism, highlighting the limitations and implications of knowledge acquisition.", 'The discussion on emergent abilities challenges the concept of emergence and extrapolation, presenting a different perspective on transient changes in phenomena.', 'The paper examines the unpredictable phenomena referred to as the emergent ability of large language models and the relationship between small and large models.', 'The debate on emergent abilities and extrapolation in language models, raising questions about the sudden improvement in perplexity and its implications.', 'The concept of emergence and its significance, challenging the idea of recursive, self-improving artificial general intelligence.']}, {'end': 4195.15, 'start': 3589.91, 'title': 'Empiricism vs rationalism in knowledge acquisition', 'summary': 'Discusses the debate between empiricists and rationalists in knowledge acquisition, highlighting the degree of innate knowledge, the role of neural networks in learning, and the existence of cognitive templates.', 'duration': 605.24, 'highlights': ["Neural networks and the role of prior knowledge The debate focuses on the role of prior knowledge in neural networks, with LeCun advocating for minimal prior knowledge and emphasizing the system's ability to learn through empirical observation.", 'Debate on the sufficiency of empirical learning The discussion involves the contention that empirical learning alone may not provide sufficient data, computational resources, or time to acquire certain types of knowledge, leading to skepticism about the pragmatic feasibility of solely relying on empirical learning.', 'Existence of cognitive templates and innate knowledge The conversation delves into the existence of cognitive templates, with considerations about the innate concepts and the potential for individuals to empirically develop new cognitive templates based on alternative realities.']}], 'duration': 1622.037, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q2573113.jpg', 'highlights': ['The presence of normalization is an integral aspect of Bayesian probability and is acknowledged by practitioners.', 'Bayesians understand the presence of a normalization constant and can work without it in certain scenarios, but there are many useful scenarios where normalization is required.', 'Lacoon acknowledges the necessity of normalization in various scenarios and utilizes it in certain places within the paper.', 'The challenge of modeling symbolic systems with non-symbolic, differentiable systems The speaker expresses skepticism about the ability of non-symbolic, differentiable systems to effectively model symbolic computational systems, emphasizing the need to embrace the challenge of searching over discrete spaces outside of the differentially accessible space.', 'The need to focus on learning how to search the space of all possible programs outside of the differentially accessible space The chapter emphasizes the necessity of dedicating more effort to learning how to conduct searches over the space of all potential programs that are not differentially accessible, highlighting the potential for significant progress in this area.', 'Critique of AI hype and different perspectives on artificial general intelligence The speaker criticizes the prevalent hype in the AI community and discusses the various perspectives on artificial general intelligence, highlighting the unfairness in criticizing certain perspectives and the need for diverse viewpoints in the field.', "Chomsky's critique of behavioral psychology and the implications of empiricism, highlighting the limitations and implications of knowledge acquisition.", 'The concept of emergence and its significance, challenging the idea of recursive, self-improving artificial general intelligence.', 'The paper examines the unpredictable phenomena referred to as the emergent ability of large language models and the relationship between small and large models.', 'The debate on emergent abilities and extrapolation in language models, raising questions about the sudden improvement in perplexity and its implications.', "Neural networks and the role of prior knowledge The debate focuses on the role of prior knowledge in neural networks, with LeCun advocating for minimal prior knowledge and emphasizing the system's ability to learn through empirical observation.", 'Debate on the sufficiency of empirical learning The discussion involves the contention that empirical learning alone may not provide sufficient data, computational resources, or time to acquire certain types of knowledge, leading to skepticism about the pragmatic feasibility of solely relying on empirical learning.', 'Existence of cognitive templates and innate knowledge The conversation delves into the existence of cognitive templates, with considerations about the innate concepts and the potential for individuals to empirically develop new cognitive templates based on alternative realities.']}, {'end': 4924.064, 'segs': [{'end': 4237.449, 'src': 'embed', 'start': 4195.25, 'weight': 0, 'content': [{'end': 4203.175, 'text': 'So, I mean, you get all kinds of almost antithetical or opposite, you know, behaviors of things that happen all throughout the natural world.', 'start': 4195.25, 'duration': 7.925}, {'end': 4209.486, 'text': "And I don't see people as having any trouble really learning like a new set of rules, if you will.", 'start': 4203.275, 'duration': 6.211}, {'end': 4216.552, 'text': 'I mean I play a lot of video games, or, or used to when I back before we started doing the show and I had time to play video games.', 'start': 4209.546, 'duration': 7.006}, {'end': 4221.736, 'text': 'but you know, for example, you run into a video game where, like it, has a time travel element.', 'start': 4216.552, 'duration': 5.184}, {'end': 4225.498, 'text': "It doesn't take long to learn how to work in this area,", 'start': 4221.896, 'duration': 3.602}, {'end': 4232.083, 'text': 'where you can move in a dimension that corresponds to to time or you learn kind of different sets of of world rules.', 'start': 4225.498, 'duration': 6.585}, {'end': 4234.685, 'text': "Like people somehow don't really have.", 'start': 4232.123, 'duration': 2.562}, {'end': 4237.449, 'text': 'trouble necessarily,', 'start': 4235.666, 'duration': 1.783}], 'summary': 'Observing diverse behaviors in natural world, easy to learn new rules, like adapting to video game scenarios.', 'duration': 42.199, 'max_score': 4195.25, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q4195250.jpg'}, {'end': 4378.241, 'src': 'embed', 'start': 4348.118, 'weight': 3, 'content': [{'end': 4352.88, 'text': 'And I think what was interesting talking to Chomsky was You know,', 'start': 4348.118, 'duration': 4.762}, {'end': 4358.665, 'text': 'when we asked is it possible that there are really just these limits to human cognition??', 'start': 4352.88, 'duration': 5.785}, {'end': 4361.748, 'text': 'You know this horizon beyond which we may never go.', 'start': 4358.845, 'duration': 2.903}, {'end': 4366.231, 'text': 'And he brought up the example of rats and prime number mazes.', 'start': 4361.828, 'duration': 4.403}, {'end': 4373.317, 'text': 'You know, that if you have a prime number maze, so this is a maze where at every prime intersection, you take a right, for example.', 'start': 4366.472, 'duration': 6.845}, {'end': 4378.241, 'text': 'And if you do that, then you get to the cheese or you escape the maze or whatever the goal is.', 'start': 4373.357, 'duration': 4.884}], 'summary': 'Chomsky discusses limits to human cognition using rats and prime number mazes.', 'duration': 30.123, 'max_score': 4348.118, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q4348118.jpg'}, {'end': 4815.579, 'src': 'embed', 'start': 4786.84, 'weight': 2, 'content': [{'end': 4789.722, 'text': 'But then of course you can prove that any true statement is true.', 'start': 4786.84, 'duration': 2.882}, {'end': 4792.624, 'text': 'So you got to kind of pick your poison here.', 'start': 4790.583, 'duration': 2.041}, {'end': 4800.469, 'text': "You're either incomplete or inconsistent, but it's important to keep in mind that this applies to closed formal systems.", 'start': 4792.664, 'duration': 7.805}, {'end': 4807.394, 'text': 'And Technically and mathematically, you know, human beings and human cognition are not closed.', 'start': 4800.709, 'duration': 6.685}, {'end': 4811.236, 'text': 'Like this is in fact what science does, right? It poses a question.', 'start': 4807.734, 'duration': 3.502}, {'end': 4815.579, 'text': "Okay I don't know if this, if the statement is true or false, I don't know.", 'start': 4811.477, 'duration': 4.102}], 'summary': 'In closed formal systems, one must choose between incompleteness or inconsistency, but human cognition is not closed.', 'duration': 28.739, 'max_score': 4786.84, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q4786840.jpg'}], 'start': 4195.25, 'title': 'Adapting to new rules and limits of human cognition', 'summary': "Discusses adapting to new rules with examples from video games and explores the limits of human cognition using the illustration of rats and prime number mazes, and the implications of gödel's proof on human cognition.", 'chapters': [{'end': 4237.449, 'start': 4195.25, 'title': 'Adapting to new rules', 'summary': 'Discusses how people can easily adapt to learning new rules, citing the example of learning different sets of rules in video games, such as time travel elements.', 'duration': 42.199, 'highlights': ['People can easily learn new sets of rules, as seen in video games with time travel elements.', 'The natural world exhibits antithetical behaviors that people can adapt to.']}, {'end': 4924.064, 'start': 4237.449, 'title': 'Limits of human cognition', 'summary': "Discusses the limits of human cognition and the possibility that there are concepts necessary to understand the physical universe that may be beyond human cognition, illustrating this with the example of rats and prime number mazes, and explores the implications of gödel's proof on the completeness and consistency of formal systems and its relevance to human cognition.", 'duration': 686.615, 'highlights': ['The chapter discusses the limits of human cognition and the possibility that there are concepts necessary to understand the physical universe that may be beyond human cognition. Discussion on the potential limits of human cognition and the existence of concepts beyond human understanding.', 'The example of rats and prime number mazes is used to illustrate the potential limitations of human cognition, suggesting that there may be concepts necessary to understand the physical universe that are beyond human cognition. Illustrating the limitations of human cognition through the example of rats and prime number mazes.', "Exploration of Gödel's proof and its implications on the completeness and consistency of formal systems, and its relevance to human cognition. Analyzing Gödel's proof and its relevance to the completeness and consistency of formal systems and its connection to human cognition."]}], 'duration': 728.814, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q4195250.jpg', 'highlights': ['People can easily learn new sets of rules, as seen in video games with time travel elements.', 'The natural world exhibits antithetical behaviors that people can adapt to.', "Exploration of Gödel's proof and its implications on the completeness and consistency of formal systems, and its relevance to human cognition.", 'The example of rats and prime number mazes is used to illustrate the potential limitations of human cognition, suggesting that there may be concepts necessary to understand the physical universe that are beyond human cognition.']}, {'end': 5786.395, 'segs': [{'end': 5016.692, 'src': 'embed', 'start': 4987.636, 'weight': 0, 'content': [{'end': 4992.741, 'text': 'but just just the mere fact of survival seems to play a role in imparting that.', 'start': 4987.636, 'duration': 5.105}, {'end': 4996.085, 'text': "Um, You know, and so it'll.", 'start': 4993.562, 'duration': 2.523}, {'end': 5005.488, 'text': "it'll be interesting to see how, over time, we learn more and more about how to apply survival, which we do and say EA algorithms, right?", 'start': 4996.085, 'duration': 9.403}, {'end': 5008.129, 'text': 'Like in EA algorithms, you have a population.', 'start': 5005.808, 'duration': 2.321}, {'end': 5010.69, 'text': "some may survive to reproduce, some don't.", 'start': 5008.129, 'duration': 2.561}, {'end': 5016.692, 'text': 'versus versus, you know observation and reasoning and that type of dynamic you know learning that you do.', 'start': 5010.69, 'duration': 6.002}], 'summary': 'Survival plays a role in learning and applying ea algorithms over time.', 'duration': 29.056, 'max_score': 4987.636, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q4987636.jpg'}, {'end': 5053.136, 'src': 'embed', 'start': 5030.501, 'weight': 3, 'content': [{'end': 5041.328, 'text': 'The human cognitive system has learned to take that those seeds, those seeds of knowledge, and generate with them all kinds of mathematics,', 'start': 5030.501, 'duration': 10.827}, {'end': 5047.052, 'text': 'like all kinds of abstract ideas that have no correspondence to anything that we know of in reality.', 'start': 5041.328, 'duration': 5.724}, {'end': 5048.933, 'text': "And that's pretty, pretty weird too.", 'start': 5048.032, 'duration': 0.901}, {'end': 5050.094, 'text': "Yeah, it's almost like the bootstrap.", 'start': 5048.953, 'duration': 1.141}, {'end': 5053.136, 'text': "Right, Like, like, like it's.", 'start': 5050.974, 'duration': 2.162}], 'summary': 'Human cognitive system can generate abstract ideas and mathematics from knowledge seeds.', 'duration': 22.635, 'max_score': 5030.501, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q5030501.jpg'}, {'end': 5197.37, 'src': 'embed', 'start': 5172.552, 'weight': 2, 'content': [{'end': 5179.974, 'text': "fortunately for us, we're rigidly pre-programmed with extremely rich systems that are part of our biological endowment.", 'start': 5172.552, 'duration': 7.422}, {'end': 5193.728, 'text': 'Correspondingly, a small amount of rather degenerate experience allows a kind of a great leap into a rich cognitive system,', 'start': 5180.578, 'duration': 13.15}, {'end': 5197.37, 'text': 'essentially uniform in a community and in fact roughly uniform for the species.', 'start': 5193.728, 'duration': 3.642}], 'summary': 'Biological endowment enables rapid cognitive development in humans.', 'duration': 24.818, 'max_score': 5172.552, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q5172552.jpg'}, {'end': 5622.502, 'src': 'embed', 'start': 5596.139, 'weight': 1, 'content': [{'end': 5604.791, 'text': 'One comes from the evolution of life in general, which is of course encoded genetically in us, but it comes from the evolution of life in general,', 'start': 5596.139, 'duration': 8.652}, {'end': 5606.153, 'text': 'the fact that it survived.', 'start': 5604.791, 'duration': 1.362}, {'end': 5608.597, 'text': "It's an existential form of knowledge.", 'start': 5606.634, 'duration': 1.963}, {'end': 5611.339, 'text': 'But the interesting thing I mean.', 'start': 5610.199, 'duration': 1.14}, {'end': 5611.839, 'text': 'first of all,', 'start': 5611.339, 'duration': 0.5}, {'end': 5622.502, 'text': "Friston is an empiricist and it's quite easy to look over the history of our evolution to explain how do these cognitive functions get implanted into our brain.", 'start': 5611.839, 'duration': 10.663}], 'summary': 'Existential knowledge encoded genetically drives cognitive functions in our brain.', 'duration': 26.363, 'max_score': 5596.139, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q5596139.jpg'}], 'start': 4924.491, 'title': 'Empiricism and human freedom', 'summary': "Discusses empiricism, encoding of innate knowledge, evolutionary process, and the basis for human freedom and creativity, highlighting survival's role and limits of human cognition.", 'chapters': [{'end': 5136.796, 'start': 4924.491, 'title': 'Empiricism and cognitive templates', 'summary': 'Discusses the concept of empiricism and the encoding of innate knowledge, the evolutionary process of life, and the emergence of abstract ideas, highlighting the role of survival in imparting knowledge and the existence of separate worlds in mathematics and reality.', 'duration': 212.305, 'highlights': ['The human cognitive system has learned to take the seeds of knowledge and generate abstract ideas with no correspondence to reality, while there are things happening in the physical universe for which we have no mapping to mathematics for.', 'Observation, interaction with the environment, and survival play a role in imparting knowledge, and learning to apply survival is demonstrated in EA algorithms with populations and reproduction.', "The chapter discusses the issue of what's the mechanism by which knowledge becomes encoded in the human circuitry or the circuitry of life, expanding the unit of analysis to the species and life in general.", "There's a vast chunk of mathematics that doesn't map to anything in reality, and there are things happening in the physical universe for which we have no sufficient mathematics to describe."]}, {'end': 5786.395, 'start': 5137.516, 'title': 'The basis for human freedom and creativity', 'summary': 'Delves into the concept of rigid genetic programming providing the basis for human freedom and creativity, exploring the role of pre-programming in allowing for rich cognitive systems and the existential form of knowledge encoded into life through survival, while also touching upon the limits of human cognition.', 'duration': 648.879, 'highlights': ['The existence of rigid genetic constraint provides the basis for human freedom and creativity, allowing individuals to construct extremely rich cognitive systems. The rigid genetic constraint is what enables human freedom and creativity, as it allows individuals to construct rich cognitive systems with a small amount of evidence.', 'The rich cognitive systems are part of our biological endowment, developed over countless evolutionary ages, and allow for free and creative human behavior. The rich cognitive systems, developed over evolutionary ages, enable free and creative human behavior, and are uniform within a community and species.', 'The existential form of knowledge is encoded into life through survival, creating an epistemic resonance between the cognitive templates humans have and the reality they live in. The existential form of knowledge is encoded into life through survival, creating a resonance between cognitive templates and the reality humans live in, enhancing their ability to learn and apply new cognitive templates.', 'The discussion also addresses the limits of human cognition, exploring the innate drive to project human-level understandings to different levels of understanding, while questioning the potential deep limits to learning new cognitive templates. The discussion explores the limits of human cognition, addressing the innate drive to project human-level understanding and questioning the potential deep limits to learning new cognitive templates.']}], 'duration': 861.904, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q4924491.jpg', 'highlights': ['Observation, interaction with the environment, and survival play a role in imparting knowledge, demonstrated in EA algorithms with populations and reproduction.', 'The existential form of knowledge is encoded into life through survival, creating a resonance between cognitive templates and the reality humans live in.', 'The rich cognitive systems, developed over evolutionary ages, enable free and creative human behavior, and are uniform within a community and species.', 'The human cognitive system has learned to generate abstract ideas with no correspondence to reality, and there are things happening in the physical universe for which we have no mapping to mathematics for.']}, {'end': 7117.634, 'segs': [{'end': 5954.297, 'src': 'embed', 'start': 5926.044, 'weight': 1, 'content': [{'end': 5934.466, 'text': 'since the mechanical laws explain movements in space, other laws must explain some of the non-spatial workings of the mind,', 'start': 5926.044, 'duration': 8.422}, {'end': 5937.327, 'text': 'which is to say the ghost in the machine.', 'start': 5934.466, 'duration': 2.861}, {'end': 5945.912, 'text': "I'll talk some about Isaac Newton and his contributions to a study of mine that he's not known for that,", 'start': 5937.907, 'duration': 8.005}, {'end': 5954.297, 'text': 'but I think a case can be made that he did make substantial, indirect but nevertheless substantial contributions.', 'start': 5945.912, 'duration': 8.385}], 'summary': "Newton's indirect contributions to non-spatial workings of the mind are substantial.", 'duration': 28.253, 'max_score': 5926.044, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q5926044.jpg'}, {'end': 6299.511, 'src': 'embed', 'start': 6270.273, 'weight': 2, 'content': [{'end': 6274.496, 'text': "that there are scales or regimes of physics where it's just, it's not the way it is.", 'start': 6270.273, 'duration': 4.223}, {'end': 6277.398, 'text': "Like our intuition doesn't apply to that.", 'start': 6274.616, 'duration': 2.782}, {'end': 6281.48, 'text': 'And that was a problem with the action at a distance, like you know gravity, right?', 'start': 6277.478, 'duration': 4.002}, {'end': 6292.547, 'text': "You know what I mean? That it's exerting this force that behaves as if it was pointing at the instantaneous location of that object and acting over,", 'start': 6281.54, 'duration': 11.007}, {'end': 6295.87, 'text': 'you know, vast distances instantaneously, right?', 'start': 6292.547, 'duration': 3.323}, {'end': 6299.511, 'text': "And it doesn't help that, sure, years later.", 'start': 6296.63, 'duration': 2.881}], 'summary': "Physics operates in scales where intuition doesn't apply, such as action at a distance in gravity.", 'duration': 29.238, 'max_score': 6270.273, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q6270273.jpg'}, {'end': 6699.761, 'src': 'embed', 'start': 6669.474, 'weight': 0, 'content': [{'end': 6676.84, 'text': 'And it kind of gets at some of the aspects that you know you and kind of Stanley talked about pretty often too, right?', 'start': 6669.474, 'duration': 7.366}, {'end': 6682.986, 'text': 'Which is, know we should be free right to experiment and have serendipity and creativity.', 'start': 6676.88, 'duration': 6.106}, {'end': 6690.372, 'text': 'utilize those aspects of human cognition when understanding even fundamental physics.', 'start': 6682.986, 'duration': 7.386}, {'end': 6699.761, 'text': 'And you know this analytic idea right of kind of splitting up things and drilling down on the one component that does something,', 'start': 6691.113, 'duration': 8.648}], 'summary': 'Encouraging freedom to experiment and utilize human cognition in understanding physics.', 'duration': 30.287, 'max_score': 6669.474, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q6669474.jpg'}, {'end': 6783.952, 'src': 'embed', 'start': 6753.947, 'weight': 3, 'content': [{'end': 6758.511, 'text': "Which is this ability, unique to humans, that can't possibly be duplicated by machines?", 'start': 6753.947, 'duration': 4.564}, {'end': 6766.297, 'text': 'So he said that you know language was innovative, without bounds, appropriate to the circumstances but not caused by them,', 'start': 6758.571, 'duration': 7.726}, {'end': 6771.321, 'text': 'and can engender thoughts in others which they recognize that they could have expressed themselves.', 'start': 6766.297, 'duration': 5.024}, {'end': 6774.423, 'text': 'So, and this is actually a creative principle of the mind.', 'start': 6771.821, 'duration': 2.602}, {'end': 6775.605, 'text': 'He called res cogitans.', 'start': 6774.444, 'duration': 1.161}, {'end': 6783.952, 'text': 'I remember Chomsky arose that which stood alongside res extensa, you know, which is this Cartesian dualism, you know, the two substances.', 'start': 6775.645, 'duration': 8.307}], 'summary': 'Human language is an innovative, creative principle of the mind, recognized by others, distinct from machines.', 'duration': 30.005, 'max_score': 6753.947, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q6753947.jpg'}], 'start': 5787.095, 'title': 'Beauty, science, and philosophical challenges', 'summary': 'Explores the intersection of beauty and science, critiques mind-body dualism, discusses challenges in representing physical theories, and examines language, mind, and intelligibility, offering diverse perspectives and thought-provoking questions.', 'chapters': [{'end': 5880.503, 'start': 5787.095, 'title': 'Beauty and science', 'summary': "Delves into richard feynman's perspective on appreciating the beauty of a flower, emphasizing the ability of science to enhance rather than diminish aesthetic appreciation and raising thought-provoking questions about the intersection of beauty and science.", 'duration': 93.408, 'highlights': ['Richard Feynman discusses his disagreement with an artist friend who perceives scientific analysis as diminishing the beauty of a flower, highlighting his belief in the coexistence of scientific understanding and aesthetic appreciation.', 'Feynman emphasizes his capability, as a scientist, to appreciate the beauty of a flower while also recognizing the intricate inner workings and evolutionary processes that contribute to its beauty.', 'The chapter concludes with Feynman expressing his view that scientific inquiry adds to the wonder and mystery of a flower, rather than detracting from its beauty, and prompts intriguing questions about the intersection of aesthetics and science.']}, {'end': 6150.458, 'start': 5883.513, 'title': 'Descartes, newton, and the ghost in the machine', 'summary': "Discusses gilbert ryle's critique of descartes' mind-body dualism, newton's indirect contributions to the study of mind, and the impact of the scientific revolution on human understanding, highlighting the limitations and complexity of the world and the lowering of scientific expectations.", 'duration': 266.945, 'highlights': ["Newton's indirect but substantial contributions to the study of mind. Newton is discussed for his substantial, indirect contributions to the study of mind, challenging the traditional view and leading to a significant lowering of scientific expectations.", "Descartes and subsequent philosophers' erroneous belief in the escape route from mechanical laws to explain non-spatial workings of the mind. Descartes and subsequent philosophers erroneously believed that non-spatial workings of the mind could be explained by laws other than mechanical laws, leading to the concept of 'the ghost in the machine.'", 'The impact of the scientific revolution on human understanding, demonstrating the limitations and complexity of the world and the lowering of scientific expectations. The scientific revolution had a significant impact on human understanding, revealing the limitations and complexity of the world, and lowering scientific expectations.']}, {'end': 6732.053, 'start': 6150.458, 'title': 'Challenges in representing physical theories', 'summary': 'Discusses the challenges of representing physical theories, questioning the ability of mathematical equations to accurately represent physical phenomena and highlighting the limitations of human cognition in understanding the underlying reality, as well as emphasizing the role of mathematics as a bridge between the mysterious physical reality and abstract concepts.', 'duration': 581.595, 'highlights': ['The universe is not intelligible to us at certain scales or regimes of physics, challenging our common-sense mechanical intuition, as illustrated by the action at a distance in gravity. Challenges the common-sense mechanical intuition in understanding certain scales or regimes of physics, such as the action at a distance in gravity.', 'Mathematics is a bridge between the unintelligible physical reality and the equally mysterious abstractions and concepts within human cognition. Emphasizes the role of mathematics as a bridge between the mysterious physical reality and abstract concepts within human cognition.', "Science thrives on reductionism by separating phenomena to model and understand them, challenging the mechanical philosophy's view of the world as a machine. Discusses the impact of reductionism on science and challenges the mechanical philosophy's view of the world as a machine."]}, {'end': 7117.634, 'start': 6732.813, 'title': 'Language, mind, and intelligibility', 'summary': "Examines chomsky's view on language and mind, discussing descartes' creative principle, the mystery of free will, and the shift from the world being a machine to constructing intelligible theories, raising questions about the intelligibility of theories themselves.", 'duration': 384.821, 'highlights': ["Chomsky's discussion on Descartes' view of language and thought, emphasizing the creative aspect unique to humans and the ability to engender thoughts in others. Chomsky's emphasis on language's creative aspect and uniqueness to humans, and Descartes' recognition of this creative principle as res cogitans.", 'The mystery of how ideas and linguistic expressions can be generated without being directly caused by external inputs, and the question of the freedom of the will in the context of determinism. The mystery of generating ideas and linguistic expressions not directly caused by inputs, and the conflict between determinism and the freedom of the will.', "Chomsky's perspective on the shift from viewing the world as a machine to constructing intelligible theories, raising questions about the intelligibility of theories themselves. Chomsky's view on the shift from viewing the world as a machine to constructing intelligible theories, and the raised questions about the intelligibility of theories themselves."]}], 'duration': 1330.539, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q5787095.jpg', 'highlights': ['Feynman emphasizes coexistence of scientific understanding and aesthetic appreciation.', "Newton's substantial, indirect contributions to the study of mind challenge traditional views.", 'Challenges common-sense mechanical intuition in understanding certain scales or regimes of physics.', "Chomsky emphasizes language's creative aspect unique to humans and the ability to engender thoughts in others."]}, {'end': 8705.991, 'segs': [{'end': 7230.744, 'src': 'embed', 'start': 7185.012, 'weight': 0, 'content': [{'end': 7189.375, 'text': 'He knew the difference between knowing the name of something and knowing something.', 'start': 7185.012, 'duration': 4.363}, {'end': 7195.739, 'text': 'Feynman said, something can only ever be explained by taking something else for granted.', 'start': 7190.635, 'duration': 5.104}, {'end': 7197, 'text': 'And at some point,', 'start': 7196.159, 'duration': 0.841}, {'end': 7204.745, 'text': 'you need to stop this infinite regression and just admit that you need to take something as true on faith or simply admit that you cannot know.', 'start': 7197, 'duration': 7.745}, {'end': 7210.289, 'text': 'Chapter 6.', 'start': 7208.708, 'duration': 1.581}, {'end': 7219.794, 'text': 'This is a discussion of the Fodor Felician paper in the 1980s on connectionism, where They put this massive critique forwards,', 'start': 7210.289, 'duration': 9.505}, {'end': 7221.335, 'text': "which we think hasn't been answered yet.", 'start': 7219.794, 'duration': 1.541}, {'end': 7226.16, 'text': 'And their main argument centered around productivity and systematicity.', 'start': 7222.236, 'duration': 3.924}, {'end': 7227.521, 'text': "So here's a fact.", 'start': 7226.801, 'duration': 0.72}, {'end': 7230.744, 'text': 'Trains of thought are often like arguments.', 'start': 7228.382, 'duration': 2.362}], 'summary': "Fodor and felician's critique on connectionism remains unanswered, focusing on productivity and systematicity.", 'duration': 45.732, 'max_score': 7185.012, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q7185012.jpg'}, {'end': 7654.472, 'src': 'embed', 'start': 7628.112, 'weight': 2, 'content': [{'end': 7634.156, 'text': 'And really, it kind of hinges on, know, i guess, three main, you know, pillars,', 'start': 7628.112, 'duration': 6.044}, {'end': 7637.178, 'text': "although there's much more in the paper than just the discussion of this.", 'start': 7634.156, 'duration': 3.022}, {'end': 7648.345, 'text': 'but the one is that that, uh, symbolic systems are productive, productive systems by which they mean you can like, take, like, say, a formula,', 'start': 7637.178, 'duration': 11.167}, {'end': 7654.472, 'text': 'for example, that has variables and you can substitute in structures and you can produce,', 'start': 7648.345, 'duration': 6.127}], 'summary': 'Symbolic systems are productive, allowing substitution of structures to produce new outputs.', 'duration': 26.36, 'max_score': 7628.112, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q7628112.jpg'}, {'end': 8182.248, 'src': 'embed', 'start': 8148.263, 'weight': 3, 'content': [{'end': 8152.166, 'text': 'It just has this simple grammar that it can use to figure that out.', 'start': 8148.263, 'duration': 3.903}, {'end': 8154.327, 'text': 'Yeah Okay.', 'start': 8153.326, 'duration': 1.001}, {'end': 8155.487, 'text': 'Well, this makes a lot of sense.', 'start': 8154.587, 'duration': 0.9}, {'end': 8160.29, 'text': 'So, based on reading the Lacoon paper and this connectionism paper so far,', 'start': 8155.507, 'duration': 4.783}, {'end': 8167.595, 'text': 'Lacoon is saying we need to have a probabilistic-ish interpretation of possible futures.', 'start': 8160.29, 'duration': 7.305}, {'end': 8173.258, 'text': 'The connectionism paper is saying that we need to have composable, recomposable, decomposable abstractions.', 'start': 8168.055, 'duration': 5.203}, {'end': 8182.248, 'text': "Right And this is very, it's almost analogous to what Francois Chalet talks about with his library of modules and type two traversal, et cetera.", 'start': 8173.318, 'duration': 8.93}], 'summary': 'Lacoon emphasizes probabilistic interpretation, while connectionism focuses on composable abstractions.', 'duration': 33.985, 'max_score': 8148.263, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q8148263.jpg'}], 'start': 7118.686, 'title': 'Understanding inertia and cognitive science', 'summary': "Delves into understanding inertia through feynman's analogy and discusses fodor felician's 1980s critique of connectionism in cognitive science, highlighting unresolved issues, the importance of symbolic systems, and challenges in natural language understanding involving probabilistic interpretation and memory separation.", 'chapters': [{'end': 7210.289, 'start': 7118.686, 'title': 'Understanding inertia', 'summary': "Discusses a deep understanding of inertia by explaining the concept through a child's observation of a ball's movement in a wagon and emphasizing the difference between knowing the name of something and truly understanding it, as conveyed by feynman. chapter 6.", 'duration': 91.603, 'highlights': ["The concept of inertia is explained through a child's observation of a ball's movement in a wagon, where the ball rushes to the back when the wagon is pulled and rolls to the front when the wagon stops.", 'Feynman emphasizes the difference between knowing the name of something and truly understanding it, stating that something can only be explained by taking something else for granted and highlighting the need to stop infinite regression in understanding. Chapter 6.', "The chapter delves into the idea that true understanding cannot be achieved by simply knowing the name of something, as exemplified by Feynman's insight that explanations require taking something for granted and acknowledging the limits of knowledge. Chapter 6."]}, {'end': 7598.141, 'start': 7210.289, 'title': "Fodor felician's 1980s connectionism critique", 'summary': "Discusses fodor felician's 1980s critique of connectionism, highlighting the unresolved issues around productivity and systematicity in cognitive science, along with the key points that define the field and the argument for the finiteness of the mind, emphasizing the importance of a combinatorial syntax and semantics for mental representation, and the systematic interrelations among human thoughts.", 'duration': 387.852, 'highlights': ['Fodor and Filishin wrote a seminal critique of connectionism in the late 1980s, addressing unresolved issues around productivity and systematicity in cognitive science.', "Turing's idea that mental states are syntactically structured and determine the causal role in mental processes defines the field of cognitive science.", 'The argument for the finiteness of the mind challenges the assumption of infinite capacity and highlights the importance of a combinatorial syntax and semantics for mental representation.', 'Systematic interrelations among human thoughts, exemplified by the property of systematicity in natural language, demand a principled explanation and imply the systematic nature of mental representation.']}, {'end': 8081.92, 'start': 7598.141, 'title': 'Symbolic systems vs connectionist architectures', 'summary': 'Discusses the fundamental differences between symbolic systems and connectionist architectures, emphasizing the productive and compositional nature of symbolic systems, and the limitations of connectionist networks in capturing and analyzing individual parts of a structure.', 'duration': 483.779, 'highlights': ['Symbolic systems are productive, generating infinite structures from a simple set of rules, maintaining consistency and unbounded in time. Symbolic systems are productive, allowing for the generation of an infinite set of sentences from a simple set of rules, ensuring consistency and unbounded computation.', 'Symbolic systems are compositional, enabling the analysis and comparison of individual parts of a structure, while connectionist networks collapse calculations into a single output, lacking the ability to capture and analyze individual components. Symbolic systems are compositional, allowing for the analysis and comparison of individual components of a structure, while connectionist networks collapse calculations into a single output, hindering the capture and analysis of individual parts.', 'Intention vs extension: Symbolic systems maintain intention as formulas, enabling the generation of extensions for different circumstances, while materialized outputs lack generalizability and understanding. Symbolic systems maintain intention as formulas, allowing the generation of extensions for different circumstances, preserving generalizability, while materialized outputs lack the same level of understanding and generalizability.']}, {'end': 8705.991, 'start': 8081.94, 'title': 'Challenges in natural language understanding', 'summary': "Discusses challenges in natural language understanding, including the need for probabilistic interpretation of possible futures, composable abstractions, and the separation of algorithm and memory in symbolic systems. it also details the efforts to salvage a corrupted recording of chomsky's interview through transcription and voice cloning.", 'duration': 624.051, 'highlights': ['Challenges in Natural Language Understanding The chapter discusses the need for a probabilistic interpretation of possible futures and composable, recomposable, decomposable abstractions in neural networks, along with the separation of algorithm and memory in symbolic systems.', "Salvaging Corrupted Recording of Chomsky's Interview Efforts were made to salvage a corrupted recording through transcription and voice cloning, involving the synthesis of a transcript, creation of a voice clone model, and the use of time warping algorithm to align the original recording with the synthesized version.", "Chomsky's Interview Mishap The interview with Chomsky suffered from a recording failure, which led to the use of technology to recover the content, including transcription, voice cloning, and alignment with the original recording using a time warping algorithm."]}], 'duration': 1587.305, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q7118686.jpg', 'highlights': ["Feynman explains inertia through a child's observation of a ball's movement in a wagon, emphasizing the need to stop infinite regression in understanding.", 'Fodor and Filishin critique connectionism in the late 1980s, addressing unresolved issues around productivity and systematicity in cognitive science.', 'Symbolic systems are productive, generating infinite structures from a simple set of rules, maintaining consistency and unbounded in time.', 'Challenges in Natural Language Understanding involve the need for a probabilistic interpretation of possible futures and memory separation in symbolic systems.']}, {'end': 9523.588, 'segs': [{'end': 8738.176, 'src': 'embed', 'start': 8706.571, 'weight': 0, 'content': [{'end': 8711.996, 'text': 'you know the minimum cost alignment between the two tracks and then between all of the aligned words.', 'start': 8706.571, 'duration': 5.425}, {'end': 8714.077, 'text': 'we just did a linear frame interpolation.', 'start': 8711.996, 'duration': 2.081}, {'end': 8720.203, 'text': 'so when he was saying hello in one script and hello in another script, we just did a linear frame interpolation.', 'start': 8714.077, 'duration': 6.126}, {'end': 8727.128, 'text': "and, as you can imagine, there's all sorts of things that can go wrong when you do that kind of coding, because there's numerical precision problems,", 'start': 8720.203, 'duration': 6.925}, {'end': 8732.413, 'text': 'you know, because we were dealing with hundreds of thousands of frames, so it was drifting over time.', 'start': 8727.128, 'duration': 5.285}, {'end': 8738.176, 'text': "maybe in the future we'll make another video about how we did that, but Needless to say, it was in linear time complexity,", 'start': 8732.413, 'duration': 5.763}], 'summary': 'Achieved minimum cost alignment and linear frame interpolation for hundreds of thousands of frames.', 'duration': 31.605, 'max_score': 8706.571, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q8706571.jpg'}, {'end': 8810.462, 'src': 'embed', 'start': 8780.312, 'weight': 2, 'content': [{'end': 8782.333, 'text': 'And it really was Chomsky saying all of that stuff.', 'start': 8780.312, 'duration': 2.021}, {'end': 8788.335, 'text': "So there are a couple of occasions where we've actually inserted in a little bit of the original Chomsky, even though it was corrupted,", 'start': 8782.893, 'duration': 5.442}, {'end': 8791.936, 'text': 'just to capture him chuckling or, or you know, saying some words.', 'start': 8788.335, 'duration': 3.601}, {'end': 8796.778, 'text': 'I mean, I think there is a point where he said language models have achieved zero, zero.', 'start': 8791.956, 'duration': 4.822}, {'end': 8797.818, 'text': "They've done nothing.", 'start': 8796.838, 'duration': 0.98}, {'end': 8803.02, 'text': 'And obviously we just wanted to capture the original sentiment of Chomsky when he was saying those things.', 'start': 8798.518, 'duration': 4.502}, {'end': 8810.462, 'text': 'In certain parts, the generated voice is roughly twice the speed of Chomsky, and we felt that was fair, to be honest,', 'start': 8804.14, 'duration': 6.322}], 'summary': "Chomsky's sentiments on language models, with some original audio inserted, and voice speed adjusted.", 'duration': 30.15, 'max_score': 8780.312, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q8780312.jpg'}, {'end': 8860.338, 'src': 'embed', 'start': 8834.615, 'weight': 3, 'content': [{'end': 8841.381, 'text': "We've basically used engineering technology, which is something that Chomsky talks about in the podcast, to recover a broken interview.", 'start': 8834.615, 'duration': 6.766}, {'end': 8843.883, 'text': "And we got Chomsky's full permission.", 'start': 8841.861, 'duration': 2.022}, {'end': 8846.065, 'text': 'So this is the email that we got back from Chomsky.', 'start': 8843.943, 'duration': 2.122}, {'end': 8848.568, 'text': 'By the way, this email means so much to me personally.', 'start': 8846.446, 'duration': 2.122}, {'end': 8850.009, 'text': "I'm going to frame this email.", 'start': 8848.588, 'duration': 1.421}, {'end': 8851.791, 'text': 'I should put this email in my CV.', 'start': 8850.329, 'duration': 1.462}, {'end': 8854.273, 'text': 'Just doing all of this work,', 'start': 8853.092, 'duration': 1.181}, {'end': 8860.338, 'text': 'having this story to tell and getting this response from Chomsky to me is something that I could tell my grandchildren about.', 'start': 8854.273, 'duration': 6.065}], 'summary': 'Using engineering technology, we recovered a broken interview with chomsky, obtaining his full permission, which means a lot personally and professionally.', 'duration': 25.723, 'max_score': 8834.615, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q8834615.jpg'}, {'end': 9082.707, 'src': 'embed', 'start': 9053.716, 'weight': 4, 'content': [{'end': 9059.482, 'text': 'Chomsky is a big believer in autonomy, free will, creativity, and novelty.', 'start': 9053.716, 'duration': 5.766}, {'end': 9063.806, 'text': "Shouldn't be entirely surprising given that he isn't a narco-syndicalist.", 'start': 9060.202, 'duration': 3.604}, {'end': 9070.973, 'text': "It's really important for him that we are individual actors that have free will that is not determined by the situation we're in,", 'start': 9064.366, 'duration': 6.607}, {'end': 9077.118, 'text': 'which is why he quite often says that language use is appropriate to the situation but not caused by the situation.', 'start': 9070.973, 'duration': 6.145}, {'end': 9082.707, 'text': "But the really interesting thing that he says about language is that it's an expression of human creativity.", 'start': 9077.881, 'duration': 4.826}], 'summary': 'Chomsky emphasizes autonomy, free will, and creativity in language use.', 'duration': 28.991, 'max_score': 9053.716, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q9053716.jpg'}, {'end': 9213.44, 'src': 'embed', 'start': 9180.965, 'weight': 5, 'content': [{'end': 9184.626, 'text': 'especially as they relate to many hot topics in artificial intelligence.', 'start': 9180.965, 'duration': 3.661}, {'end': 9189.628, 'text': "Chomsky's goal as a linguist is to find principles that are common to all languages,", 'start': 9184.806, 'duration': 4.822}, {'end': 9193.669, 'text': 'which allow people to creatively speak freely and understand each other.', 'start': 9189.628, 'duration': 4.041}, {'end': 9196.735, 'text': "Chomsky's work is so much more than linguistics.", 'start': 9194.795, 'duration': 1.94}, {'end': 9205.398, 'text': 'He actually thinks that linguistics should be a branch of psychology and that so much about our language actually determines how we behave as human beings?', 'start': 9197.076, 'duration': 8.322}, {'end': 9213.44, 'text': 'I think finding the principles common to all languages and understanding what enables us to speak freely and, importantly,', 'start': 9206.118, 'duration': 7.322}], 'summary': 'Chomsky aims to find common language principles for free speech and understanding.', 'duration': 32.475, 'max_score': 9180.965, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q9180965.jpg'}], 'start': 8706.571, 'title': 'Language and script alignment', 'summary': "Discusses linear frame interpolation for script alignment, highlighting potential numerical precision issues and linear time complexity. it also delves into chomsky's deepfake interview, covering ethical concerns, linguistic insights, and critique of large language models, with a comparison to a physics paper.", 'chapters': [{'end': 8738.176, 'start': 8706.571, 'title': 'Linear frame interpolation in script alignment', 'summary': 'Discusses the linear frame interpolation used to align words in different scripts, with potential numerical precision issues when dealing with hundreds of thousands of frames, resulting in linear time complexity.', 'duration': 31.605, 'highlights': ['The linear frame interpolation was used to align words in different scripts, potentially resulting in numerical precision problems when dealing with hundreds of thousands of frames.', 'The process had a linear time complexity, which may lead to potential numerical precision issues due to dealing with hundreds of thousands of frames.']}, {'end': 9523.588, 'start': 8738.176, 'title': "Chomsky's deepfake interview and language insights", 'summary': "Discusses the creation of a deepfake interview with chomsky, addressing ethical concerns, the linguistic insights shared by chomsky, and his critique of large language models, emphasizing their failure to achieve anything in the domain, with a comparison to a physics paper that says 'anything goes.'", 'duration': 785.412, 'highlights': ["Chomsky's critique of large language models, stating that they have achieved zero in the domain, likening them to a physics paper that says 'anything goes,' highlighting their deficiency and lack of contribution to science.", "The discussion of the creation of a deepfake interview with Chomsky, addressing ethical concerns, emphasizing the use of engineering technology to recover a broken interview and obtaining Chomsky's permission for the synthesis, showcasing the team's emotional investment and the unique insights gained from the original interview.", "Insights into language and human creativity, discussing Chomsky's belief in autonomy, free will, creativity, and novelty in language use, reflecting on the capacity of language to produce and understand new thoughts without limit and without control, and the importance of language as an expression of human creativity and infinite space of possible expression.", 'The introduction and background of Chomsky, highlighting his influence and contributions to linguistics, cognitive science, and social criticism, and the significance of the interview with him as a dream come true for the team and a rare opportunity to discuss contemporary issues and hot topics in artificial intelligence.']}], 'duration': 817.017, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q8706571.jpg', 'highlights': ['The linear frame interpolation was used to align words in different scripts, potentially resulting in numerical precision problems when dealing with hundreds of thousands of frames.', 'The process had a linear time complexity, which may lead to potential numerical precision issues due to dealing with hundreds of thousands of frames.', "Chomsky's critique of large language models, stating that they have achieved zero in the domain, likening them to a physics paper that says 'anything goes,' highlighting their deficiency and lack of contribution to science.", "The discussion of the creation of a deepfake interview with Chomsky, addressing ethical concerns, emphasizing the use of engineering technology to recover a broken interview and obtaining Chomsky's permission for the synthesis, showcasing the team's emotional investment and the unique insights gained from the original interview.", "Insights into language and human creativity, discussing Chomsky's belief in autonomy, free will, creativity, and novelty in language use, reflecting on the capacity of language to produce and understand new thoughts without limit and without control, and the importance of language as an expression of human creativity and infinite space of possible expression.", 'The introduction and background of Chomsky, highlighting his influence and contributions to linguistics, cognitive science, and social criticism, and the significance of the interview with him as a dream come true for the team and a rare opportunity to discuss contemporary issues and hot topics in artificial intelligence.']}, {'end': 11043.95, 'segs': [{'end': 9607.455, 'src': 'embed', 'start': 9578.992, 'weight': 0, 'content': [{'end': 9587.801, 'text': "I'd like to know your thoughts on the current rise of connectionism or the resurgence of connectionism,", 'start': 9578.992, 'duration': 8.809}, {'end': 9592.265, 'text': "let's put it that way and the ostensible success of deep learning.", 'start': 9587.801, 'duration': 4.464}, {'end': 9595.028, 'text': "And specifically, I'd like to know.", 'start': 9593.186, 'duration': 1.842}, {'end': 9607.455, 'text': 'Do you think the classic Fodor and Pylyshyn critique that was written in the classic paper Connectionism in Cognitive Architecture a Critical Analysis?', 'start': 9596.149, 'duration': 11.306}], 'summary': 'Requesting thoughts on rise of connectionism and success of deep learning', 'duration': 28.463, 'max_score': 9578.992, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q9578992.jpg'}, {'end': 10024.787, 'src': 'embed', 'start': 9995.831, 'weight': 1, 'content': [{'end': 9997.712, 'text': 'That was the time when AI was beginning.', 'start': 9995.831, 'duration': 1.881}, {'end': 10007.499, 'text': 'Morven Minsky, Herb Simon, other people, Alan Turing, who were, in their view, AI was supposed to be a study of the nature of intelligence.', 'start': 9997.992, 'duration': 9.507}, {'end': 10009.66, 'text': 'It was a scientific field.', 'start': 10008.219, 'duration': 1.441}, {'end': 10011.641, 'text': "By now that's disappeared.", 'start': 10010.32, 'duration': 1.321}, {'end': 10013.602, 'text': "Not anybody's interested.", 'start': 10012.302, 'duration': 1.3}, {'end': 10017.065, 'text': 'But MIT at that time was an engineering school.', 'start': 10014.163, 'duration': 2.902}, {'end': 10022.226, 'text': 'There were great people in math and physics, but they were basically teachers and engineers.', 'start': 10017.685, 'duration': 4.541}, {'end': 10024.787, 'text': 'It changed in about 10 years.', 'start': 10023.327, 'duration': 1.46}], 'summary': 'Ai was once a scientific field but has now disappeared, mit transformed to an engineering school in about 10 years.', 'duration': 28.956, 'max_score': 9995.831, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q9995831.jpg'}, {'end': 10157.707, 'src': 'embed', 'start': 10131.171, 'weight': 2, 'content': [{'end': 10136.737, 'text': 'Our biological wetware implements a kind of hybrid analog and digital computation,', 'start': 10131.171, 'duration': 5.566}, {'end': 10143.784, 'text': 'which might might realize aspects that are effectively impossible to replicate in digital circuits alone.', 'start': 10136.737, 'duration': 7.047}, {'end': 10154.442, 'text': 'So Roger Penrose goes as far as to hypothesize that our brains take advantage of quantum properties to access non-computable oracles,', 'start': 10144.605, 'duration': 9.837}, {'end': 10157.707, 'text': 'making our brains what Turing would have called oracle machines.', 'start': 10154.442, 'duration': 3.265}], 'summary': 'Our brains may use quantum properties to access non-computable oracles, as per roger penrose.', 'duration': 26.536, 'max_score': 10131.171, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q10131171.jpg'}, {'end': 10477.229, 'src': 'embed', 'start': 10446.462, 'weight': 4, 'content': [{'end': 10448.183, 'text': 'I just wondered what do you think of this view?', 'start': 10446.462, 'duration': 1.721}, {'end': 10455.553, 'text': 'Well, first, are there questions that can be formulated that are outside of our cognitive range?', 'start': 10449.249, 'duration': 6.304}, {'end': 10458.515, 'text': "I think it would be a miracle if it's not true.", 'start': 10456.114, 'duration': 2.401}, {'end': 10461.677, 'text': "Unless we're angels, that's going to be true.", 'start': 10459.116, 'duration': 2.561}, {'end': 10468.262, 'text': "If we are organic creatures, part of the organic world, then there'll be scope and limits to our capacities.", 'start': 10461.917, 'duration': 6.345}, {'end': 10477.229, 'text': 'In fact, the scope and limits are related So I have the capacity to walk much faster than a chimpanzee, much better than an eagle.', 'start': 10468.582, 'duration': 8.647}], 'summary': 'Discussion on cognitive capacities and organic limitations.', 'duration': 30.767, 'max_score': 10446.462, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q10446462.jpg'}, {'end': 10639.624, 'src': 'embed', 'start': 10608.066, 'weight': 3, 'content': [{'end': 10610.407, 'text': 'Newton showed that the universe is not intelligible.', 'start': 10608.066, 'duration': 2.341}, {'end': 10613.589, 'text': 'And what happened after Newton took a long time.', 'start': 10611.027, 'duration': 2.562}, {'end': 10616.43, 'text': 'Science just reduced its aspirations.', 'start': 10614.489, 'duration': 1.941}, {'end': 10619.732, 'text': "It doesn't seek to find an intelligible universe.", 'start': 10617.11, 'duration': 2.622}, {'end': 10623.714, 'text': 'It just seeks to find intelligible theories about the universe.', 'start': 10620.592, 'duration': 3.122}, {'end': 10627.195, 'text': "So Leibniz could understand Newton's theories.", 'start': 10624.674, 'duration': 2.521}, {'end': 10629.056, 'text': 'They were not unintelligible.', 'start': 10627.715, 'duration': 1.341}, {'end': 10632.258, 'text': 'It was the world that we were describing that was unintelligible.', 'start': 10629.356, 'duration': 2.902}, {'end': 10635.6, 'text': "Well, that's a big shift in the nature of science.", 'start': 10633.118, 'duration': 2.482}, {'end': 10639.624, 'text': "It wasn't particularly recognized, but it just became tacit.", 'start': 10636.161, 'duration': 3.463}], 'summary': "Newton's theories led to a shift in science, focusing on intelligible theories about the universe, not seeking an intelligible universe itself.", 'duration': 31.558, 'max_score': 10608.066, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q10608066.jpg'}], 'start': 9524.248, 'title': 'Evolving science and cognition', 'summary': "Discusses the rise of connectionism and the critique of deep learning, the evolution of engineering and science, the potential quantum properties of the human brain, newton's impact on the nature of science, and the mystery of cognitive processes and semantics theories.", 'chapters': [{'end': 9924.351, 'start': 9524.248, 'title': 'Rise of connectionism and the illusion of deep learning', 'summary': 'Discusses the rise of connectionism, the critique of deep learning, and the innate nature of empiricism, arguing that the entire framework of connectionism is wrong and suggesting the need to break out of the current paradigm in cognitive science.', 'duration': 400.103, 'highlights': ['The critique of deep learning and the rise of connectionism The chapter delves into the critique of deep learning and the resurgence of connectionism, suggesting that the entire framework of these approaches is wrong and refuted by evidence.', 'The innate nature of empiricism The transcript discusses a study showing that empiricism is innate, with evidence indicating that it comes naturally to humans, akin to an instinct, challenging the success of deep learning.', "The need to break out of the current paradigm in cognitive science The chapter argues for the necessity to break out of the current cognitive science paradigm, likening the current state to that of the 16th century in physics and suggesting the difficulty in doing so, drawing parallels to the resistance faced by Newton's theories in the field of physics."]}, {'end': 10130.511, 'start': 9925.331, 'title': 'Evolution of engineering and science', 'summary': 'Discusses the evolution of engineering and science, highlighting the transition from ai as a scientific field to an engineering one, the impact of engineering techniques like deep learning on fields such as protein folding, and the changing dynamics at mit in the 1950s and 1960s.', 'duration': 205.18, 'highlights': ['The transition of AI from a scientific field to an engineering one at MIT in the 1950s and 1960s, leading to a significant impact on the nature of the institution and the integration of humanities into the curriculum.', 'The contribution of engineering techniques like deep learning to fields such as protein folding, live transcription, and speech recognition, showcasing their impact on advancing understanding and technological applications.', 'The diminishing focus on the nature of intelligence and cognitive sciences in AI, with a shift towards an engineering-centric approach, leading to notable achievements but a lack of contribution to science.']}, {'end': 10536.765, 'start': 10131.171, 'title': "Brain's quantum properties & human cognition", 'summary': 'Discusses the potential quantum properties of the human brain, the limitations of digital replication in silico, and the relevance of open-ended exploration in scientific progress.', 'duration': 405.594, 'highlights': ['The human brain might utilize quantum properties to access non-computable oracles, making it akin to an oracle machine. The chapter explores the hypothesis that the human brain takes advantage of quantum properties to access non-computable oracles, suggesting a unique computational capability.', "The limitations of digitally replicating human cognition in a silicon system are discussed, emphasizing the organic nature of the brain's functioning. The discussion highlights the challenge of digitally replicating human cognition in a silicon system and emphasizes the organic nature of the brain's functioning, suggesting a fundamental difference that cannot be easily simulated.", "The concept of open-ended exploration, termed as 'treasure hunting,' is proposed as essential for scientific progress, potentially surpassing the limitations of formal objectives. The view on open-ended exploration, referred to as 'treasure hunting,' is presented as necessary for scientific progress, suggesting that it can lead to valuable discoveries beyond the constraints of formal objectives."]}, {'end': 10722.573, 'start': 10537.325, 'title': "Newton's impact on the nature of science", 'summary': "Discusses how newton's theories led to the shift in the nature of science, reducing its aspirations to find an intelligible universe and focusing on intelligible theories, with examples such as gpt-3 and deep learning, while acknowledging the existence of mysteries beyond human understanding, like the 17th century hard problem of motion.", 'duration': 185.248, 'highlights': ["Newton's theories led to a shift in the nature of science, reducing its aspirations to find an intelligible universe and focusing on intelligible theories, such as GPT-3 and deep learning, while acknowledging the existence of mysteries beyond human understanding.", "The 17th century hard problem of motion is one of the mysteries beyond human understanding, where zero progress has been made, and physicists' explanations, such as gravitons in a quantum system, are accepted beyond human capacity to conceive.", 'Science no longer seeks to find an intelligible universe but instead seeks intelligible theories about the universe, a significant shift from the goal of early modern science to find an intelligible universe directed by a super skilled artisan.']}, {'end': 11043.95, 'start': 10723.853, 'title': 'Semantics and cognitive horizon', 'summary': 'Explores the profound mystery of cognitive processes and the limitations of human understanding, discusses the challenges of semantics theories, and delves into the intricate nature of word meanings and conceptual structures within language.', 'duration': 320.097, 'highlights': ['The profound mystery of cognitive processes and the limitations of human understanding The speaker discusses the mystery of how humans construct thoughts and communicate them, highlighting the inability to comprehend even simple actions like lifting a finger, showcasing the profound limitations of human understanding.', "Challenges of semantics theories and limitations in understanding word meanings The chapter delves into the challenges of semantic theories, emphasizing that the meanings of words are largely based on conceptual structures and mental objects, as demonstrated through examples like 'house' and 'river.'", "Intricate nature of word meanings and conceptual structures within language The intricate nature of word meanings and conceptual structures within language is explored, with examples such as the fluidity of the meaning of 'river' and the influence of conceptual structures on word meanings."]}], 'duration': 1519.702, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q9524248.jpg', 'highlights': ['The critique of deep learning and the rise of connectionism', 'The transition of AI from a scientific field to an engineering one at MIT in the 1950s and 1960s', 'The human brain might utilize quantum properties to access non-computable oracles', "Newton's theories led to a shift in the nature of science, reducing its aspirations to find an intelligible universe", 'The profound mystery of cognitive processes and the limitations of human understanding']}, {'end': 12031.772, 'segs': [{'end': 11171.96, 'src': 'embed', 'start': 11098.402, 'weight': 0, 'content': [{'end': 11107.887, 'text': "And the proposed answer is, there's an event reading, there's an agent John, there's an agent book, and there's an adverb, the modifier of the event.", 'start': 11098.402, 'duration': 9.485}, {'end': 11114.591, 'text': "Quickly, if you analyze it that way, that's just a concatenation, and you get the inferences.", 'start': 11108.428, 'duration': 6.163}, {'end': 11121.577, 'text': "that's been developed extensively by people like Piotrowski and Shane, among others.", 'start': 11115.672, 'duration': 5.905}, {'end': 11130.164, 'text': "that happens to fit very naturally to what I think we're coming to understand as pure syntax.", 'start': 11121.577, 'duration': 8.587}, {'end': 11139.171, 'text': "It seems that that's the way pure syntax provides structures of that nature, which fits very naturally into events and semantics.", 'start': 11131.205, 'duration': 7.966}, {'end': 11142.675, 'text': 'But notice the event semantics, the syntax.', 'start': 11140.252, 'duration': 2.423}, {'end': 11146.478, 'text': "When you talk about an event, it's not anything in the world.", 'start': 11143.235, 'duration': 3.243}, {'end': 11149.121, 'text': "It's something that we construct in our minds.", 'start': 11147.039, 'duration': 2.082}, {'end': 11154.426, 'text': 'There was a gentleman named Zeno who taught us something about that.', 'start': 11150.322, 'duration': 4.104}, {'end': 11160.712, 'text': 'How many events are there when I cross a room? As many as you decide to put there.', 'start': 11155.787, 'duration': 4.925}, {'end': 11164.535, 'text': "You know, there's no end up to the power of the continuum.", 'start': 11161.492, 'duration': 3.043}, {'end': 11169.619, 'text': 'So event semantics, I think, is productive as a form of syntax.', 'start': 11165.836, 'duration': 3.783}, {'end': 11171.96, 'text': 'Then comes another question.', 'start': 11170.58, 'duration': 1.38}], 'summary': 'Event semantics in syntax, as proposed by piotrowski and shane, fits naturally and provides productive structures.', 'duration': 73.558, 'max_score': 11098.402, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q11098402.jpg'}, {'end': 11234.548, 'src': 'embed', 'start': 11201.08, 'weight': 5, 'content': [{'end': 11213.049, 'text': "What do you think is the relationship between what you have called universal grammar, or the I language, and Fodor's language of thought,", 'start': 11201.08, 'duration': 11.969}, {'end': 11221.635, 'text': 'which has been also quite a theory in linguistics and cognitive science?', 'start': 11213.049, 'duration': 8.586}, {'end': 11231.606, 'text': 'And if we can suppose that, both as human innate systems endowed by genetics and or the laws of nature?', 'start': 11222.601, 'duration': 9.005}, {'end': 11234.548, 'text': 'if that is a similarity, or is it??', 'start': 11231.606, 'duration': 2.942}], 'summary': "Examining the connection between universal grammar and fodor's language of thought in linguistics and cognitive science.", 'duration': 33.468, 'max_score': 11201.08, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q11201080.jpg'}, {'end': 11429.121, 'src': 'embed', 'start': 11402.529, 'weight': 4, 'content': [{'end': 11410.411, 'text': 'It has the capacity to acquire a language, any infant, as far as we know, can acquire any language with equal facility.', 'start': 11402.529, 'duration': 7.882}, {'end': 11414.072, 'text': "So it's probably uniform, which would not be very surprising.", 'start': 11411.032, 'duration': 3.04}, {'end': 11418.834, 'text': 'Humans are a very recent species, a couple hundred thousand years.', 'start': 11414.793, 'duration': 4.041}, {'end': 11421.996, 'text': "That's a flick of an eye in evolutionary time.", 'start': 11419.514, 'duration': 2.482}, {'end': 11429.121, 'text': 'And we know from genomic evidence that humans began to separate on the order of 150, 000 years ago.', 'start': 11422.656, 'duration': 6.465}], 'summary': 'Humans can acquire any language with equal facility, evolving for 150,000 years.', 'duration': 26.592, 'max_score': 11402.529, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q11402529.jpg'}, {'end': 11545.882, 'src': 'embed', 'start': 11503.705, 'weight': 6, 'content': [{'end': 11505.566, 'text': "So that's the way Mother Nature works.", 'start': 11503.705, 'duration': 1.861}, {'end': 11511.367, 'text': "Can't really give an explanation, but it's so overwhelmingly supported that nobody even questions it.", 'start': 11506.266, 'duration': 5.101}, {'end': 11515.568, 'text': "And if you don't have the simplest solution, you figure you're wrong.", 'start': 11512.047, 'duration': 3.521}, {'end': 11518.849, 'text': "You know, that's the ordinary scientific way.", 'start': 11515.988, 'duration': 2.861}, {'end': 11522.87, 'text': 'And as I say, Einstein just called it the magical creed.', 'start': 11519.309, 'duration': 3.561}, {'end': 11524.351, 'text': "That's the way it is.", 'start': 11523.49, 'duration': 0.861}, {'end': 11526.051, 'text': "That's how nature is.", 'start': 11524.371, 'duration': 1.68}, {'end': 11537.177, 'text': "So we'd expect that when Mother Nature some event took place, random event, which provided Homo sapiens with the capacity for recursive enumeration,", 'start': 11526.832, 'duration': 10.345}, {'end': 11542.28, 'text': 'the fundamental property of the computational system, No other organism has it.', 'start': 11537.177, 'duration': 5.103}, {'end': 11544.361, 'text': "It's nowhere, you know.", 'start': 11542.96, 'duration': 1.401}, {'end': 11545.882, 'text': "It's uniform in humans.", 'start': 11544.641, 'duration': 1.241}], 'summary': "Mother nature's random event enabled humans' unique computational capacity.", 'duration': 42.177, 'max_score': 11503.705, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q11503705.jpg'}, {'end': 11719.318, 'src': 'heatmap', 'start': 11581.299, 'weight': 0.706, 'content': [{'end': 11593.264, 'text': 'So the tasks of researchers in linguistics and cognitive science ought to be see if you can show that the simplest possible solution to this conundrum yields explanation for the phenomenon of language.', 'start': 11581.299, 'duration': 11.965}, {'end': 11594.784, 'text': "That's the task of the field.", 'start': 11593.484, 'duration': 1.3}, {'end': 11597.045, 'text': "Almost nobody's interested in it.", 'start': 11595.425, 'duration': 1.62}, {'end': 11602.227, 'text': "You can count the number of people on the fingers of one hand, but that's what the field ought to be.", 'start': 11597.605, 'duration': 4.622}, {'end': 11604.568, 'text': "And I think they're progressing in that.", 'start': 11602.247, 'duration': 2.321}, {'end': 11610.39, 'text': "I think we're maybe entering a new era where, for the first time, first time ever,", 'start': 11604.828, 'duration': 5.562}, {'end': 11614.331, 'text': 'we can give genuine explanations for fundamental properties of language.', 'start': 11610.39, 'duration': 3.941}, {'end': 11616.632, 'text': 'In fact, one of them I mentioned.', 'start': 11615.151, 'duration': 1.481}, {'end': 11624.178, 'text': "the most striking, dramatic feature of language is what's called structure dependence the fact that from infancy,", 'start': 11616.632, 'duration': 7.546}, {'end': 11631.022, 'text': 'every human understands unconsciously that all the rules of language, all operations in language,', 'start': 11624.178, 'duration': 6.844}, {'end': 11634.365, 'text': 'have to ignore the order of words and deal just with structures.', 'start': 11631.022, 'duration': 3.343}, {'end': 11639.248, 'text': "So ignore everything you've heard, deal with the abstract structures in your mind.", 'start': 11635.205, 'duration': 4.043}, {'end': 11642.491, 'text': 'You can demonstrate this directly, overwhelmingly.', 'start': 11639.849, 'duration': 2.642}, {'end': 11644.272, 'text': "That's the way it works.", 'start': 11643.271, 'duration': 1.001}, {'end': 11646.234, 'text': 'We now have an explanation for it.', 'start': 11644.552, 'duration': 1.682}, {'end': 11651.319, 'text': "Turns out that that's what follows from the simplest combinatorial operation.", 'start': 11647.355, 'duration': 3.964}, {'end': 11659.768, 'text': "The simplest combinatorial operation happens to be binary set formation, what's called merge in contemporary literature.", 'start': 11651.339, 'duration': 8.429}, {'end': 11667.093, 'text': 'Well, if language is based on binary set formation, you get this property, no linear largest structures.', 'start': 11660.408, 'duration': 6.685}, {'end': 11676.98, 'text': 'So we had a, for the first time ever, a deep explanation for the most fundamental property of language, which is a very surprising property,', 'start': 11667.953, 'duration': 9.027}, {'end': 11679.642, 'text': 'which tells you something about learning cognition.', 'start': 11676.98, 'duration': 2.662}, {'end': 11683.063, 'text': "and so on, almost nobody's interested in it.", 'start': 11680.422, 'duration': 2.641}, {'end': 11686.284, 'text': 'You take a look at the literature in cognitive science,', 'start': 11683.283, 'duration': 3.001}, {'end': 11695.006, 'text': "there's an endless number of papers trying to show that by massive statistical analysis of huge amounts of data, you can begin to approximate,", 'start': 11686.284, 'duration': 8.722}, {'end': 11696.306, 'text': 'but you can explain nothing.', 'start': 11695.006, 'duration': 1.3}, {'end': 11698.447, 'text': 'I mean, of course they all failed.', 'start': 11696.646, 'duration': 1.801}, {'end': 11700.367, 'text': "It's not interesting.", 'start': 11699.407, 'duration': 0.96}, {'end': 11705.449, 'text': 'Why try it in the first? They have a perfect explanation, the best possible.', 'start': 11700.908, 'duration': 4.541}, {'end': 11708.05, 'text': 'for some fundamental mental property.', 'start': 11706.149, 'duration': 1.901}, {'end': 11716.996, 'text': "What's the point of trying to see if a couple of supercomputers and massive amounts of data can approximate it? I mean, it's madness, you know.", 'start': 11708.511, 'duration': 8.485}, {'end': 11719.318, 'text': "But that's the field that we're in.", 'start': 11717.857, 'duration': 1.461}], 'summary': 'Progress in linguistics and cognitive science may lead to genuine explanations for fundamental language properties, such as structure dependence, based on simple combinatorial operations, despite limited interest in the field.', 'duration': 138.019, 'max_score': 11581.299, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q11581299.jpg'}, {'end': 11676.98, 'src': 'embed', 'start': 11647.355, 'weight': 8, 'content': [{'end': 11651.319, 'text': "Turns out that that's what follows from the simplest combinatorial operation.", 'start': 11647.355, 'duration': 3.964}, {'end': 11659.768, 'text': "The simplest combinatorial operation happens to be binary set formation, what's called merge in contemporary literature.", 'start': 11651.339, 'duration': 8.429}, {'end': 11667.093, 'text': 'Well, if language is based on binary set formation, you get this property, no linear largest structures.', 'start': 11660.408, 'duration': 6.685}, {'end': 11676.98, 'text': 'So we had a, for the first time ever, a deep explanation for the most fundamental property of language, which is a very surprising property,', 'start': 11667.953, 'duration': 9.027}], 'summary': 'Binary set formation in language leads to no linear largest structures.', 'duration': 29.625, 'max_score': 11647.355, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q11647355.jpg'}, {'end': 11808.001, 'src': 'embed', 'start': 11780.663, 'weight': 9, 'content': [{'end': 11788.928, 'text': 'Which profound misunderstandings of language and linguistics persist even at the highest levels of the scientific community?', 'start': 11780.663, 'duration': 8.265}, {'end': 11794.091, 'text': 'Do you mind that many of your own scientific ideas are widely misunderstood?', 'start': 11789.428, 'duration': 4.663}, {'end': 11795.932, 'text': 'Totally misunderstood.', 'start': 11794.771, 'duration': 1.161}, {'end': 11797.412, 'text': "It's amazing to me.", 'start': 11796.492, 'duration': 0.92}, {'end': 11799.454, 'text': "So there's one paper of mine.", 'start': 11797.432, 'duration': 2.022}, {'end': 11801.575, 'text': 'that I presume you know about.', 'start': 11800.254, 'duration': 1.321}, {'end': 11808.001, 'text': 'It was in 1956, three models for the description of language that did enter the literature.', 'start': 11802.236, 'duration': 5.765}], 'summary': 'Profound misunderstandings of language persist in scientific community, including one paper from 1956.', 'duration': 27.338, 'max_score': 11780.663, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q11780663.jpg'}, {'end': 11991.488, 'src': 'embed', 'start': 11963.456, 'weight': 10, 'content': [{'end': 11965.537, 'text': 'which holds not just for computer science.', 'start': 11963.456, 'duration': 2.081}, {'end': 11970.319, 'text': 'It holds for philosophy, linguistics, total misunderstanding.', 'start': 11966.217, 'duration': 4.102}, {'end': 11975.382, 'text': "I think probably for our experiential reasons we're instinctively radical and persist.", 'start': 11971, 'duration': 4.382}, {'end': 11977.563, 'text': "And it's hard to get out of it.", 'start': 11976.182, 'duration': 1.381}, {'end': 11982.065, 'text': 'It took the 17th and 18th century for physics to break out of it.', 'start': 11977.583, 'duration': 4.482}, {'end': 11985.726, 'text': "Another problem is what's sometimes called Darwin's problem.", 'start': 11982.765, 'duration': 2.961}, {'end': 11989.728, 'text': "How do we get this language system? It's unique.", 'start': 11986.306, 'duration': 3.422}, {'end': 11991.488, 'text': "It's common to humans.", 'start': 11990.388, 'duration': 1.1}], 'summary': 'Challenges in various fields, language uniqueness, human commonality.', 'duration': 28.032, 'max_score': 11963.456, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q11963456.jpg'}], 'start': 11044.971, 'title': 'Challenges in natural language semantics, language of thought, and language evolution', 'summary': "Discusses challenges in natural language semantics, highlights absence of formal semantics, concept of event semantics, and its relation to pure syntax. it also explores the relationship between universal grammar and the language of thought, and discusses nature's simplicity in language evolution with emphasis on the three stages of evolution, recursive enumeration in homo sapiens, and fundamental properties of language.", 'chapters': [{'end': 11171.96, 'start': 11044.971, 'title': 'Challenges in natural language semantics', 'summary': 'Discusses the challenges in natural language semantics, highlighting the absence of formal semantics, the concept of event semantics, and its relation to pure syntax, emphasizing the mental construction of events and the productivity of event semantics as a form of syntax.', 'duration': 126.989, 'highlights': ['Event semantics, such as the Davidsonian approach, addresses the absence of formal semantics in natural language, providing a productive framework for understanding language structure and inferences.', "The proposed answer to the question of why we can infer 'John read the book' from 'John read the book quickly' involves analyzing the event reading, the agent, the book, and the adverb as a concatenation, illustrating the productivity of event semantics in making inferences.", 'The concept of event semantics aligns with the understanding of pure syntax and provides natural structures for language, demonstrating its relevance to language comprehension and interpretation.', "Zeno's concept of constructing events in the mind emphasizes the mental nature of events, highlighting the subjective and constructivist aspect of event semantics in language understanding."]}, {'end': 11452.759, 'start': 11172.781, 'title': 'Language of thought and universal grammar', 'summary': 'Discusses the relationship between universal grammar and the language of thought, emphasizing the debate about the existence of universal grammar and the inborn nature of human language faculty.', 'duration': 279.978, 'highlights': ['There is a debate in the cognitive science literature about whether universal grammar exists, illustrating the pre-scientific character of cognitive science.', 'The inborn nature of human language faculty is emphasized, with the capacity for any infant to acquire any language with equal facility.', 'The chapter delves into the relationship between universal grammar and the language of thought, emphasizing the debate about their similarity and differences.', 'Humans began to separate on the order of 150,000 years ago, sharing the language faculty equally, with no new evidence of its existence before modern Homo sapiens.']}, {'end': 12031.772, 'start': 11454.14, 'title': "Nature's simplicity in language evolution", 'summary': 'Discusses the three stages of evolution and how mother nature finds the simplest solutions, emphasizing on the capacity for recursive enumeration in homo sapiens and the fundamental properties of language, providing a deep explanation for structure dependence and the misunderstandings in the scientific community.', 'duration': 577.632, 'highlights': ['Mother Nature finds the simplest solutions in evolution, supported by overwhelming evidence, emphasizing the importance of simplicity in scientific explanation. Mother Nature operates by finding the simplest solutions in evolution, supported by overwhelming evidence, emphasizing the importance of simplicity in scientific explanation.', 'The capacity for recursive enumeration, a unique property in Homo sapiens, is a fundamental aspect of language evolution, with barely any interest from researchers in linguistics and cognitive science. The capacity for recursive enumeration, a unique property in Homo sapiens, is a fundamental aspect of language evolution, with barely any interest from researchers in linguistics and cognitive science.', 'The explanation for the structure dependence in language lies in the simplest combinatorial operation of binary set formation, providing a deep understanding of this fundamental property. The explanation for the structure dependence in language lies in the simplest combinatorial operation of binary set formation, providing a deep understanding of this fundamental property.', "Misunderstandings persist at the highest levels of the scientific community, including the misinterpretation of Chomsky's paper on three models for the description of language, and the failure to recognize the third model's significance. Misunderstandings persist at the highest levels of the scientific community, including the misinterpretation of Chomsky's paper on three models for the description of language, and the failure to recognize the third model's significance.", "The Plato problem and Darwin's problem are major challenges in understanding language evolution, with common misunderstandings prevailing in various fields such as computer science, philosophy, and linguistics. The Plato problem and Darwin's problem are major challenges in understanding language evolution, with common misunderstandings prevailing in various fields such as computer science, philosophy, and linguistics."]}], 'duration': 986.801, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q11044971.jpg', 'highlights': ['Event semantics addresses the absence of formal semantics in natural language, providing a productive framework for understanding language structure and inferences.', 'The concept of event semantics aligns with the understanding of pure syntax and provides natural structures for language, demonstrating its relevance to language comprehension and interpretation.', "The proposed answer to the question of why we can infer 'John read the book' from 'John read the book quickly' involves analyzing the event reading, the agent, the book, and the adverb as a concatenation, illustrating the productivity of event semantics in making inferences.", "Zeno's concept of constructing events in the mind emphasizes the mental nature of events, highlighting the subjective and constructivist aspect of event semantics in language understanding.", 'The inborn nature of human language faculty is emphasized, with the capacity for any infant to acquire any language with equal facility.', 'The chapter delves into the relationship between universal grammar and the language of thought, emphasizing the debate about their similarity and differences.', 'Mother Nature finds the simplest solutions in evolution, supported by overwhelming evidence, emphasizing the importance of simplicity in scientific explanation.', 'The capacity for recursive enumeration, a unique property in Homo sapiens, is a fundamental aspect of language evolution, with barely any interest from researchers in linguistics and cognitive science.', 'The explanation for the structure dependence in language lies in the simplest combinatorial operation of binary set formation, providing a deep understanding of this fundamental property.', "Misunderstandings persist at the highest levels of the scientific community, including the misinterpretation of Chomsky's paper on three models for the description of language, and the failure to recognize the third model's significance.", "The Plato problem and Darwin's problem are major challenges in understanding language evolution, with common misunderstandings prevailing in various fields such as computer science, philosophy, and linguistics."]}, {'end': 13013.411, 'segs': [{'end': 12060.202, 'src': 'embed', 'start': 12032.412, 'weight': 2, 'content': [{'end': 12036.455, 'text': "That's part of the overwhelming irrationality of the way the human mind is studied.", 'start': 12032.412, 'duration': 4.043}, {'end': 12038.317, 'text': "That's Darwin's problem.", 'start': 12037.096, 'duration': 1.221}, {'end': 12042.353, 'text': "Now, the last problem is what's now called Descartes' problem.", 'start': 12039.372, 'duration': 2.981}, {'end': 12044.795, 'text': "How can we do what we're doing right now?", 'start': 12042.954, 'duration': 1.841}, {'end': 12050.357, 'text': 'How can we speak in ways which are appropriate to situations but not caused by them?', 'start': 12045.595, 'duration': 4.762}, {'end': 12055.68, 'text': "That's a huge problem, which is basically the question of freedom of choice.", 'start': 12051.458, 'duration': 4.222}, {'end': 12060.202, 'text': "How can we do it, you see? This is, we're involved in it all our lives.", 'start': 12056.26, 'duration': 3.942}], 'summary': "Studying human mind's irrationality, darwin's and descartes' problems, and freedom of choice.", 'duration': 27.79, 'max_score': 12032.412, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q12032412.jpg'}, {'end': 12119.47, 'src': 'embed', 'start': 12076.585, 'weight': 0, 'content': [{'end': 12082.267, 'text': "You look at the behavior of the same people, 100% of the time they act as if they're not determined.", 'start': 12076.585, 'duration': 5.682}, {'end': 12084.888, 'text': "So, and it's this way constantly.", 'start': 12082.987, 'duration': 1.901}, {'end': 12090.011, 'text': "And even the people who give arguments saying determined are tacitly assuming we're not determined.", 'start': 12085.349, 'duration': 4.662}, {'end': 12096.655, 'text': "Otherwise, why give the argument? You know if it's determined, if it's a thermostat, why bother?", 'start': 12090.732, 'duration': 5.923}, {'end': 12102.479, 'text': "So it goes all the way up to Einstein, who gave arguments trying to show that everything's determined,", 'start': 12097.176, 'duration': 5.303}, {'end': 12105.16, 'text': 'proving by his effort to do it that he was not determined.', 'start': 12102.479, 'duration': 2.681}, {'end': 12108.222, 'text': 'Okay, so here we have a kind of a paradox.', 'start': 12105.721, 'duration': 2.501}, {'end': 12110.504, 'text': "Everybody says we're automatic.", 'start': 12108.943, 'duration': 1.561}, {'end': 12113.786, 'text': "Everybody acts all the time as if we're not automated.", 'start': 12111.144, 'duration': 2.642}, {'end': 12117.088, 'text': 'Well, does science say anything about this? Nothing.', 'start': 12114.286, 'duration': 2.802}, {'end': 12119.47, 'text': "Science says we can't deal with it.", 'start': 12117.809, 'duration': 1.661}], 'summary': "People act as if not determined, contradicting arguments for determinism. science can't address this paradox.", 'duration': 42.885, 'max_score': 12076.585, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q12076585.jpg'}, {'end': 12197.536, 'src': 'embed', 'start': 12169.046, 'weight': 3, 'content': [{'end': 12174.207, 'text': "If you look at the philosophical literature today, there's something that everyone's obsessed with.", 'start': 12169.046, 'duration': 5.161}, {'end': 12176.288, 'text': "It's called the hard problem.", 'start': 12174.947, 'duration': 1.341}, {'end': 12178.368, 'text': 'The problem is consciousness.', 'start': 12176.908, 'duration': 1.46}, {'end': 12185.63, 'text': "What's it like to see the sun rise? Let's go back to the 17th century, which is an interesting century.", 'start': 12178.948, 'duration': 6.682}, {'end': 12187.21, 'text': "It's the birth of modern science.", 'start': 12185.69, 'duration': 1.52}, {'end': 12191.531, 'text': 'They had a hard problem, as it was called, the motion.', 'start': 12187.871, 'duration': 3.66}, {'end': 12193.152, 'text': 'The hard problem was motion.', 'start': 12191.772, 'duration': 1.38}, {'end': 12197.536, 'text': 'That was a hard problem dealt with in the 17th century.', 'start': 12194.213, 'duration': 3.323}], 'summary': "Philosophers are obsessed with the hard problem of consciousness, akin to the 17th century's hard problem of motion.", 'duration': 28.49, 'max_score': 12169.046, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q12169046.jpg'}, {'end': 12388.012, 'src': 'embed', 'start': 12359.344, 'weight': 4, 'content': [{'end': 12362.646, 'text': 'Well, I think there are questions right at the border of research.', 'start': 12359.344, 'duration': 3.302}, {'end': 12367.107, 'text': 'which I think are very exciting, like the one I mentioned.', 'start': 12363.866, 'duration': 3.241}, {'end': 12372.568, 'text': 'I mentioned that I think for the first time ever for thousands of years,', 'start': 12367.887, 'duration': 4.681}, {'end': 12379.17, 'text': 'we can now give deep explanations of some fundamental properties of language and thought which are probably the same thing.', 'start': 12372.568, 'duration': 6.602}, {'end': 12383.551, 'text': "Let's assume that language and thought are two different ways of looking at the same thing.", 'start': 12379.65, 'duration': 3.901}, {'end': 12388.012, 'text': "Thought's what is generated by language, and language is what generates thought.", 'start': 12384.171, 'duration': 3.841}], 'summary': 'Exciting research reveals deep explanations of language and thought.', 'duration': 28.668, 'max_score': 12359.344, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q12359344.jpg'}, {'end': 12665.576, 'src': 'embed', 'start': 12638.684, 'weight': 5, 'content': [{'end': 12642.745, 'text': 'Well, among the problems of language are what the neural basis?', 'start': 12638.684, 'duration': 4.061}, {'end': 12644.585, 'text': "what's the neural basis for language?", 'start': 12642.745, 'duration': 1.84}, {'end': 12648.086, 'text': "I mean something's going on in our brains when we're doing this.", 'start': 12645.245, 'duration': 2.841}, {'end': 12653.399, 'text': "What's going on? It's very hard to find out, partly for ethical reasons.", 'start': 12648.908, 'duration': 4.491}, {'end': 12656.767, 'text': "You just can't do the experiments that might give you some answers.", 'start': 12653.92, 'duration': 2.847}, {'end': 12659.191, 'text': 'because they were unethical.', 'start': 12657.95, 'duration': 1.241}, {'end': 12665.576, 'text': "You can't stick an electrode into a particular neuron in Broca's area to find out what's going on.", 'start': 12659.571, 'duration': 6.005}], 'summary': 'Challenges in studying neural basis of language due to ethical constraints.', 'duration': 26.892, 'max_score': 12638.684, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q12638684.jpg'}, {'end': 12779.638, 'src': 'embed', 'start': 12755.206, 'weight': 7, 'content': [{'end': 12761.873, 'text': "That's the kind of topic you can investigate from the outside, and any area of science you know about, you've got many questions like this.", 'start': 12755.206, 'duration': 6.667}, {'end': 12764.456, 'text': "So there's no shortage of things to study.", 'start': 12762.434, 'duration': 2.022}, {'end': 12772.124, 'text': "It's just a shame that huge amount of effort, money, scarce energy are wasted in doing things that make absolutely no sense.", 'start': 12764.716, 'duration': 7.408}, {'end': 12779.638, 'text': "Well, Professor Chomsky, we're going to wrap the conversation here, but can I just thank you so much for coming on our podcast?", 'start': 12772.932, 'duration': 6.706}], 'summary': 'Unlimited research opportunities exist in science, but resources are wasted on senseless activities.', 'duration': 24.432, 'max_score': 12755.206, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q12755206.jpg'}, {'end': 12855.545, 'src': 'embed', 'start': 12819.484, 'weight': 6, 'content': [{'end': 12821.765, 'text': 'Right And so when I realized how.', 'start': 12819.484, 'duration': 2.281}, {'end': 12832.51, 'text': 'down to earth he is and, and, and somebody that you could just hang out with and joke, joke about and talk about things concretely in the real world.', 'start': 12823.41, 'duration': 9.1}, {'end': 12833.351, 'text': "And yet he's.", 'start': 12832.71, 'duration': 0.641}, {'end': 12839.352, 'text': 'you know, the foremost intellectual of our era.', 'start': 12835.049, 'duration': 4.303}, {'end': 12842.115, 'text': 'Right I think, um, it, it was nice.', 'start': 12839.653, 'duration': 2.462}, {'end': 12845.698, 'text': 'It was nice to see that, like you, can be both at the same time.', 'start': 12842.215, 'duration': 3.483}, {'end': 12846.718, 'text': 'you know you can be.', 'start': 12845.698, 'duration': 1.02}, {'end': 12855.545, 'text': 'you can be an ordinary human being and an extraordinary intellectual, all rolled into one.', 'start': 12846.718, 'duration': 8.827}], 'summary': 'Realizing the down-to-earth nature of a prominent intellectual and the ability to be both ordinary and extraordinary at the same time.', 'duration': 36.061, 'max_score': 12819.484, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q12819484.jpg'}], 'start': 12032.412, 'title': 'Understanding philosophy and science', 'summary': "Delves into the paradox of determinism, misunderstandings in philosophy and science, and the study of mind and language. it also reflects on chomsky's views on science, offering deep insights into these subjects.", 'chapters': [{'end': 12145.468, 'start': 12032.412, 'title': 'The paradox of determinism', 'summary': "Discusses darwin's problem, descartes' problem, and the paradox of determinism, highlighting the contradiction between the belief in determinism and the behavior of individuals acting as if they have freedom of choice.", 'duration': 113.056, 'highlights': ["The paradox of determinism is evident in the contradiction between people's belief in determinism and their behavior, with 100% of individuals acting as if they have freedom of choice despite arguing for determinism.", 'Science is unable to address the paradox of determinism, as it can deal with determinism and randomness but not with things that are not determined or random.', "Einstein's arguments for determinism paradoxically demonstrate his own freedom of choice, indicating the inherent contradiction in individuals advocating for determinism while acting as if they are not automated.", "The chapter explores Darwin's problem and Descartes' problem, addressing the irrationality in studying the human mind and the question of freedom of choice in human actions."]}, {'end': 12468.952, 'start': 12146.269, 'title': 'Misunderstandings in philosophy and science', 'summary': 'Discusses the profound misunderstandings dominating philosophy and science, focusing on the hard problem of consciousness and the irrationality in addressing unanswerable questions, while highlighting the ability to provide deep explanations for fundamental properties of language and thought.', 'duration': 322.683, 'highlights': ['The hard problem of consciousness poses an unanswerable question, leading to a total obsession in philosophy of mind, reminiscent of a strange and irrational period in history. The chapter emphasizes the unanswerable nature of the hard problem of consciousness and criticizes the total obsession in philosophy of mind, reflecting a strange and irrational period in history.', 'The ability to give deep explanations of fundamental properties of language and thought, particularly the property of structure dependence, is considered as one of the most exciting areas of research. The chapter highlights the ability to provide deep explanations for fundamental properties of language and thought, focusing on the exciting area of research involving structure dependence.', 'The chapter mentions the ability to account for the property of structure dependence by assuming that nature acts in the simplest way, leading to a perfect answer for this fundamental property. The chapter discusses the ability to account for the property of structure dependence by assuming that nature acts in the simplest way, providing a perfect answer for this fundamental property.']}, {'end': 12754.925, 'start': 12468.972, 'title': 'The study of mind and language', 'summary': 'Explores the exciting prospects of understanding the mind and language, delving into the mysteries of consciousness and the neural basis of language, while highlighting the challenges of ethical limitations on experimentation in this field.', 'duration': 285.953, 'highlights': ['The mysteries of consciousness and the neural basis of language are explored, with a focus on the ethical limitations of experimentation in this field. Exploration of consciousness and neural basis of language, ethical limitations on experimentation.', "The discussion of the unknown nature of matter and the question 'What is a particle?' is highlighted, demonstrating the vast gaps in understanding in physics. Unknown nature of matter, question of 'What is a particle?', gaps in understanding in physics.", 'The challenges of studying the neural basis for language due to ethical reasons are emphasized, with a specific example of limitations on experiments with humans. Challenges of studying neural basis for language, limitations on experiments with humans.']}, {'end': 13013.411, 'start': 12755.206, 'title': "Chomsky's views on science", 'summary': 'Reflects on the conversation with chomsky, highlighting his down-to-earth nature while being the foremost intellectual, his perspective on the wastage of effort and resources in nonsensical pursuits, and the technical challenges faced in producing and restoring the show.', 'duration': 258.205, 'highlights': ["Chomsky's down-to-earth nature and ability to engage in concrete discussions despite being the foremost intellectual Chomsky's ability to discuss concrete and down-to-earth topics while being the foremost intellectual of our era, demonstrating the dual nature of being an ordinary human being and an extraordinary intellectual.", "Chomsky's perspective on the wastage of effort and resources in nonsensical pursuits Chomsky's viewpoint on the huge amount of effort, money, and scarce energy being wasted in pursuits that make absolutely no sense, reflecting on the inefficiency in certain scientific endeavors.", 'Technical challenges faced in producing and restoring the show The extensive production cycle and technological failure resulting in an epic and satisfying journey to restore the show, involving a tremendous amount of effort and technical challenges.']}], 'duration': 980.999, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/axuGfh4UR9Q/pics/axuGfh4UR9Q12032412.jpg', 'highlights': ['100% of individuals act as if they have freedom of choice despite arguing for determinism.', 'Science is unable to address the paradox of determinism.', "The chapter explores Darwin's problem and Descartes' problem, addressing the irrationality in studying the human mind and the question of freedom of choice in human actions.", 'The hard problem of consciousness poses an unanswerable question, leading to a total obsession in philosophy of mind.', 'The ability to provide deep explanations for fundamental properties of language and thought is considered as one of the most exciting areas of research.', 'The mysteries of consciousness and the neural basis of language are explored, with a focus on the ethical limitations of experimentation in this field.', "Chomsky's ability to discuss concrete and down-to-earth topics while being the foremost intellectual of our era.", "Chomsky's viewpoint on the huge amount of effort, money, and scarce energy being wasted in pursuits that make absolutely no sense, reflecting on the inefficiency in certain scientific endeavors."]}], 'highlights': ['The emotional significance of interviewing Noam Chomsky at Lake Como in Italy is highlighted as a dream come true, adding a personal touch to the episode.', "Interviewing Professor Noam Chomsky, the most important intellectual of the 20th century, is an incredible honor and a significant event in the podcast's history.", 'Lacoon emphasizes the importance of learning in creating intelligent systems and critiques current AI approaches, stating that our best systems still fall short of matching human reliability in real-world tasks, despite being fed with large amounts of data and having behaviors hardwired into the model.', 'The ability to take any two objects and select from an infinite set of possible abstractions is the core of human cognition.', 'Lacoon criticizes supervised learning for its limitations, advocating for the need to embrace uncertainty in machine learning.', 'The presence of normalization is an integral aspect of Bayesian probability and is acknowledged by practitioners.', 'The critique of energy-based models and the comparison to arithmetic with fractions underscores the importance of addressing normalization and embracing fundamental principles in modeling.', 'The critique of deep learning and the rise of connectionism', 'The human brain might utilize quantum properties to access non-computable oracles', 'The mysteries of consciousness and the neural basis of language are explored, with a focus on the ethical limitations of experimentation in this field.']}