title
KARL FRISTON - INTELLIGENCE 3.0

description
This special edition show is sponsored by Numerai, please visit them here with our sponsor link (we would really appreciate it) http://numer.ai/mlst Prof. Karl Friston recently proposed a vision of artificial intelligence that goes beyond machines and algorithms, and embraces humans and nature as part of a cyber-physical ecosystem of intelligence. This vision is based on the principle of active inference, which states that intelligent systems can learn from their observations and act on their environment to reduce uncertainty and achieve their goals. This leads to a formal account of collective intelligence that rests on shared narratives and goals. To realize this vision, Friston suggests developing a shared hyper-spatial modelling language and transaction protocol, as well as novel methods for measuring and optimizing collective intelligence. This could harness the power of artificial intelligence for the common good, without compromising human dignity or autonomy. It also challenges us to rethink our relationship with technology, nature, and each other, and invites us to join a global community of sense-makers who are curious about the world and eager to improve it. Pod version: https://podcasters.spotify.com/pod/show/machinelearningstreettalk/episodes/106---Prof--KARL-FRISTON-3-0---Collective-Intelligence-Special-Edition-e208f50 Support us! https://www.patreon.com/mlst MLST Discord: https://discord.gg/aNPkGUQtc5 TOC: Intro [00:00:00] Numerai (Sponsor segment) [00:07:10] Designing Ecosystems of Intelligence from First Principles (Friston et al) [00:09:48] Information / Infosphere and human agency [00:18:30] Intelligence [00:31:38] Reductionism [00:39:36] Universalism [00:44:46] Emergence [00:54:23] Markov blankets [01:02:11] Whole part relationships / structure learning [01:22:33] Enactivism [01:29:23] Knowledge and Language [01:43:53] ChatGPT [01:50:56] Ethics (is-ought) [02:07:55] Can people be evil? [02:35:06] Ethics in Al, subjectiveness [02:39:05] Final thoughts [02:57:00] References: Designing Ecosystems of Intelligence from First Principles (Friston et al) https://arxiv.org/abs/2212.01354 GLOM - How to represent part-whole hierarchies in a neural network (Hinton) https://arxiv.org/pdf/2102.12627.pdf Seven Brief Lessons on Physics (Carlo Rovelli) https://www.amazon.co.uk/Seven-Brief-Lessons-Physics-Rovelli/dp/0141981725 How Emotions Are Made: The Secret Life of the Brain (Lisa Feldman Barrett) https://www.amazon.co.uk/How-Emotions-Are-Made-Secret/dp/B01N3D4OON Am I Self-Conscious? (Or Does Self-Organization Entail Self-Consciousness?) (Karl Friston) https://www.frontiersin.org/articles/10.3389/fpsyg.2018.00579/full Integrated information theory (Giulio Tononi) https://en.wikipedia.org/wiki/Integrated_information_theory

detail
{'title': 'KARL FRISTON - INTELLIGENCE 3.0', 'heatmap': [], 'summary': "The video, 'karl friston - intelligence 3.0,' discusses creating a cyber-physical ecosystem of intelligence based on physics and biology, exploring active inference as a physics of intelligence, existential imperatives in industry, information impact on humanity, understanding intelligence and agency, temporal scales and markov blankets, evolution of machine learning architectures, impact of chat gpt and generative ai, morality, values, society, ai ethics, and sentient behavior.", 'chapters': [{'end': 572.178, 'segs': [{'end': 44.556, 'src': 'embed', 'start': 0.109, 'weight': 0, 'content': [{'end': 7.274, 'text': "Hello everyone, I'm here to talk to you about a vision of artificial intelligence that goes beyond machines and algorithms.", 'start': 0.109, 'duration': 7.165}, {'end': 15.5, 'text': 'A vision which embraces humans and nature as integral parts of a cyber-physical ecosystem of intelligence.', 'start': 7.835, 'duration': 7.665}, {'end': 20.224, 'text': 'A vision that is based on first principles derived from physics and biology.', 'start': 16.161, 'duration': 4.063}, {'end': 26.789, 'text': 'Professor Carl Friston just dropped a paper called Designing Ecosystems of Intelligence from First Principles.', 'start': 21.084, 'duration': 5.705}, {'end': 36.033, 'text': 'One of the principles is active inference, a formulation of adaptive behavior which can be read as a physics of intelligence.', 'start': 27.689, 'duration': 8.344}, {'end': 44.556, 'text': 'Active inference says that intelligent systems are those which can accumulate evidence for a generative model of their sensed world.', 'start': 36.893, 'duration': 7.663}], 'summary': 'Vision of ai integrates humans and nature, based on physics and biology principles.', 'duration': 44.447, 'max_score': 0.109, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/V_VXOdf1NMw/pics/V_VXOdf1NMw109.jpg'}, {'end': 113.705, 'src': 'embed', 'start': 66.655, 'weight': 2, 'content': [{'end': 72.759, 'text': 'This leads to a formal account of collective intelligence that rests on shared narratives and goals.', 'start': 66.655, 'duration': 6.104}, {'end': 75.581, 'text': 'So how do we realize this vision?', 'start': 74.12, 'duration': 1.461}, {'end': 85.508, 'text': 'Friston proposes a research agenda for the next decade and beyond which aims to design such ecosystems of intelligence from scratch.', 'start': 76.242, 'duration': 9.266}, {'end': 92.072, 'text': 'They suggest developing a shared hyperspatial modeling language and transaction protocols,', 'start': 86.609, 'duration': 5.463}, {'end': 95.975, 'text': 'as well as novel methods for measuring and optimizing collective intelligence.', 'start': 92.072, 'duration': 3.903}, {'end': 98.576, 'text': 'So why should we care about this vision?', 'start': 97.116, 'duration': 1.46}, {'end': 108.062, 'text': "It's because it offers a way to harness the power of artificial intelligence for the common good without compromising human dignity or autonomy.", 'start': 99.117, 'duration': 8.945}, {'end': 113.705, 'text': 'And because it challenges us to rethink our relationship with technology, nature, and each other.', 'start': 108.622, 'duration': 5.083}], 'summary': 'Friston proposes a research agenda to design ecosystems of intelligence, aiming to harness ai for the common good.', 'duration': 47.05, 'max_score': 66.655, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/V_VXOdf1NMw/pics/V_VXOdf1NMw66655.jpg'}, {'end': 212.948, 'src': 'embed', 'start': 178.624, 'weight': 4, 'content': [{'end': 187.728, 'text': 'It has the potential to reshape how we view the connection between inanimate matter and living things,', 'start': 178.624, 'duration': 9.104}, {'end': 197.952, 'text': 'and even to answer the question of how and why consciousness and intelligence might emerge from physical matter and processes.', 'start': 187.728, 'duration': 10.224}, {'end': 212.948, 'text': 'That means a world where all things from particles to people to the largest systems all move or evolve according to two processes which combine.', 'start': 199.593, 'duration': 13.355}], 'summary': 'Potential to reshape connection between inanimate matter and living things, and answer the question of how consciousness and intelligence might emerge.', 'duration': 34.324, 'max_score': 178.624, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/V_VXOdf1NMw/pics/V_VXOdf1NMw178624.jpg'}], 'start': 0.109, 'title': 'Uniting humans and nature with active inference', 'summary': 'Discusses creating a cyber-physical ecosystem of intelligence based on physics and biology, aiming to harness ai for the common good. it presents active inference as a physics of intelligence, explains the formal account of collective intelligence, and proposes a research agenda for designing ecosystems of intelligence, offering a way to reshape our relationship with technology, nature, and each other.', 'chapters': [{'end': 288.665, 'start': 0.109, 'title': 'Ecosystems of intelligence: uniting humans and nature', 'summary': 'Discusses the vision of a cyber-physical ecosystem of intelligence, based on principles from physics and biology, aiming to harness the power of artificial intelligence for the common good without compromising human dignity or autonomy. it presents active inference as a physics of intelligence, explains the formal account of collective intelligence, and proposes a research agenda for designing ecosystems of intelligence from scratch, offering a way to reshape our relationship with technology, nature, and each other.', 'duration': 288.556, 'highlights': ['The vision of a cyber-physical ecosystem of intelligence embraces humans and nature and is based on first principles derived from physics and biology. Emphasizes the fundamental principles and the holistic approach of the envisioned ecosystem.', 'Active inference is presented as a physics of intelligence, where intelligent systems accumulate evidence for a generative model of their sensed world to reduce uncertainty and achieve their goals. Highlights the concept of active inference as a crucial component of the envisioned ecosystem, explaining its role in intelligent systems.', 'The proposal of a research agenda for designing ecosystems of intelligence from scratch involves developing shared hyperspatial modeling language, transaction protocols, and novel methods for measuring and optimizing collective intelligence. Outlines the practical steps and tools suggested for realizing the vision of ecosystems of intelligence, emphasizing the need for collaborative and innovative approaches.', 'The vision aims to harness the power of artificial intelligence for the common good without compromising human dignity or autonomy, challenging individuals to rethink their relationship with technology, nature, and each other. Underlines the broader societal implications and ethical considerations associated with the envisioned ecosystem, urging a shift in perspectives and behaviors.', 'The potential of the free energy principle to reshape views on the connection between inanimate matter and living things, and to provide insights into the emergence of consciousness and intelligence from physical matter and processes, is discussed. Highlights the transformative implications of the free energy principle, particularly in understanding the connection between inanimate matter and living entities.']}, {'end': 572.178, 'start': 290.446, 'title': 'Active inference and existence', 'summary': 'Discusses the concept of active inference and its implications, such as the dynamics of systems maintaining coherence and identity, the application of the free energy principle at all scales of size and time, and the sponsorship message from numeri offering a data science competition platform for predicting the stock market.', 'duration': 281.732, 'highlights': ['The dynamics of systems maintaining coherence and identity are based on the concept of a Markov blanket and a pullback attractor, ensuring integrity and coherence over time.', 'The free energy principle applies at all scales of size and time, leading to an ecosystem of things interacting across scales, potentially providing insights into emergence and consciousness.', 'Numeri sponsors a data science competition platform for predicting the stock market, offering over 50 million dollars for 5,000 models and providing back testing data for fine-tuning and testing models.']}], 'duration': 572.069, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/V_VXOdf1NMw/pics/V_VXOdf1NMw109.jpg', 'highlights': ['The vision of a cyber-physical ecosystem of intelligence embraces humans and nature and is based on first principles derived from physics and biology.', 'Active inference is presented as a physics of intelligence, where intelligent systems accumulate evidence for a generative model of their sensed world to reduce uncertainty and achieve their goals.', 'The proposal of a research agenda for designing ecosystems of intelligence from scratch involves developing shared hyperspatial modeling language, transaction protocols, and novel methods for measuring and optimizing collective intelligence.', 'The vision aims to harness the power of artificial intelligence for the common good without compromising human dignity or autonomy, challenging individuals to rethink their relationship with technology, nature, and each other.', 'The potential of the free energy principle to reshape views on the connection between inanimate matter and living things, and to provide insights into the emergence of consciousness and intelligence from physical matter and processes, is discussed.']}, {'end': 1182.474, 'segs': [{'end': 676.195, 'src': 'embed', 'start': 640.675, 'weight': 0, 'content': [{'end': 644.576, 'text': "So he's very gutted that he couldn't come today to be part of this.", 'start': 640.675, 'duration': 3.901}, {'end': 646.096, 'text': "It's very gracious of him.", 'start': 644.796, 'duration': 1.3}, {'end': 653.299, 'text': "So you just wrote a paper, and it's called Designing Ecosystems of Intelligence from First Principles.", 'start': 646.977, 'duration': 6.322}, {'end': 662.246, 'text': 'You led with this white paper lays out a vision of research and development in the field of artificial intelligence for the next decade and beyond.', 'start': 654.161, 'duration': 8.085}, {'end': 663.767, 'text': "It's denouement.", 'start': 662.946, 'duration': 0.821}, {'end': 665.408, 'text': "I wasn't sure if I was going to say that word.", 'start': 663.787, 'duration': 1.621}, {'end': 676.195, 'text': 'is a cyber-physical ecosystem of natural and synthetic sense-making in which humans are integral participants, what we call shared intelligence.', 'start': 666.764, 'duration': 9.431}], 'summary': 'Paper outlines vision for ai research and development, focusing on shared intelligence.', 'duration': 35.52, 'max_score': 640.675, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/V_VXOdf1NMw/pics/V_VXOdf1NMw640675.jpg'}, {'end': 733.535, 'src': 'embed', 'start': 709.651, 'weight': 1, 'content': [{'end': 715.772, 'text': 'and crucially, you said, active inference foregrounds an existential imperative of intelligent systems,', 'start': 709.651, 'duration': 6.121}, {'end': 719.893, 'text': 'namely curiosity or the resolution of uncertainty.', 'start': 715.772, 'duration': 4.121}, {'end': 726.094, 'text': 'and the same imperative underwrites belief sharing in ensembles of agents in which certain aspects,', 'start': 719.893, 'duration': 6.201}, {'end': 733.535, 'text': "which is to say factors of each agent's generative model, provide common ground or a frame of reference.", 'start': 726.094, 'duration': 7.441}], 'summary': 'Active inference emphasizes curiosity and belief sharing in ensembles of agents.', 'duration': 23.884, 'max_score': 709.651, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/V_VXOdf1NMw/pics/V_VXOdf1NMw709651.jpg'}, {'end': 935.147, 'src': 'embed', 'start': 903.632, 'weight': 2, 'content': [{'end': 911.738, 'text': 'Often I hear underwriting the move from AI artificial intelligence to IA,', 'start': 903.632, 'duration': 8.106}, {'end': 922.582, 'text': "where it's intelligent agents the artifacts with agency that do have behaviors that have a sentient kind of behavior or at least an intelligent kind of behavior.", 'start': 911.738, 'duration': 10.844}, {'end': 924.923, 'text': 'Close brackets.', 'start': 923.422, 'duration': 1.501}, {'end': 935.147, 'text': "So we're now moving from thinking about a single thing, a particle or a person or a computer,", 'start': 925.363, 'duration': 9.784}], 'summary': 'Transitioning from ai to ia, focusing on intelligent agents with sentient or intelligent behavior.', 'duration': 31.515, 'max_score': 903.632, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/V_VXOdf1NMw/pics/V_VXOdf1NMw903632.jpg'}], 'start': 572.718, 'title': 'Designing intelligent ecosystems and existential imperatives in industry', 'summary': 'Discusses the concept of active inference in designing ecosystems of intelligence, emphasizing the existential imperative of intelligent systems and the vision of research and development in the field of artificial intelligence for the next decade and beyond. it also examines the existential imperatives of intelligent behavior and self-organization in industry, emphasizing the shift from ai to ia, distributed intelligence, and the emergence of the age of intelligence, while highlighting the potential impact on human agency and control in the infosphere.', 'chapters': [{'end': 733.535, 'start': 572.718, 'title': 'Designing ecosystems of intelligence', 'summary': 'Discusses the concept of active inference in designing ecosystems of intelligence, emphasizing the existential imperative of intelligent systems and the vision of research and development in the field of artificial intelligence for the next decade and beyond.', 'duration': 160.817, 'highlights': ["The vision laid out in the white paper 'Designing Ecosystems of Intelligence from First Principles' outlines a vision for research and development in artificial intelligence for the next decade and beyond.", 'Active inference foregrounds an existential imperative of intelligent systems, namely curiosity or the resolution of uncertainty.', 'The chapter delves into the existential imperative of intelligent systems, emphasizing the capacity to accumulate evidence for a generative model and the resolution of uncertainty.', 'Professor Friston introduces the concept of active inference as the physics of intelligence, inheriting from the physics of self-organization, providing greater flexibility to adapt to new information.']}, {'end': 1182.474, 'start': 733.535, 'title': 'Existential imperatives in industry', 'summary': 'Discusses the existential imperatives of intelligent behavior and self-organization in industry, emphasizing the shift from ai to ia, distributed intelligence, and the emergence of the age of intelligence, while highlighting the potential impact on human agency and control in the infosphere.', 'duration': 448.939, 'highlights': ['The shift from AI to IA involves the transition to intelligent agents with sentient behavior, distributed cognition, and belief sharing, impacting the interaction and coexistence of multiple entities in the world of communication and technology deployment.', 'The emergence of the age of intelligence signifies a shift from the age of information, focusing on realizing the potential of shared intelligence and equipping people with the right technology and infrastructure for the future.', 'The discussion on the diminishment of human ontology and agency in the infosphere, highlighting the potential erosion of human control and self-determination as information becomes the substrate.']}], 'duration': 609.756, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/V_VXOdf1NMw/pics/V_VXOdf1NMw572718.jpg', 'highlights': ["The vision laid out in the white paper 'Designing Ecosystems of Intelligence from First Principles' outlines a vision for research and development in artificial intelligence for the next decade and beyond.", 'Active inference foregrounds an existential imperative of intelligent systems, namely curiosity or the resolution of uncertainty.', 'The shift from AI to IA involves the transition to intelligent agents with sentient behavior, distributed cognition, and belief sharing, impacting the interaction and coexistence of multiple entities in the world of communication and technology deployment.', 'The chapter delves into the existential imperative of intelligent systems, emphasizing the capacity to accumulate evidence for a generative model and the resolution of uncertainty.']}, {'end': 2077.886, 'segs': [{'end': 1211.058, 'src': 'embed', 'start': 1183.095, 'weight': 3, 'content': [{'end': 1187.943, 'text': 'And I think what he meant by that is being ensconced and enmeshed in technology to such a high degree.', 'start': 1183.095, 'duration': 4.848}, {'end': 1193.591, 'text': 'And information being the first class citizen in our society is kind of truncating our very existence.', 'start': 1188.383, 'duration': 5.208}, {'end': 1198.195, 'text': 'I mean, what would you say to that? Beautifully, beautifully expressed.', 'start': 1193.611, 'duration': 4.584}, {'end': 1199.155, 'text': 'Very poetic.', 'start': 1198.475, 'duration': 0.68}, {'end': 1201.556, 'text': "I'd have to think about that very carefully.", 'start': 1199.955, 'duration': 1.601}, {'end': 1204.456, 'text': "I'm sure there are deep truths there.", 'start': 1202.336, 'duration': 2.12}, {'end': 1211.058, 'text': 'And certainly from the point of view of a physicist, your information is obviously the thing.', 'start': 1205.236, 'duration': 5.822}], 'summary': "Technology's dominance truncates human existence. information is vital.", 'duration': 27.963, 'max_score': 1183.095, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/V_VXOdf1NMw/pics/V_VXOdf1NMw1183095.jpg'}, {'end': 1371.726, 'src': 'embed', 'start': 1347.077, 'weight': 2, 'content': [{'end': 1356.001, 'text': "But I guess reading between the lines and the twinkle in your eye, the idea that he's trying to get across, that may be quite bad for humanity.", 'start': 1347.077, 'duration': 8.924}, {'end': 1358.182, 'text': "Is that the idea? That's what he thinks.", 'start': 1356.021, 'duration': 2.161}, {'end': 1358.642, 'text': 'I mean, he thinks.', 'start': 1358.202, 'duration': 0.44}, {'end': 1362.503, 'text': "You're a computationalist.", 'start': 1361.523, 'duration': 0.98}, {'end': 1367.345, 'text': 'You think that information is the kind of primary substrate.', 'start': 1363.624, 'duration': 3.721}, {'end': 1371.726, 'text': "And in a way, he's also worried that information is becoming the primary substrate.", 'start': 1367.745, 'duration': 3.981}], 'summary': 'The speaker expresses concern about information becoming the primary substrate, potentially having negative consequences for humanity.', 'duration': 24.649, 'max_score': 1347.077, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/V_VXOdf1NMw/pics/V_VXOdf1NMw1347077.jpg'}, {'end': 1678.512, 'src': 'embed', 'start': 1640.248, 'weight': 0, 'content': [{'end': 1655.615, 'text': "There are two sort of metaphors here which might sort of ground the framework or the perspective that I'm trying to think about this worry.", 'start': 1640.248, 'duration': 15.367}, {'end': 1665.621, 'text': 'One which I found very helpful, again coming from my colleagues in industry,', 'start': 1656.756, 'duration': 8.865}, {'end': 1674.368, 'text': 'is just trying to explain the nature of shared intelligence and distributed intelligence.', 'start': 1665.621, 'duration': 8.747}, {'end': 1678.512, 'text': 'And the analogy here would be the brain.', 'start': 1675.79, 'duration': 2.722}], 'summary': 'Exploring shared and distributed intelligence, using the brain as an analogy.', 'duration': 38.264, 'max_score': 1640.248, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/V_VXOdf1NMw/pics/V_VXOdf1NMw1640248.jpg'}, {'end': 1905.318, 'src': 'embed', 'start': 1879.04, 'weight': 1, 'content': [{'end': 1889.606, 'text': 'The agenda, remember, is just to resolve uncertainty, to sate curiosity, and to move towards a state of greater mutual understanding.', 'start': 1879.04, 'duration': 10.566}, {'end': 1890.907, 'text': "So that's the sort of the.", 'start': 1889.626, 'duration': 1.281}, {'end': 1897.932, 'text': 'the non-dystopian view of information sharing.', 'start': 1892.948, 'duration': 4.984}, {'end': 1905.318, 'text': "So Shane Legg said that his definition of intelligence is the ability of an agent, and we're using words here.", 'start': 1898.373, 'duration': 6.945}], 'summary': 'Agenda: resolve uncertainty, sate curiosity, and move towards mutual understanding.', 'duration': 26.278, 'max_score': 1879.04, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/V_VXOdf1NMw/pics/V_VXOdf1NMw1879040.jpg'}], 'start': 1183.095, 'title': 'Information impact on humanity and shared intelligence', 'summary': 'Delves into the impact of information on society, highlighting its influence on human existence, potential impact on autonomy, and the limitations of artificial intelligence. it also explores shared and distributed intelligence, aiming for mutual understanding and resolution of uncertainty.', 'chapters': [{'end': 1609.628, 'start': 1183.095, 'title': 'The impact of information on humanity', 'summary': 'Discusses the impact of information being the primary substrate in our society, its influence on human existence, and the potential truncation of human autonomy and curiosity, emphasizing the significance of belief sharing and the limitations of artificial intelligence in embodying human traits.', 'duration': 426.533, 'highlights': ['Information as the primary substrate The discussion emphasizes the idea of information being the primary substrate in our society, with implications on human existence and autonomy.', 'Truncation of human autonomy and curiosity The potential truncation of human autonomy and curiosity due to the increasing influence of information as the primary substrate in society.', 'Significance of belief sharing The significance of belief sharing in shaping human intelligence and the potential limitations of artificial intelligence in embodying human traits.', 'Limitations of artificial intelligence The limitations of current artificial intelligence in encompassing human traits such as curiosity, planning, and existential imperatives.']}, {'end': 2077.886, 'start': 1609.628, 'title': 'Age of intelligence: shared and distributed intelligence', 'summary': "Explores the concept of shared and distributed intelligence, drawing analogies from the brain's neural network and discussing the potential for symmetrical information sharing in an ecosystem of intelligence, aiming for mutual understanding and resolving uncertainty.", 'duration': 468.258, 'highlights': ["The analogy of the brain's neural network as a framework for shared and distributed intelligence, with the potential for emergent intelligence from multiple interacting 'smart' elements. The discussion of the brain's neural network as a basis for shared and distributed intelligence, with the emergence of collective intelligence from numerous interacting 'smart' elements, exemplifies the concept of distributed intelligence.", 'The potential for symmetrical information sharing in an ecosystem of intelligence, aiming for mutual understanding and resolving uncertainty rather than pursuing autonomous goals. Exploring the shift towards symmetrical information sharing for mutual understanding and uncertainty resolution, rather than pursuing autonomous objectives, presents a non-dystopian view of information sharing in an ecosystem of intelligence.', 'Comparison of different definitions of intelligence, emphasizing a minimal, physics-based definition grounded in information and information geometry. The contrast of various definitions of intelligence, emphasizing a minimal, physics-based definition grounded in information and information geometry, provides a unique perspective on the concept of intelligence.']}], 'duration': 894.791, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/V_VXOdf1NMw/pics/V_VXOdf1NMw1183095.jpg', 'highlights': ["The analogy of the brain's neural network as a framework for shared and distributed intelligence, with the potential for emergent intelligence from multiple interacting 'smart' elements.", 'Exploring the shift towards symmetrical information sharing for mutual understanding and uncertainty resolution, rather than pursuing autonomous objectives, presents a non-dystopian view of information sharing in an ecosystem of intelligence.', 'The discussion emphasizes the idea of information being the primary substrate in our society, with implications on human existence and autonomy.', 'The potential truncation of human autonomy and curiosity due to the increasing influence of information as the primary substrate in society.']}, {'end': 3425.194, 'segs': [{'end': 2107.368, 'src': 'embed', 'start': 2078.367, 'weight': 3, 'content': [{'end': 2092.619, 'text': 'So the kind of intelligence that has this anthropomorphic feel I think people are also seeing in biotic self-organization.', 'start': 2078.367, 'duration': 14.252}, {'end': 2097.202, 'text': "that would be a long way away from the kind of psychology intelligence that we're talking about.", 'start': 2092.619, 'duration': 4.583}, {'end': 2099.905, 'text': 'And yet it rests upon exactly the same kind of mechanics.', 'start': 2097.483, 'duration': 2.422}, {'end': 2107.368, 'text': 'And for me that would be the Bayesian mechanics that come from the dynamics of systems that are self-organizing,', 'start': 2100.465, 'duration': 6.903}], 'summary': 'Biotic self-organization shows anthropomorphic intelligence, relying on bayesian mechanics.', 'duration': 29.001, 'max_score': 2078.367, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/V_VXOdf1NMw/pics/V_VXOdf1NMw2078367.jpg'}, {'end': 2174.993, 'src': 'embed', 'start': 2143.429, 'weight': 1, 'content': [{'end': 2146.432, 'text': 'accumulating device from the point of view of evolution.', 'start': 2143.429, 'duration': 3.003}, {'end': 2154.919, 'text': "You know, it stores all the information about what it would require to build a phenotype that's fit for purpose in this particular environment.", 'start': 2146.552, 'duration': 8.367}, {'end': 2157.462, 'text': "So, you know, that's a kind of belief updating.", 'start': 2155.24, 'duration': 2.222}, {'end': 2158.783, 'text': "That's a kind of intelligence.", 'start': 2157.522, 'duration': 1.261}, {'end': 2163.307, 'text': "But it doesn't have what we were talking about before, which is this capacity to plan.", 'start': 2159.463, 'duration': 3.844}, {'end': 2165.049, 'text': "Evolution doesn't plan.", 'start': 2163.988, 'duration': 1.061}, {'end': 2169.21, 'text': "You could also argue that the World Wide Web doesn't plan.", 'start': 2166.209, 'duration': 3.001}, {'end': 2174.993, 'text': 'Google Maps, you could argue, does plan to a certain extent because it certainly.', 'start': 2170.971, 'duration': 4.022}], 'summary': 'Evolutionary accumulation stores information for phenotype, lacks planning capacity.', 'duration': 31.564, 'max_score': 2143.429, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/V_VXOdf1NMw/pics/V_VXOdf1NMw2143429.jpg'}, {'end': 2298.189, 'src': 'embed', 'start': 2268.182, 'weight': 2, 'content': [{'end': 2276.809, 'text': 'They are trajectories, dynamics, narratives, paths that have this sort of future pointing aspect.', 'start': 2268.182, 'duration': 8.627}, {'end': 2289.662, 'text': 'then being able to select among different futures, becomes an emergent property of this kind of sense-making, this kind of autopoiesis,', 'start': 2278.37, 'duration': 11.292}, {'end': 2292.826, 'text': 'read as a Bayesian mechanics of self-organization.', 'start': 2289.662, 'duration': 3.164}, {'end': 2298.189, 'text': 'And just thinking about your definitions, they all have the aspect of choosing among different futures,', 'start': 2293.426, 'duration': 4.763}], 'summary': 'Trajectories and narratives lead to emergent property of choosing among different futures.', 'duration': 30.007, 'max_score': 2268.182, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/V_VXOdf1NMw/pics/V_VXOdf1NMw2268182.jpg'}, {'end': 2345.617, 'src': 'embed', 'start': 2321.228, 'weight': 0, 'content': [{'end': 2335.084, 'text': 'which is for me the big bright line between the anthropomorphic kind of intelligence and the intelligence you find in a thermostat or in a variation autoencoder.', 'start': 2321.228, 'duration': 13.856}, {'end': 2340.052, 'text': 'Yeah And later on, we will decompose the different aspects of cognition.', 'start': 2335.888, 'duration': 4.164}, {'end': 2344.216, 'text': 'And I think as well as thinking, it can be knowing and acting as well.', 'start': 2340.292, 'duration': 3.924}, {'end': 2345.617, 'text': "And we'll talk to that.", 'start': 2344.436, 'duration': 1.181}], 'summary': 'Distinguishing anthropomorphic from non-anthropomorphic intelligence, and discussing cognition and action.', 'duration': 24.389, 'max_score': 2321.228, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/V_VXOdf1NMw/pics/V_VXOdf1NMw2321228.jpg'}, {'end': 2425.609, 'src': 'embed', 'start': 2395.718, 'weight': 5, 'content': [{'end': 2399.12, 'text': 'Do single cell organisms plan into the future?', 'start': 2395.718, 'duration': 3.402}, {'end': 2409.492, 'text': 'And to the previous point is planning necessarily a reductive lens of intelligence??', 'start': 2400.14, 'duration': 9.352}, {'end': 2411.254, 'text': 'You brought up loads of interesting things there.', 'start': 2409.512, 'duration': 1.742}, {'end': 2413.116, 'text': 'You call me by the reductive lens.', 'start': 2411.615, 'duration': 1.501}, {'end': 2413.937, 'text': "That's a lovely phrase.", 'start': 2413.156, 'duration': 0.781}, {'end': 2414.458, 'text': 'What does that mean?', 'start': 2413.957, 'duration': 0.501}, {'end': 2425.609, 'text': "Well, I didn't mean it in a pejorative sense, but when we use words, like you know, like we say, intelligence must do X, Y and Z.", 'start': 2415.441, 'duration': 10.168}], 'summary': 'Discussion on intelligence and planning in single-cell organisms.', 'duration': 29.891, 'max_score': 2395.718, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/V_VXOdf1NMw/pics/V_VXOdf1NMw2395718.jpg'}, {'end': 2565.133, 'src': 'embed', 'start': 2544.78, 'weight': 4, 'content': [{'end': 2555.907, 'text': 'And we have this way of summarizing and classifying certain kinds of behavior which may not truly reflect the underlying complexity,', 'start': 2544.78, 'duration': 11.127}, {'end': 2557.368, 'text': "the beauty of what's going on underneath.", 'start': 2555.907, 'duration': 1.461}, {'end': 2560.63, 'text': 'At that point, I would go the other way.', 'start': 2557.388, 'duration': 3.242}, {'end': 2565.133, 'text': "I would actually say that do not properly reflect the underlying simplicity of what's going on.", 'start': 2560.65, 'duration': 4.483}], 'summary': 'Classifying behavior may not reflect complexity or simplicity accurately.', 'duration': 20.353, 'max_score': 2544.78, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/V_VXOdf1NMw/pics/V_VXOdf1NMw2544780.jpg'}, {'end': 2635.101, 'src': 'embed', 'start': 2610.378, 'weight': 6, 'content': [{'end': 2615.683, 'text': 'as if I think the free energy principle is another example of this kind of reductive thing.', 'start': 2610.378, 'duration': 5.305}, {'end': 2622.91, 'text': "It's looking at something which is inherently much more simple than the lens through which you're looking at, which is the Bayesian mechanics itself.", 'start': 2616.604, 'duration': 6.306}, {'end': 2624.411, 'text': 'The free energy principle.', 'start': 2623.11, 'duration': 1.301}, {'end': 2630.917, 'text': "So I think that's absolutely right, a really interesting idea.", 'start': 2625.152, 'duration': 5.765}, {'end': 2635.101, 'text': 'And At another level, I think it speaks to some key issues.', 'start': 2630.917, 'duration': 4.184}], 'summary': 'The free energy principle is a reductive approach to bayesian mechanics, presenting an interesting idea.', 'duration': 24.723, 'max_score': 2610.378, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/V_VXOdf1NMw/pics/V_VXOdf1NMw2610378.jpg'}, {'end': 2684.611, 'src': 'embed', 'start': 2659.443, 'weight': 7, 'content': [{'end': 2668.512, 'text': 'And you know what would that look like if I stripped away different levels of meta-awareness or meta-cognition?', 'start': 2659.443, 'duration': 9.069}, {'end': 2679.825, 'text': "if you're a psychologist? and just had a minimal selfhood, is just having plans sufficient to call me an agent, even if I don't know if I'm an agent.", 'start': 2668.512, 'duration': 11.313}, {'end': 2684.611, 'text': 'If I have plans, who else is going to enact those plans?', 'start': 2680.145, 'duration': 4.466}], 'summary': 'Exploring levels of meta-awareness and selfhood in psychology.', 'duration': 25.168, 'max_score': 2659.443, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/V_VXOdf1NMw/pics/V_VXOdf1NMw2659443.jpg'}, {'end': 2790.791, 'src': 'embed', 'start': 2763.556, 'weight': 8, 'content': [{'end': 2770.881, 'text': 'he said researchers and neuroscientists alike, such as Carl Friston And Jeff Hawkins, and even Andrew Ng,', 'start': 2763.556, 'duration': 7.325}, {'end': 2777.963, 'text': 'have flirted with the idea that there might be a single algorithm which underpins intelligence, with the brain acting like a massive TPU,', 'start': 2770.881, 'duration': 7.082}, {'end': 2781.885, 'text': 'repeating instructions ad nauseum to generate complex behavior.', 'start': 2777.963, 'duration': 3.922}, {'end': 2783.785, 'text': "So it's a fascinating idea.", 'start': 2782.365, 'duration': 1.42}, {'end': 2790.791, 'text': "Is that a fair summation? Yes, what's a TPU in this? Oh, a Tensor Programming Unit.", 'start': 2783.865, 'duration': 6.926}], 'summary': 'Researchers explore the idea of a single algorithm for intelligence, likening the brain to a massive tpu.', 'duration': 27.235, 'max_score': 2763.556, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/V_VXOdf1NMw/pics/V_VXOdf1NMw2763556.jpg'}, {'end': 3029.664, 'src': 'embed', 'start': 3000.615, 'weight': 9, 'content': [{'end': 3008.197, 'text': 'So from that point of view, you have a particular kind of universalism that is actually scale-free,', 'start': 3000.615, 'duration': 7.582}, {'end': 3010.637, 'text': 'because you get the same principle emerging at every level.', 'start': 3008.197, 'duration': 2.44}, {'end': 3021.14, 'text': 'And that is basically one way of looking at the deployment of the free energy principle is asking what would it look like when deployed at different scales?', 'start': 3011.178, 'duration': 9.962}, {'end': 3029.664, 'text': 'So you can deploy it at the level of dendritic self-organization, you can deploy it at the level of neuroscience,', 'start': 3021.22, 'duration': 8.444}], 'summary': 'Universalism at every level with free energy principle deployed at different scales.', 'duration': 29.049, 'max_score': 3000.615, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/V_VXOdf1NMw/pics/V_VXOdf1NMw3000615.jpg'}, {'end': 3229.696, 'src': 'embed', 'start': 3201.046, 'weight': 10, 'content': [{'end': 3208.668, 'text': 'So you should be able to predict the sparsity from the first principles at every scale.', 'start': 3201.046, 'duration': 7.622}, {'end': 3212.77, 'text': 'So that sparsity defines the nature of a graph.', 'start': 3209.129, 'duration': 3.641}, {'end': 3216.951, 'text': "And indeed, if you're talking about anything that's deep in a hierarchical sense,", 'start': 3212.81, 'duration': 4.141}, {'end': 3221.573, 'text': "all you're saying is there's a particular kind of graph that I have in mind.", 'start': 3216.951, 'duration': 4.622}, {'end': 3229.696, 'text': "And it has a certain sparsity structure, but crucially, it's a sparsity structure that allows me to call it a hierarchy.", 'start': 3222.113, 'duration': 7.583}], 'summary': 'Predict sparsity to define hierarchical graphs.', 'duration': 28.65, 'max_score': 3201.046, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/V_VXOdf1NMw/pics/V_VXOdf1NMw3201046.jpg'}, {'end': 3291.574, 'src': 'embed', 'start': 3267.975, 'weight': 11, 'content': [{'end': 3275.442, 'text': 'so morphogenesis and the rungs of the emergence ladder and the causal pressures between those rungs.', 'start': 3267.975, 'duration': 7.467}, {'end': 3282.209, 'text': 'And philosophers like George Ellis said that you only have causation between the levels.', 'start': 3275.482, 'duration': 6.727}, {'end': 3286.333, 'text': 'And Douglas Hofstadter in Gödel, Escher, Bach and the strange loop.', 'start': 3282.909, 'duration': 3.424}, {'end': 3291.574, 'text': "He thinks that there's a very complex panoply of causal pressures between the scales.", 'start': 3286.673, 'duration': 4.901}], 'summary': 'Transcript discusses causation between different levels and scales.', 'duration': 23.599, 'max_score': 3267.975, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/V_VXOdf1NMw/pics/V_VXOdf1NMw3267975.jpg'}], 'start': 2078.367, 'title': 'Understanding intelligence and agency', 'summary': 'Explores the connection between biotic self-organization and bayesian mechanics, the limitations of a reductive lens of intelligence, the concept of agency and universal algorithms, and the importance of graph sparsity in defining hierarchical structures. it emphasizes the distinction between anthropomorphic and biological intelligence and the role of counterfactual futures, while discussing the scale-free deployment of the free energy principle and the emergence of multicellular organisms.', 'chapters': [{'end': 2395.298, 'start': 2078.367, 'title': 'Biotic self-organization and bayesian mechanics', 'summary': 'Discusses the connection between biotic self-organization and bayesian mechanics, emphasizing the distinction between anthropomorphic intelligence and biological intelligence, and the role of counterfactual futures in shaping autonomy and sense-making.', 'duration': 316.931, 'highlights': ['The distinction between anthropomorphic intelligence and biological intelligence lies in the concept of counterfactual futures and the capacity to plan.', 'Biological intelligence, exemplified by DNA, involves belief updating and information storage for evolutionary fitness.', 'The discussion emphasizes the role of counterfactual futures in autonomy and emergent properties of sense-making, highlighting the importance of imagining different futures and curiosity.', 'The chapter also touches on the role of cognitive priors and the information traversal points over a geometry, suggesting a deeper complexity beyond conventional understanding.', 'The relevance of Bayesian mechanics and autonomy in open systems that are self-organizing is highlighted, emphasizing the role of trajectories and narratives in shaping intelligence and autonomy.']}, {'end': 2635.101, 'start': 2395.718, 'title': 'Reductive lens of intelligence', 'summary': 'Explores the concept of a reductive lens of intelligence, discussing the limitations of language and models in capturing the complexity and beauty of underlying behaviors and the fundamental question of whether intelligent behavior is deducible from the models and words used.', 'duration': 239.383, 'highlights': ['The chapter discusses the limitations of language and models in capturing the complexity and beauty of underlying behaviors, questioning the deducibility of intelligent behavior from the models and words used.', 'The speaker introduces the concept of a reductive lens of intelligence, emphasizing the way we make sense of the world through coarse-graining, reduction, and intuitive models, which may not truly reflect the underlying complexity.', 'The free energy principle is presented as another example of a reductive story, looking at something inherently much simpler through a lens, such as Bayesian mechanics.', 'The discussion touches on the impenetrable realm of subjective experience, beyond function, dynamics, and behavior, raising the question of the extent to which intelligent behavior is deducible from models and words.', 'The chapter highlights the concept of existential imperatives and self-evidencing, portraying belief updating, sentience, and intelligence as reductive stories that make sense of what we must or can.']}, {'end': 3130.14, 'start': 2635.101, 'title': 'Exploring universal algorithms and agency', 'summary': 'Discusses the concept of agency, the idea of a universal algorithm underlying intelligence, and the scale-free deployment of the free energy principle across different levels, emphasizing the importance of structure learning and the coupling between levels in various disciplines.', 'duration': 495.039, 'highlights': ['The chapter delves into the concept of agency and the boundaries of selfhood, questioning whether having plans is sufficient to be considered an agent and discussing the causal pressures between agents. Questioning the sufficiency of having plans to be considered an agent, discussing causal pressures between agents.', 'The discussion presents the idea of a universal algorithm underlying intelligence, with proponents such as Carl Friston, Jeff Hawkins, and Andrew Ng suggesting that the brain may act like a massive TPU repeating instructions to generate complex behavior. Proposing the existence of a single algorithm underlying intelligence, mentioning proponents like Carl Friston, Jeff Hawkins, and Andrew Ng.', 'The chapter explores the scale-free deployment of the free energy principle, emphasizing its application at different scales, from dendritic self-organization to neuroscience and morphogenesis, and the coupling between levels in various disciplines. Emphasizing the application of the free energy principle at different scales, discussing the coupling between levels in various disciplines.']}, {'end': 3425.194, 'start': 3130.2, 'title': 'Graph sparsity and hierarchical structures', 'summary': 'Discusses the importance of sparsity in defining the nature of a graph, the principles of self-evidencing, and the causal pressures between hierarchical scales, emphasizing the role of sparsity in hierarchical structures and the emergence of multicellular organisms.', 'duration': 294.994, 'highlights': ['The chapter emphasizes the importance of sparsity in defining the nature of a graph and suggests that adaptive systems must comply with the principles of self-evidencing, where evidence is the marginal likelihood, leading to minimal complexity and the smallest degrees of freedom. Importance of sparsity in defining the nature of a graph, principles of self-evidencing, minimal complexity, smallest degrees of freedom', 'It discusses the causal pressures between hierarchical scales, citing examples from philosophers like George Ellis and Douglas Hofstadter, and explores the emergence of multicellular organisms, highlighting the paradoxical nature of cell specialization and sacrifice for the greater good. Causal pressures between hierarchical scales, emergence of multicellular organisms, cell specialization and sacrifice for the greater good', 'The discussion also touches on the importance of sparsity in hierarchical structures, the role of sparsity in defining a hierarchy, and the significance of sparsity in the emergence of multicellular organisms. Importance of sparsity in hierarchical structures, defining a hierarchy, significance in the emergence of multicellular organisms']}], 'duration': 1346.827, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/V_VXOdf1NMw/pics/V_VXOdf1NMw2078367.jpg', 'highlights': ['The distinction between anthropomorphic and biological intelligence lies in the concept of counterfactual futures and the capacity to plan.', 'Biological intelligence involves belief updating and information storage for evolutionary fitness.', 'The role of counterfactual futures in autonomy and emergent properties of sense-making is emphasized.', 'The relevance of Bayesian mechanics and autonomy in self-organizing open systems is highlighted.', 'The limitations of language and models in capturing the complexity and beauty of underlying behaviors are discussed.', 'The concept of a reductive lens of intelligence is introduced, questioning the deducibility of intelligent behavior from models and words.', 'The free energy principle is presented as an example of a reductive story, looking at something inherently much simpler through a lens.', 'The chapter delves into the concept of agency and the boundaries of selfhood, questioning whether having plans is sufficient to be considered an agent.', 'The idea of a universal algorithm underlying intelligence is presented, with proponents such as Carl Friston, Jeff Hawkins, and Andrew Ng.', 'The scale-free deployment of the free energy principle is explored, emphasizing its application at different scales and the coupling between levels in various disciplines.', 'The importance of sparsity in defining the nature of a graph and its compliance with the principles of self-evidencing is emphasized.', 'The causal pressures between hierarchical scales and the emergence of multicellular organisms are discussed.', 'The significance of sparsity in hierarchical structures and its role in the emergence of multicellular organisms is touched upon.']}, {'end': 5082.252, 'segs': [{'end': 3889.456, 'src': 'embed', 'start': 3864.408, 'weight': 1, 'content': [{'end': 3870.011, 'text': "is to say well, That's not the right question, because we've just talked about separation of temporal scales.", 'start': 3864.408, 'duration': 5.603}, {'end': 3872.492, 'text': 'So you have to say at what temporal scale?', 'start': 3870.011, 'duration': 2.481}, {'end': 3879.373, 'text': 'Well, you operationally define the Markov blanket as the time over which it exists.', 'start': 3873.152, 'duration': 6.221}, {'end': 3889.456, 'text': 'And what would that look like then, when you suddenly now think about situating that temporal scale within the context of a larger temporal scale?', 'start': 3880.834, 'duration': 8.622}], 'summary': 'Discussing the operational definition of markov blanket in relation to temporal scales.', 'duration': 25.048, 'max_score': 3864.408, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/V_VXOdf1NMw/pics/V_VXOdf1NMw3864408.jpg'}, {'end': 4125.359, 'src': 'embed', 'start': 4107.225, 'weight': 2, 'content': [{'end': 4119.314, 'text': 'It commits to the notion that there are states and that those states have a separation of temporal scales that disambiguates or separates systemic states from random fluctuations.', 'start': 4107.225, 'duration': 12.089}, {'end': 4121.216, 'text': "But that's all it does.", 'start': 4120.356, 'duration': 0.86}, {'end': 4125.359, 'text': 'There are no more assumptions and then everything else follows from that.', 'start': 4122.256, 'duration': 3.103}], 'summary': 'States have a separation of temporal scales, disambiguating systemic states from random fluctuations.', 'duration': 18.134, 'max_score': 4107.225, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/V_VXOdf1NMw/pics/V_VXOdf1NMw4107225.jpg'}, {'end': 4581.657, 'src': 'embed', 'start': 4564.91, 'weight': 3, 'content': [{'end': 4581.657, 'text': 'so you know the very hierarchy could also be described now in terms of Markov blankets within Markov blankets within Markov blankets and some really interesting questions about what that means for what part of the brain can know about another part of the brain and what What action becomes and cause it becomes attention?', 'start': 4564.91, 'duration': 16.747}], 'summary': 'Hierarchy described in terms of nested markov blankets, raising questions about brain communication.', 'duration': 16.747, 'max_score': 4564.91, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/V_VXOdf1NMw/pics/V_VXOdf1NMw4564910.jpg'}, {'end': 4745, 'src': 'embed', 'start': 4719.971, 'weight': 0, 'content': [{'end': 4737.617, 'text': "if you look at weight sharing as just the kind of structure learning that's trying to minimize the complexity part of your learning free energy or your negative log marginal likelihood then what you are doing is finding the right kind of sparsity fit for explaining these kinds of data.", 'start': 4719.971, 'duration': 17.646}, {'end': 4745, 'text': "What is the architectural principle? It's the translational invariance, the translational symmetry.", 'start': 4738.657, 'duration': 6.343}], 'summary': 'Weight sharing minimizes complexity and finds sparsity fit for explaining data with translational invariance.', 'duration': 25.029, 'max_score': 4719.971, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/V_VXOdf1NMw/pics/V_VXOdf1NMw4719971.jpg'}], 'start': 3425.194, 'title': 'Temporal scales and markov blankets', 'summary': 'Explores the concept of temporal scales and structural learning, demonstrating how the same principles apply across different levels, and discusses the role of markov blankets in various contexts including system dynamics, brain architecture, and neuroscience.', 'chapters': [{'end': 3815.202, 'start': 3425.194, 'title': 'Temporal scales and structural learning', 'summary': 'Explores the concept of temporal scales and structural learning, demonstrating how the same principles of free energy minimization and information geometry apply across different levels of temporal scales and structural learning, with examples from belief-updating processes, learning, and structural learning in machine learning and complex systems.', 'duration': 390.008, 'highlights': ['The same principles of free energy minimization and information geometry apply across different levels of temporal scales, such as belief-updating processes, learning, and structural learning in machine learning and complex systems, demonstrating a conservation of principles over temporal scales and structural learning.', 'The concept of structural learning in machine learning and complex systems demonstrates a progression of the same principles over years, leveraging free energy minimization and information geometry for formal descriptions of natural selection and Bayesian model selection, showcasing the conservation of principles over extended temporal scales and structural learning.', 'The discussion delves into the notion of Markov boundaries facilitating the interaction between internal and external states, highlighting the significance of boundaries in complex systems and prompting consideration of boundaries in time and space, emphasizing the evolving perspective on boundaries and their relevance in modeling complex systems.']}, {'end': 4085.572, 'start': 3815.542, 'title': 'Markov blankets and temporal scales', 'summary': "Discusses the concept of markov blankets as temporal scales, from large-scale institutions to molecular levels, and explores the idea of observer-relative blankets and their understanding in the context of one's perspective.", 'duration': 270.03, 'highlights': ['Markov blankets as temporal scales The concept of Markov blankets is discussed in the context of temporal scales, from big blankets defining nations and institutions to smaller scale blankets at molecular levels lasting for nanoseconds or milliseconds, suggesting a hierarchical organization.', "Observer-relative blankets The idea of observer-relative Markov blankets is explored, considering how they could be understood in the context of one's perspective, raising questions about their description of reality and their understanding as a lens.", 'Metaphysical nature of Markov blankets The discussion acknowledges the metaphysical nature of Markov blankets, with the speaker admitting a lack of expertise in philosophy and suggesting consulting philosophers for a deeper understanding of the concept.']}, {'end': 4543.42, 'start': 4085.592, 'title': 'Markov blankets and system dynamics', 'summary': 'Discusses the role of markov blankets in disambiguating systemic states from random fluctuations and their implications for simulating large dynamical systems, emphasizing the unanswerable nature of understanding internal states from an external perspective.', 'duration': 457.828, 'highlights': ['The separation of temporal scales disambiguates systemic states from random fluctuations, forming the basis of the free energy principle. The free energy principle commits to the notion of states with a separation of temporal scales, disambiguating systemic states from random fluctuations.', 'For sufficiently large systems, the probability of there not being a Markov blanket is zero, implying a sparse coupling conjecture. The probability of absence of a Markov blanket in sufficiently large systems is zero, supporting the sparse coupling conjecture.', 'The unanswerable nature of understanding internal states from an external perspective is emphasized, as external observers can never know the internal states of a Markov blanket. The chapter emphasizes the unknowable nature of understanding internal states from an external perspective, highlighting the impenetrable nature of being something else.', 'Markov blankets can be deployed as a device to understand and simulate self-organization at different scales in multi-agent simulations. Markov blankets can serve as a device for understanding and simulating self-organization at various scales in multi-agent simulations.', 'The sparsity structure and the dynamics of Markov blankets play a role in social interactions and information sharing within complex systems. The sparsity structure and dynamics of Markov blankets influence social interactions and information sharing within complex systems.']}, {'end': 5082.252, 'start': 4543.42, 'title': 'Understanding markov blankets in the brain', 'summary': 'Discusses the concept of markov blankets within the brain, their hierarchical nature, and their implications for information processing and architecture, emphasizing the relevance to neuroscience and machine learning.', 'duration': 538.832, 'highlights': ["Markov blankets delineating me from a world and the hierarchical nature of Markov blankets within Markov blankets are discussed, raising interesting questions about the brain's information processing and architecture. ", 'The concept of Markov blankets is applied to message passing in factor graphs, influencing the complexity of message passing and defining a hierarchy. ', 'The relevance of Markov blankets to the structure of certain neural networks, such as convolutional neural networks, and their role in minimizing complexity and explaining data is explained. ', 'The discussion extends to the emergence of architectures in intelligent machines related to contiguity properties in the causes of the sensorium and their representation in metric spaces. ', 'The relationship between Markov blankets nesting into hierarchies and the philosophical study of myriology is considered, along with the impact of inductive priors in deep learning on minimizing complexity. ']}], 'duration': 1657.058, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/V_VXOdf1NMw/pics/V_VXOdf1NMw3425194.jpg', 'highlights': ['The same principles of free energy minimization and information geometry apply across different levels of temporal scales, such as belief-updating processes, learning, and structural learning in machine learning and complex systems, demonstrating a conservation of principles over temporal scales and structural learning.', 'Markov blankets as temporal scales The concept of Markov blankets is discussed in the context of temporal scales, from big blankets defining nations and institutions to smaller scale blankets at molecular levels lasting for nanoseconds or milliseconds, suggesting a hierarchical organization.', 'The separation of temporal scales disambiguates systemic states from random fluctuations, forming the basis of the free energy principle. The free energy principle commits to the notion of states with a separation of temporal scales, disambiguating systemic states from random fluctuations.', "Markov blankets delineating me from a world and the hierarchical nature of Markov blankets within Markov blankets are discussed, raising interesting questions about the brain's information processing and architecture."]}, {'end': 6876.523, 'segs': [{'end': 5110.397, 'src': 'embed', 'start': 5083.013, 'weight': 0, 'content': [{'end': 5091.844, 'text': 'You could actually think of much of evolution and the trajectory of machine learning architectures as this game of finding the right structure.', 'start': 5083.013, 'duration': 8.831}, {'end': 5094.527, 'text': 'There are just these structural priors.', 'start': 5092.305, 'duration': 2.222}, {'end': 5099.31, 'text': 'that are apt for describing the kind of data you know.', 'start': 5095.908, 'duration': 3.402}, {'end': 5102.512, 'text': "so if you've spent your entire life doing MNIST images,", 'start': 5099.31, 'duration': 3.202}, {'end': 5110.397, 'text': "then it's going to be the kind of structural prize that inherit from being sampled from a metric space and you'll have all the convolution,", 'start': 5102.512, 'duration': 7.885}], 'summary': 'Evolution of machine learning architectures involves finding structural priors based on the type of data, like mnist images.', 'duration': 27.384, 'max_score': 5083.013, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/V_VXOdf1NMw/pics/V_VXOdf1NMw5083013.jpg'}, {'end': 5166.742, 'src': 'embed', 'start': 5114.459, 'weight': 2, 'content': [{'end': 5131.269, 'text': "I always like listening to Jeff's latest ideas because he's always got the right kind of intuitions and these intuitions are part of the I think part of our reflection of Ashby's law of requisite varieties,", 'start': 5114.459, 'duration': 16.81}, {'end': 5134.331, 'text': 'explored in hypothesis space about the very structures.', 'start': 5131.269, 'duration': 3.062}, {'end': 5149.549, 'text': 'But certainly the capsule stuff, and the one aspect of structures, of graphical models or implicit factor graphs when it comes to implementation,', 'start': 5137.294, 'duration': 12.255}, {'end': 5160.197, 'text': 'which I think is often neglected, is the orthogonal aspect in relation to hierarchical composition.', 'start': 5149.549, 'duration': 10.648}, {'end': 5166.742, 'text': "And that's the breadth of a model in terms of having factors that can be separated.", 'start': 5160.578, 'duration': 6.164}], 'summary': "Exploring ashby's law of requisite varieties in graphical model implementation.", 'duration': 52.283, 'max_score': 5114.459, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/V_VXOdf1NMw/pics/V_VXOdf1NMw5114459.jpg'}, {'end': 5799.371, 'src': 'embed', 'start': 5775.331, 'weight': 3, 'content': [{'end': 5782.394, 'text': "Every little agent's a good scientist designing the right experiments to get the smart data that resolves its uncertainty about what it doesn't know.", 'start': 5775.331, 'duration': 7.063}, {'end': 5785.255, 'text': 'Job done if you can do it like that.', 'start': 5783.674, 'duration': 1.581}, {'end': 5794.806, 'text': 'So that would be one, if you like, sort of answer to your question the implications and the importance of inactivism,', 'start': 5785.775, 'duration': 9.031}, {'end': 5799.371, 'text': 'used really just as a euphemism for an agent that can gather its own data.', 'start': 5794.806, 'duration': 4.565}], 'summary': 'Agents design experiments to resolve uncertainty and gather data independently.', 'duration': 24.04, 'max_score': 5775.331, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/V_VXOdf1NMw/pics/V_VXOdf1NMw5775331.jpg'}, {'end': 5890.686, 'src': 'embed', 'start': 5862.522, 'weight': 5, 'content': [{'end': 5867.445, 'text': "Yes We'll resist the urge that we'll go there immediately after this, I promise.", 'start': 5862.522, 'duration': 4.923}, {'end': 5873.929, 'text': 'So clearly we think that intelligence does necessitate embodiment.', 'start': 5869.086, 'duration': 4.843}, {'end': 5875.069, 'text': "I think that's clear.", 'start': 5874.029, 'duration': 1.04}, {'end': 5882.153, 'text': 'But I want to just explore this continuum between an activism and representationalism.', 'start': 5875.109, 'duration': 7.044}, {'end': 5885.034, 'text': 'This is really interesting.', 'start': 5883.894, 'duration': 1.14}, {'end': 5890.686, 'text': 'This comes down to grounding to a certain extent.', 'start': 5888.324, 'duration': 2.362}], 'summary': 'Discussion on the relation between intelligence, embodiment, and activism vs. representationalism.', 'duration': 28.164, 'max_score': 5862.522, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/V_VXOdf1NMw/pics/V_VXOdf1NMw5862522.jpg'}, {'end': 6244.178, 'src': 'embed', 'start': 6215.876, 'weight': 6, 'content': [{'end': 6226.062, 'text': 'of the kind that has evolved in terms of cultural niche construction and evolutionary psychology that speaks to cultural niche construction that we enjoy,', 'start': 6215.876, 'duration': 10.186}, {'end': 6231.425, 'text': 'in terms of the words that we use and the exchanges that we use.', 'start': 6226.062, 'duration': 5.363}, {'end': 6232.806, 'text': 'Yes, yes, indeed.', 'start': 6231.685, 'duration': 1.121}, {'end': 6238.792, 'text': 'There are two things to bring in, knowledge and language.', 'start': 6234.948, 'duration': 3.844}, {'end': 6244.178, 'text': 'You did invoke Andy Clarke in another interview, which is the extended theory of mind.', 'start': 6238.912, 'duration': 5.266}], 'summary': 'Evolutionary psychology and cultural niche construction shape our language and knowledge.', 'duration': 28.302, 'max_score': 6215.876, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/V_VXOdf1NMw/pics/V_VXOdf1NMw6215876.jpg'}, {'end': 6314.78, 'src': 'embed', 'start': 6293.224, 'weight': 7, 'content': [{'end': 6301.772, 'text': 'then embodied in the structure and presumably the parameters of the connections that constitute that graphical model, that would be knowledge.', 'start': 6293.224, 'duration': 8.548}, {'end': 6313.9, 'text': 'So for me, in a very non-mysterious and possibly too simple-minded way Knowledge is just the product of belief, updating of a slow kind,', 'start': 6302.252, 'duration': 11.648}, {'end': 6314.78, 'text': 'which is learning.', 'start': 6313.9, 'duration': 0.88}], 'summary': 'Knowledge is the product of belief and learning.', 'duration': 21.556, 'max_score': 6293.224, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/V_VXOdf1NMw/pics/V_VXOdf1NMw6293224.jpg'}, {'end': 6830.003, 'src': 'embed', 'start': 6805.584, 'weight': 8, 'content': [{'end': 6813.11, 'text': "And now it's been integrated into GPT and that does this retrieval, augmented generation, which means rather than just being a snapshot in time,", 'start': 6805.584, 'duration': 7.526}, {'end': 6819.916, 'text': 'it will also go out to Bing, get some relevant search results, incorporate that into the prompt and then generate from there.', 'start': 6813.11, 'duration': 6.806}, {'end': 6825.48, 'text': 'And there was this incident a couple of days ago where Bing had this successful launch to much fanfare.', 'start': 6820.416, 'duration': 5.064}, {'end': 6830.003, 'text': 'And then people looked at the results it was generating, including, you know,', 'start': 6826.221, 'duration': 3.782}], 'summary': 'Gpt has integrated retrieval and generation, incorporating bing search results for better prompt generation.', 'duration': 24.419, 'max_score': 6805.584, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/V_VXOdf1NMw/pics/V_VXOdf1NMw6805584.jpg'}], 'start': 5083.013, 'title': 'Evolution of machine learning architectures', 'summary': 'Explores the evolution of machine learning architectures, emphasizing the role of structural priors and inactivism, and discussing the significance of bayesian beliefs and language in cultural niche construction.', 'chapters': [{'end': 5134.331, 'start': 5083.013, 'title': 'Evolution of machine learning architectures', 'summary': "Explores the evolution of machine learning architectures in finding the right structure based on structural priors apt for describing specific data, reflecting ashby's law of requisite varieties and exploring hypothesis space.", 'duration': 51.318, 'highlights': ['The evolution of machine learning architectures is akin to finding the right structure based on structural priors apt for describing specific data, such as MNIST images.', "Jeff's intuitions in machine learning are reflective of Ashby's law of requisite varieties and are essential in exploring hypothesis space about the very structures.", 'The trajectory of machine learning architectures is influenced by the structural priors that are apt for describing specific types of data, like MNIST images.']}, {'end': 6195.445, 'start': 5137.294, 'title': 'Structural learning and inactivism in machine learning', 'summary': 'Discusses the importance of structural learning and inactivism in machine learning, highlighting the role of hierarchical graphical models, conditional independences, and the implications for ai, such as the contrast between representationalism and inactivism. it emphasizes the significance of inactivism in simulating action perception cycles and the shift towards small, agile, intelligent agents for smart data mining in contrast to big data approaches, with a focus on complexity minimization and sustainability.', 'duration': 1058.151, 'highlights': ['The chapter emphasizes the importance of inactivism in simulating action perception cycles and the shift towards small, agile, intelligent agents for smart data mining in contrast to big data approaches. Importance of inactivism in simulating action perception cycles, shift towards small, agile, intelligent agents in smart data mining, contrast to big data approaches.', 'The discussion delves into the implications of inactivism on the contrast between representationalism and inactivism, emphasizing the grounding of cognition in affordances and the need for representations for intelligent action. Implications of inactivism on contrast between representationalism and inactivism, grounding of cognition in affordances, the need for representations for intelligent action.', 'The chapter highlights the significance of hierarchical graphical models, conditional independences, and the implications for AI. Significance of hierarchical graphical models, importance of conditional independences, implications for AI.']}, {'end': 6876.523, 'start': 6195.485, 'title': 'Communication and knowledge in cultural niche construction', 'summary': 'Discusses the communication of bayesian beliefs through factor graphs, the role of knowledge and language in cultural niche construction, and the challenges and implications of using gpt for language generation and understanding.', 'duration': 681.038, 'highlights': ['The role of knowledge and language in cultural niche construction and evolutionary psychology is discussed, highlighting the ability for proper communication and the sharing of beliefs through factor graphs. communication of Bayesian beliefs, probability distributions, factor graph, cultural niche construction, evolutionary psychology', 'The concept of knowledge is explored, linking it to a probabilistic graphical model and the process of belief updating and learning. knowledge, probabilistic graphical model, belief updating, learning', "The challenges and implications of using GPT for language generation and understanding are outlined, including issues with code generation and the integration of ChatGPT into Bing's search engine. GPT language generation, ChatGPT, issues with code generation, integration into Bing search engine"]}], 'duration': 1793.51, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/V_VXOdf1NMw/pics/V_VXOdf1NMw5083013.jpg', 'highlights': ['The evolution of machine learning architectures is akin to finding the right structure based on structural priors apt for describing specific data, such as MNIST images.', 'The trajectory of machine learning architectures is influenced by the structural priors that are apt for describing specific types of data, like MNIST images.', "Jeff's intuitions in machine learning are reflective of Ashby's law of requisite varieties and are essential in exploring hypothesis space about the very structures.", 'The chapter emphasizes the importance of inactivism in simulating action perception cycles and the shift towards small, agile, intelligent agents for smart data mining in contrast to big data approaches.', 'The chapter highlights the significance of hierarchical graphical models, conditional independences, and the implications for AI.', 'The discussion delves into the implications of inactivism on the contrast between representationalism and inactivism, emphasizing the grounding of cognition in affordances and the need for representations for intelligent action.', 'The role of knowledge and language in cultural niche construction and evolutionary psychology is discussed, highlighting the ability for proper communication and the sharing of beliefs through factor graphs.', 'The concept of knowledge is explored, linking it to a probabilistic graphical model and the process of belief updating and learning.', "The challenges and implications of using GPT for language generation and understanding are outlined, including issues with code generation and the integration of ChatGPT into Bing's search engine."]}, {'end': 8137.109, 'segs': [{'end': 6932.474, 'src': 'embed', 'start': 6905.519, 'weight': 0, 'content': [{'end': 6910.241, 'text': 'But whatever, you know, those are not really sort of quantum leaps.', 'start': 6905.519, 'duration': 4.722}, {'end': 6913.183, 'text': "They're not massive technological innovations.", 'start': 6910.281, 'duration': 2.902}, {'end': 6916.284, 'text': 'But what the innovation was of the accessibility.', 'start': 6914.303, 'duration': 1.981}, {'end': 6921.427, 'text': 'So I think you know, just standing back, why that was a moment and it has been a moment, I think you know,', 'start': 6916.304, 'duration': 5.123}, {'end': 6924.149, 'text': 'in terms of selling AI to investors and the like.', 'start': 6921.907, 'duration': 2.242}, {'end': 6927.851, 'text': "They all now, oh, you're talking about chat GTP-like stuff.", 'start': 6924.189, 'duration': 3.662}, {'end': 6928.932, 'text': 'I know about that.', 'start': 6928.211, 'duration': 0.721}, {'end': 6929.952, 'text': "That's really exciting.", 'start': 6928.952, 'duration': 1}, {'end': 6932.474, 'text': 'So it has changed the landscape, I think.', 'start': 6930.593, 'duration': 1.881}], 'summary': 'The innovation of accessibility changed the landscape for selling ai to investors, making it a moment in technological advancements.', 'duration': 26.955, 'max_score': 6905.519, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/V_VXOdf1NMw/pics/V_VXOdf1NMw6905519.jpg'}, {'end': 6998.522, 'src': 'embed', 'start': 6973.077, 'weight': 1, 'content': [{'end': 6982.078, 'text': "And, of course, you ask yourself, okay, what's generative AI brought to the table? Well, it's generating the kind of stuff that I would see.", 'start': 6973.077, 'duration': 9.001}, {'end': 6987.079, 'text': "And, you know, basically it's an interpolation machine.", 'start': 6982.939, 'duration': 4.14}, {'end': 6992.74, 'text': "Sort of if I give it enough stuff, it'll interpolate and generate the kind of things that I've given it.", 'start': 6987.099, 'duration': 5.641}, {'end': 6998.522, 'text': 'So why is that useful? Well, you can now select from the stuff it generates.', 'start': 6993.641, 'duration': 4.881}], 'summary': 'Generative ai can interpolate and generate content based on input, enabling selection from the generated material.', 'duration': 25.445, 'max_score': 6973.077, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/V_VXOdf1NMw/pics/V_VXOdf1NMw6973077.jpg'}, {'end': 7364.609, 'src': 'embed', 'start': 7333.19, 'weight': 2, 'content': [{'end': 7336.211, 'text': "most of it will be generated by machines and we won't know the difference.", 'start': 7333.19, 'duration': 3.021}, {'end': 7345.555, 'text': 'And that reminds me of, you know, one thing which the ambivalence, that whole issue induces in people.', 'start': 7337.99, 'duration': 7.565}, {'end': 7356.923, 'text': 'So you know this tendency to write in meaning and anthropomorphize the content generated by generative AI.', 'start': 7346.016, 'duration': 10.907}, {'end': 7364.609, 'text': "I've heard, actually by the second author of the white paper we started with it, smashed the Turing test, you know.", 'start': 7356.923, 'duration': 7.686}], 'summary': 'Generative ai may blur lines, prompting anthropomorphism, as seen in turing test.', 'duration': 31.419, 'max_score': 7333.19, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/V_VXOdf1NMw/pics/V_VXOdf1NMw7333190.jpg'}, {'end': 7675.217, 'src': 'embed', 'start': 7643.631, 'weight': 3, 'content': [{'end': 7648.112, 'text': 'And these models just learn very, very superficial surface statistics.', 'start': 7643.631, 'duration': 4.481}, {'end': 7650.894, 'text': "about language and how it's used,", 'start': 7648.952, 'duration': 1.942}, {'end': 7663.904, 'text': "and I think it's going to change the peer review process because now so much of this stuff slips beneath the net and the amount of due diligence and rigor that is required to weed out some of this stuff,", 'start': 7650.894, 'duration': 13.01}, {'end': 7667.272, 'text': 'because most of it has gone undetected.', 'start': 7664.69, 'duration': 2.582}, {'end': 7668.032, 'text': "I think that's the problem.", 'start': 7667.292, 'duration': 0.74}, {'end': 7675.217, 'text': "People don't realize how big the problem is because the mistakes are not immediately obvious on the surface.", 'start': 7668.052, 'duration': 7.165}], 'summary': 'Language models lack depth, impacting peer review with unnoticed errors.', 'duration': 31.586, 'max_score': 7643.631, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/V_VXOdf1NMw/pics/V_VXOdf1NMw7643631.jpg'}], 'start': 6880.705, 'title': 'Impact of chat gpt and generative ai', 'summary': 'Discusses the accessibility innovation of chat gpt, changing the ai landscape, and the significance of generative ai in content generation and user engagement. it also addresses challenges and impacts on peer review, emphasizing the need for due diligence and discerning machine-generated content.', 'chapters': [{'end': 6952.406, 'start': 6880.705, 'title': 'Chat gpt and the accessibility innovation', 'summary': 'Discusses how the accessibility innovation of chat gpt has changed the landscape of ai, making it exciting and more accessible to investors through belief sharing and participatory aspect, which has been a significant moment.', 'duration': 71.701, 'highlights': ['The innovation of the accessibility of chat GPT has changed the landscape of AI, making it exciting and more accessible to investors through belief sharing and participatory aspect, which has been a significant moment.', 'The technological structure of the generative model has played a crucial role in making chat GPT work, and the access to it has enchanted people, contributing to its importance.', 'The chat GPT moment has been crucial in terms of selling AI to investors and has changed the perception of AI, as it has become an exciting topic in the industry.']}, {'end': 7261.394, 'start': 6952.406, 'title': 'Generative ai and language models', 'summary': 'Discusses the application of generative ai, its role in content generation, user engagement, and its distinction from belief-sharing. it highlights the asymmetric dyadic interaction and the significance of generative ai in generating language content.', 'duration': 308.988, 'highlights': ["Generative AI's role in content generation and user engagement Generative AI's ability to interpolate and generate content, allowing users to select and triage, has led to its foregrounding in conversations and media, engaging users in a dyadic exchange.", 'Asymmetric dyadic interaction The asymmetric nature of the dyadic interaction with generative AI, where users take on all actions of choice and selection, has captivated people and led to its enchantment.', "Generative AI's distinction from belief-sharing Generative AI's purpose to generate sensations and content in data space, not beliefs, sets it apart from systems like Google Maps, contributing to its unique role in language generation."]}, {'end': 7643.251, 'start': 7262.174, 'title': 'Challenges of generative ai', 'summary': 'Discusses the challenges posed by generative ai, including the potential for misinformation, the difficulty in discerning content generated by machines, and the necessity for smart data mining to preclude confabulated data.', 'duration': 381.077, 'highlights': ["Generative AI pollutes the infosphere with misinformation and false news, posing a significant problem as most information will soon be generated by machines. Generative AI's capability to pollute the infosphere with misinformation and false news on a massive scale is highlighted, indicating a looming challenge as the majority of information on the internet will be machine-generated.", 'The challenge of discerning between content generated by humans and machines, as well as the tendency to anthropomorphize the output of generative AI, is discussed. The difficulty in distinguishing between human-generated and machine-generated content, along with the inclination to anthropomorphize generative AI output, is outlined as a significant challenge.', 'The necessity for equipping smart data mining machines with autonomy to select the right kind of data in order to preclude confabulated data is emphasized. The importance of empowering smart data mining machines with autonomy to select accurate data and prevent confabulated information is stressed as a crucial aspect for the future.']}, {'end': 8137.109, 'start': 7643.631, 'title': 'The impact of language models on peer review', 'summary': 'Discusses the impact of language models on the peer review process, highlighting the need for increased due diligence and rigor to detect and address undetected errors, as well as the connection between the is-ought problem and consciousness in the context of the free energy principle.', 'duration': 493.478, 'highlights': ["Language models' impact on the peer review process The language models have led to a need for increased due diligence and rigor in the peer review process to detect and address undetected errors.", 'The connection between the is-ought problem, consciousness, and the free energy principle The discussion explores the connection between the is-ought problem, consciousness, and the free energy principle, emphasizing the fundamental nature of the is-ought relationship and its implications for distributed cognition and network dynamics.', 'The role of language models in shaping beliefs and values Language models play a role in shaping beliefs and values by influencing the convergence of intelligent artifacts on the web and the resolution of uncertainty, highlighting the expected free energy as a way of specifying oughtness and multiple constraint satisfaction.']}], 'duration': 1256.404, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/V_VXOdf1NMw/pics/V_VXOdf1NMw6880705.jpg', 'highlights': ['The innovation of the accessibility of chat GPT has changed the landscape of AI, making it exciting and more accessible to investors through belief sharing and participatory aspect, which has been a significant moment.', "Generative AI's role in content generation and user engagement Generative AI's ability to interpolate and generate content, allowing users to select and triage, has led to its foregrounding in conversations and media, engaging users in a dyadic exchange.", "Generative AI pollutes the infosphere with misinformation and false news, posing a significant problem as most information will soon be generated by machines. Generative AI's capability to pollute the infosphere with misinformation and false news on a massive scale is highlighted, indicating a looming challenge as the majority of information on the internet will be machine-generated.", "Language models' impact on the peer review process The language models have led to a need for increased due diligence and rigor in the peer review process to detect and address undetected errors."]}, {'end': 9371.905, 'segs': [{'end': 8164.985, 'src': 'embed', 'start': 8137.429, 'weight': 3, 'content': [{'end': 8144.353, 'text': 'Yeah, Many people have an intuition that when designing utility functions, if you look at how markets work and so on,', 'start': 8137.429, 'duration': 6.924}, {'end': 8148.195, 'text': 'that utility and value are orthogonal.', 'start': 8144.353, 'duration': 3.842}, {'end': 8157.721, 'text': 'And then we have institutions like church and government and so on to introduce value pressures onto the utility function.', 'start': 8148.916, 'duration': 8.805}, {'end': 8164.985, 'text': "And I guess what I'm getting from you is certainly from an evolutionary perspective that they need to not be orthogonal.", 'start': 8158.521, 'duration': 6.464}], 'summary': 'Utility and value are not orthogonal from an evolutionary perspective.', 'duration': 27.556, 'max_score': 8137.429, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/V_VXOdf1NMw/pics/V_VXOdf1NMw8137429.jpg'}, {'end': 8233.022, 'src': 'embed', 'start': 8202.225, 'weight': 2, 'content': [{'end': 8209.769, 'text': 'So are they hardwired in the brain or is it just a kind of like constructivist social phenomenon??', 'start': 8202.225, 'duration': 7.544}, {'end': 8217.912, 'text': "I think it's a constructivist social phenomena and a transgenerational sort of niche construction as well, but also in the moment.", 'start': 8210.106, 'duration': 7.806}, {'end': 8223.055, 'text': "And constructivist in the sense that we're not born with these things.", 'start': 8218.451, 'duration': 4.604}, {'end': 8233.022, 'text': 'So these things are very much part of cultural revolutionary thinking and sort of, you know, nurture that we inherit not just from our mum,', 'start': 8223.135, 'duration': 9.887}], 'summary': 'Cultural constructivist social phenomena, not hardwired in the brain.', 'duration': 30.797, 'max_score': 8202.225, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/V_VXOdf1NMw/pics/V_VXOdf1NMw8202225.jpg'}, {'end': 8619.015, 'src': 'embed', 'start': 8590.764, 'weight': 1, 'content': [{'end': 8592.606, 'text': 'So precision is just the inverse uncertainty.', 'start': 8590.764, 'duration': 1.842}, {'end': 8596.469, 'text': 'The confidence with which I can assert, no, I will always do this.', 'start': 8593.066, 'duration': 3.403}, {'end': 8598.85, 'text': "Yeah, as opposed to I'm.", 'start': 8596.969, 'duration': 1.881}, {'end': 8602.711, 'text': "you know 80%, sure, I'm that kind of person.", 'start': 8598.85, 'duration': 3.861}, {'end': 8612.073, 'text': "But also noticing, you're now conditioning your moral position or your ethical position On being a particular kind of person.", 'start': 8602.711, 'duration': 9.362}, {'end': 8614.114, 'text': 'and of course we can all be a different kind of person.', 'start': 8612.073, 'duration': 2.041}, {'end': 8614.954, 'text': 'I can be a teacher.', 'start': 8614.114, 'duration': 0.84}, {'end': 8615.914, 'text': 'I can be a student.', 'start': 8614.994, 'duration': 0.92}, {'end': 8618.295, 'text': 'I can be a parent, you know I can be a friend.', 'start': 8615.954, 'duration': 2.341}, {'end': 8619.015, 'text': 'You know.', 'start': 8618.295, 'duration': 0.72}], 'summary': 'Precision is the inverse uncertainty, aiming for 100% confidence, while acknowledging the ability to embody different roles.', 'duration': 28.251, 'max_score': 8590.764, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/V_VXOdf1NMw/pics/V_VXOdf1NMw8590764.jpg'}, {'end': 8665.625, 'src': 'embed', 'start': 8639.446, 'weight': 0, 'content': [{'end': 8645.391, 'text': "I mean, as you said, you believe that it's socially constructed, which rather gets away from this notion of.", 'start': 8639.446, 'duration': 5.945}, {'end': 8647.352, 'text': "some people believe that it's kind of hardwired.", 'start': 8645.391, 'duration': 1.961}, {'end': 8655.077, 'text': "And then there's the notion of our values might be changing faster than evolution essentially,", 'start': 8648.052, 'duration': 7.025}, {'end': 8661.322, 'text': 'which necessitates the need to have some kind of societal pressures, or governance, if you like.', 'start': 8655.077, 'duration': 6.245}, {'end': 8665.625, 'text': "And I do believe that we're in a new regime now.", 'start': 8662.082, 'duration': 3.543}], 'summary': 'Belief in socially constructed values challenges hardwired beliefs, necessitating societal pressures for change.', 'duration': 26.179, 'max_score': 8639.446, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/V_VXOdf1NMw/pics/V_VXOdf1NMw8639446.jpg'}, {'end': 8937.425, 'src': 'embed', 'start': 8907.183, 'weight': 4, 'content': [{'end': 8912.305, 'text': "if things are speeding up, That usually means you've lost.", 'start': 8907.183, 'duration': 5.122}, {'end': 8918.967, 'text': "You've lost precision in your prior beliefs, which means that the precision, the informativeness, the salience,", 'start': 8912.305, 'duration': 6.662}, {'end': 8922.768, 'text': 'the reliability of information now takes precedence,', 'start': 8918.967, 'duration': 3.801}, {'end': 8930.796, 'text': 'and you increase your learning rate and You become in schizophrenia that we call as jumping to conclusions on the basis of sparse information.', 'start': 8922.768, 'duration': 8.028}, {'end': 8937.425, 'text': "You're basically looking out there, not inside, to resolve uncertainty about states of affairs,", 'start': 8931.117, 'duration': 6.308}], 'summary': 'Speeding up indicates loss of precision in prior beliefs, leading to increased learning rate and jumping to conclusions based on sparse information.', 'duration': 30.242, 'max_score': 8907.183, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/V_VXOdf1NMw/pics/V_VXOdf1NMw8907183.jpg'}], 'start': 8137.429, 'title': 'Morality, values, and society', 'summary': 'Explores the evolution of morality and values, the construction of beliefs and moral decision-making, the impact of societal changes on morals and ethics, and the influence of the information age on emotions and ethical considerations. it delves into the interplay between utility and value, the constructivist nature of belief construction, the precision in moral decision-making, and the impact of globalized connectivity. it also addresses the role of institutions like religion in introducing value pressures and the impact of culture on emotional states.', 'chapters': [{'end': 8201.564, 'start': 8137.429, 'title': 'Evolution of morality and value', 'summary': 'Discusses the interplay between utility and value, highlighting the need for non-orthogonality and the role of institutions like religion in introducing value pressures onto the utility function, emphasizing the distinction between sacred and profane in achieving morality.', 'duration': 64.135, 'highlights': ['The interplay between utility and value, emphasizing the need for non-orthogonality and the role of institutions like religion in introducing value pressures onto the utility function.', 'Emphasizing the distinction between sacred and profane in achieving morality, highlighting the role of institutions like religion in this process.', "The discussion about morality's origins and the different perspectives on achieving morality, contrasting moral realists' views with moral relativists' beliefs."]}, {'end': 8542.746, 'start': 8202.225, 'title': 'Construction of beliefs and moral decision-making', 'summary': 'Discusses the constructivist social phenomenon of belief construction, influenced by transgenerational nurture, epigenetic hardwiring, and the need for selfhood and model of others to make moral and ethical decisions.', 'duration': 340.521, 'highlights': ['The constructivist social phenomenon of belief construction is influenced by transgenerational nurture, epigenetic hardwiring, and the need for selfhood and model of others to make moral and ethical decisions.', 'Beliefs about the way one should behave are learned through interactions and are not necessarily specified at birth.', 'Epigenetically hardwired sub-personal beliefs play a role in homeostasis and allostasis, contributing to the precision and conviction in decision-making.', 'The spectrum of convictions about behavior can vary from absolutist to relativistic, and can be simulated through precise versus less precise beliefs.', "The development of selfhood and a model of others is a prerequisite for making moral or ethical decisions, involving inferring the probability of outcomes based on one's world model and the kind of person one is perceived to be."]}, {'end': 8907.183, 'start': 8542.746, 'title': 'Morals and ethics in a changing society', 'summary': 'Examines the impact of societal changes on morals and ethics, emphasizing the need for precision in moral decision-making amidst evolving values and the potential risks of over-connectivity in a globalized world.', 'duration': 364.437, 'highlights': ["The importance of precision in moral decision-making Precision in moral decision-making is crucial, as it allows for a calculus of alternative ways of responding in a given situation, reducing uncertainty and providing confidence in one's actions.", 'The impact of societal changes on values and ethics The rapid evolution of values in a fast-moving, globalized world necessitates societal pressures and governance to adapt to the changing landscape, requiring a deeper understanding of the societal impact on individual moral positions.', 'Risks of over-connectivity in a globalized world Over-connectivity in a globalized world can lead to the destruction of essential sparsity and disrupt the graceful harmony with the world, analogous to a cancerous cell that has forgotten its boundaries, emphasizing the need to maintain sparsity in communication and carefully sample data.']}, {'end': 9371.905, 'start': 8907.183, 'title': 'The information age and constructed emotions', 'summary': 'Discusses the impact of speeding up information on confidence and precision, the construction of emotions, and the role of culture in shaping emotional states. it also addresses the question of evil in psychiatry.', 'duration': 464.722, 'highlights': ['The impact of speeding up information on confidence and precision Speeding up information indicates a loss of confidence in prior beliefs, leading to an increase in learning rate and jumping to conclusions based on sparse information.', 'The construction of emotions Emotions are constructed through a process of building generative models in a neurodevelopmental context, integrating interceptive and situational awareness.', "The role of culture in shaping emotional states Cultural convention and upbringing mold the repertoire of explanations for emotional states, similar to Eskimos' refined visual discrimination of whiteness due to multiple words for snow.", 'The question of evil in psychiatry The existence of evil is not part of the diagnostic criteria in psychiatry, and the distinction between psychiatry and psychotherapy is clarified.']}], 'duration': 1234.476, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/V_VXOdf1NMw/pics/V_VXOdf1NMw8137429.jpg', 'highlights': ['The rapid evolution of values in a fast-moving, globalized world necessitates societal pressures and governance to adapt to the changing landscape, requiring a deeper understanding of the societal impact on individual moral positions.', "The importance of precision in moral decision-making Precision in moral decision-making is crucial, as it allows for a calculus of alternative ways of responding in a given situation, reducing uncertainty and providing confidence in one's actions.", 'The constructivist social phenomenon of belief construction is influenced by transgenerational nurture, epigenetic hardwiring, and the need for selfhood and model of others to make moral and ethical decisions.', 'The interplay between utility and value, emphasizing the need for non-orthogonality and the role of institutions like religion in introducing value pressures onto the utility function.', 'The impact of speeding up information on confidence and precision Speeding up information indicates a loss of confidence in prior beliefs, leading to an increase in learning rate and jumping to conclusions based on sparse information.']}, {'end': 10760.688, 'segs': [{'end': 9764.928, 'src': 'embed', 'start': 9737.405, 'weight': 0, 'content': [{'end': 9740.508, 'text': 'So become your psychotherapist, you know.', 'start': 9737.405, 'duration': 3.103}, {'end': 9744.23, 'text': 'Well, could I push back a tiny bit?', 'start': 9741.127, 'duration': 3.103}, {'end': 9751.657, 'text': 'So I think part of the reason why we have this focus on ethics is because of the centralization of AI.', 'start': 9744.27, 'duration': 7.387}, {'end': 9756.722, 'text': "Things like Facebook, they're controlled by centralized corporations.", 'start': 9753.659, 'duration': 3.063}, {'end': 9760.625, 'text': "And What you're alluding to is far more interesting.", 'start': 9757.483, 'duration': 3.142}, {'end': 9764.928, 'text': "It's this multi-agent diffusion, stratification, fractionation.", 'start': 9760.705, 'duration': 4.223}], 'summary': 'Centralization of ai leads to focus on ethics; multi-agent diffusion is more interesting.', 'duration': 27.523, 'max_score': 9737.405, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/V_VXOdf1NMw/pics/V_VXOdf1NMw9737405.jpg'}, {'end': 9832.074, 'src': 'embed', 'start': 9810.619, 'weight': 1, 'content': [{'end': 9823.186, 'text': 'So, when I was talking about belief sharing and distributed the age of intelligence, it was exactly this distributed ecosystem,', 'start': 9810.619, 'duration': 12.567}, {'end': 9826.208, 'text': 'a democratized kind of belief sharing and data sharing.', 'start': 9823.186, 'duration': 3.022}, {'end': 9832.074, 'text': 'where the data is small bits of smart data that are essential to reduce uncertainty.', 'start': 9827.089, 'duration': 4.985}], 'summary': 'Belief and data sharing create a distributed ecosystem to reduce uncertainty.', 'duration': 21.455, 'max_score': 9810.619, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/V_VXOdf1NMw/pics/V_VXOdf1NMw9810619.jpg'}, {'end': 10325.854, 'src': 'embed', 'start': 10293.815, 'weight': 2, 'content': [{'end': 10305.426, 'text': "So for me, that's a sort of bright line between sort of sentience, at least, and self-organization that is not organized.", 'start': 10293.815, 'duration': 11.611}, {'end': 10314.049, 'text': 'sentient and that distinction is exactly what is being tested in this adversarial research collaboration funded by the Templeton Foundation.', 'start': 10305.866, 'duration': 8.183}, {'end': 10325.854, 'text': "It's called Intrepid and it's contrasting IIT integrated information theory with various predictive processing,", 'start': 10314.089, 'duration': 11.765}], 'summary': 'Adversarial research collaboration testing sentience and self-organization funded by templeton foundation.', 'duration': 32.039, 'max_score': 10293.815, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/V_VXOdf1NMw/pics/V_VXOdf1NMw10293815.jpg'}, {'end': 10489.55, 'src': 'embed', 'start': 10460.229, 'weight': 3, 'content': [{'end': 10462.57, 'text': 'So you have to be able to select it.', 'start': 10460.229, 'duration': 2.341}, {'end': 10465.411, 'text': 'What does that mean? You have to be able to attend to it.', 'start': 10463.15, 'duration': 2.261}, {'end': 10471.375, 'text': 'And you have to be the source of that sort of enabling attention, that active sampling inside the brain.', 'start': 10466.532, 'duration': 4.843}, {'end': 10482.764, 'text': "So now there's a deep link between qualia or qualitative experience and attention and conscious processing of the kind that is accompanied by qualitative experience.", 'start': 10471.876, 'duration': 10.888}, {'end': 10484.586, 'text': "I think it's important.", 'start': 10483.585, 'duration': 1.001}, {'end': 10489.55, 'text': 'It also, interestingly, also speaks to your dynamic Markov blankets, because, of course,', 'start': 10485.206, 'duration': 4.344}], 'summary': 'Conscious processing requires enabling attention and active sampling inside the brain.', 'duration': 29.321, 'max_score': 10460.229, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/V_VXOdf1NMw/pics/V_VXOdf1NMw10460229.jpg'}], 'start': 9373.466, 'title': 'Ai ethics and sentient behavior', 'summary': 'Covers the intersection of psychiatry and ai ethics, the need for decentralized ai ethics and ecosystems, and the distinction between sentience and self-organization, with a focus on multi-agent diffusion and conscious behavior planning.', 'chapters': [{'end': 9712.686, 'start': 9373.466, 'title': 'The ethics of behavior and ai', 'summary': 'Delves into the intersection of psychiatry and ai ethics, discussing the potential link between psychiatric conditions and behaviors regarded as evil, as well as the ethical implications of ai decision-making and the concept of free will.', 'duration': 339.22, 'highlights': ['The expression of certain behaviors regarded as evil may arise from psychiatric conditions that preclude proper theory of mind and the ability to sympathize or empathize. Certain psychiatric conditions may lead to behaviors considered evil due to the lack of theory of mind and empathy, impacting the ability to relate and sympathize with others.', 'The concept of evil itself may be a constructed concept influenced by language and individual beliefs, leading to varying perceptions of what constitutes evil behavior. The concept of evil is influenced by language and individual beliefs, resulting in varying perceptions of what constitutes evil behavior, suggesting its subjective nature.', "The discussion explores the ethical implications of AI decision-making, questioning what AI should prioritize and who determines its objectives, raising concerns about its potential impact on society. The ethical implications of AI decision-making are examined, addressing the determination of AI priorities and objectives, and the potential societal impact of AI's actions."]}, {'end': 10251.591, 'start': 9712.706, 'title': 'Decentralized ai ethics and ecosystems', 'summary': 'Emphasizes the need for decentralized ai ethics and ecosystems to preclude dystopian outcomes, with a focus on multi-agent diffusion, belief sharing, and constraints to prevent the emergence of autocracies in the age of intelligence.', 'duration': 538.885, 'highlights': ['The emphasis is on decentralized ecosystems and multi-agent diffusion to preclude dystopian outcomes in the age of intelligence. The discussion focuses on the importance of decentralized ecosystems and multi-agent diffusion to prevent dystopian outcomes in the age of intelligence.', 'Belief sharing and data sharing in a distributed ecosystem are essential to reduce uncertainty The speaker emphasizes the importance of belief sharing and data sharing in a distributed ecosystem to reduce uncertainty.', 'The need for constraints and governance to address behaviors that emerge in the multi-agent setting with humans and machines The conversation highlights the necessity of introducing constraints and governance to address emergent behaviors in the multi-agent setting involving humans and machines.', 'Emphasis on standards, contracts, and shared agreements to prevent catastrophic, dystopian abuses in the age of intelligence The discussion stresses the importance of standards, contracts, and shared agreements to prevent catastrophic, dystopian abuses in the age of intelligence.', 'The proposal of the mind coming into being when self-evidencing has a temporal thickness, grounding inferences about the consequences of actions The speaker proposes that the mind comes into being when self-evidencing has a temporal thickness, grounding inferences about the consequences of actions.']}, {'end': 10760.688, 'start': 10252.691, 'title': 'Consciousness and planning in sentient behavior', 'summary': 'Discusses the distinction between sentience and self-organization, the role of planning in conscious behavior, and the link between attention and qualitative experience in the context of an adversarial research collaboration funded by the templeton foundation.', 'duration': 507.997, 'highlights': ['The distinction between sentience and self-organization The chapter emphasizes the bright line between sentience and self-organization, indicating that the capacity to plan and select amongst alternative courses of action is a key factor in determining sentience.', 'The role of planning in conscious behavior It discusses the importance of planning and free selection amongst alternative courses of action as essential components of conscious behavior, setting a clear distinction between conscious and non-conscious systems.', 'The link between attention and qualitative experience The chapter explores the deep link between attention, conscious processing, and qualitative experience, suggesting that the ability to select, attend to, and be the source of enabling attention is a prerequisite for qualitative experience.', 'Adversarial research collaboration funded by the Templeton Foundation It mentions an adversarial research collaboration called Intrepid, funded by the Templeton Foundation, contrasting integrated information theory with various predictive processing, highlighting the significance of this research in testing the distinction between sentience and self-organization.']}], 'duration': 1387.222, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/V_VXOdf1NMw/pics/V_VXOdf1NMw9373466.jpg', 'highlights': ["The ethical implications of AI decision-making are examined, addressing the determination of AI priorities and objectives, and the potential societal impact of AI's actions.", 'The discussion focuses on the importance of decentralized ecosystems and multi-agent diffusion to prevent dystopian outcomes in the age of intelligence.', 'The chapter emphasizes the bright line between sentience and self-organization, indicating that the capacity to plan and select amongst alternative courses of action is a key factor in determining sentience.', 'The chapter explores the deep link between attention, conscious processing, and qualitative experience, suggesting that the ability to select, attend to, and be the source of enabling attention is a prerequisite for qualitative experience.']}], 'highlights': ['The vision of a cyber-physical ecosystem of intelligence embraces humans and nature and is based on first principles derived from physics and biology.', 'Active inference is presented as a physics of intelligence, where intelligent systems accumulate evidence for a generative model of their sensed world to reduce uncertainty and achieve their goals.', 'The proposal of a research agenda for designing ecosystems of intelligence from scratch involves developing shared hyperspatial modeling language, transaction protocols, and novel methods for measuring and optimizing collective intelligence.', 'The vision aims to harness the power of artificial intelligence for the common good without compromising human dignity or autonomy, challenging individuals to rethink their relationship with technology, nature, and each other.', 'The potential of the free energy principle to reshape views on the connection between inanimate matter and living things, and to provide insights into the emergence of consciousness and intelligence from physical matter and processes, is discussed.', "The vision laid out in the white paper 'Designing Ecosystems of Intelligence from First Principles' outlines a vision for research and development in artificial intelligence for the next decade and beyond.", 'Active inference foregrounds an existential imperative of intelligent systems, namely curiosity or the resolution of uncertainty.', 'The shift from AI to IA involves the transition to intelligent agents with sentient behavior, distributed cognition, and belief sharing, impacting the interaction and coexistence of multiple entities in the world of communication and technology deployment.', "The analogy of the brain's neural network as a framework for shared and distributed intelligence, with the potential for emergent intelligence from multiple interacting 'smart' elements.", 'Exploring the shift towards symmetrical information sharing for mutual understanding and uncertainty resolution, rather than pursuing autonomous objectives, presents a non-dystopian view of information sharing in an ecosystem of intelligence.', 'The same principles of free energy minimization and information geometry apply across different levels of temporal scales, such as belief-updating processes, learning, and structural learning in machine learning and complex systems, demonstrating a conservation of principles over temporal scales and structural learning.', 'The evolution of machine learning architectures is akin to finding the right structure based on structural priors apt for describing specific data, such as MNIST images.', 'The innovation of the accessibility of chat GPT has changed the landscape of AI, making it exciting and more accessible to investors through belief sharing and participatory aspect, which has been a significant moment.', "Generative AI's role in content generation and user engagement Generative AI's ability to interpolate and generate content, allowing users to select and triage, has led to its foregrounding in conversations and media, engaging users in a dyadic exchange.", "Generative AI pollutes the infosphere with misinformation and false news, posing a significant problem as most information will soon be generated by machines. Generative AI's capability to pollute the infosphere with misinformation and false news on a massive scale is highlighted, indicating a looming challenge as the majority of information on the internet will be machine-generated.", 'The rapid evolution of values in a fast-moving, globalized world necessitates societal pressures and governance to adapt to the changing landscape, requiring a deeper understanding of the societal impact on individual moral positions.', "The ethical implications of AI decision-making are examined, addressing the determination of AI priorities and objectives, and the potential societal impact of AI's actions.", 'The discussion focuses on the importance of decentralized ecosystems and multi-agent diffusion to prevent dystopian outcomes in the age of intelligence.']}