All Segments
Tomaso Poggio: Brains, Minds, and Machines | Lex Fridman Podcast #13
42 segments available
Segments Timeline
The Fascination with Einstein
Tommaso Poggio shares his childhood admiration for Einstein and discusses the essence of his genius, particularly in relation to the theory of relativity. He reflects on how Einstein's ability to simplify complex ideas through thought experiments showcases the power of human intelligence and imagination.
"the following is a conversation with Tommaso poggio he's the professor at MIT and as a director of the Center for brains minds and machines sited over 100,000 times his work has had a profound impact ..."
Lessons from Einstein's Journey
Poggio recounts Einstein's unconventional path to success, highlighting that despite being the least successful student, he became a revolutionary thinker. This segment emphasizes the importance of nonconformity and thinking differently, drawing parallels to success in both science and the stock market.
"mentioned that in your childhood you've developed a fascination with physics especially the theory of relativity and that Einstein was also a childhood hero to you what aspect of Einstein's genius the..."
The Possibility of Time Travel
In a thought-provoking discussion, Poggio explores the concept of time travel, expressing skepticism about its feasibility. He discusses the scientific understanding of time and the potential for forward time travel, while reflecting on the dreams of engineering intelligence capable of such feats.
"fascinating was the power of intelligence the power of the mind do you think the ability to imagine to visualize as he did as a lot of great forces sister do you think that's in all of us human beings..."
The Quest for Intelligent Machines
Poggio discusses the aspiration to create machines that can think as well as humans, or even surpass human intelligence. He emphasizes the ongoing advancements in AI and the complexities of defining intelligence, consciousness, and the potential for machines to enhance human thinking.
"very unlikely so do you still hold the underlying dream of the engineering intelligence that will build systems that are able to do such huge leaps like discovering the kind of mechanism that would be..."
Understanding Intelligence: The Ultimate Question
Poggio reflects on his lifelong fascination with the nature of intelligence, considering it the greatest problem in science. He discusses the interplay between human intelligence and the scientific process, emphasizing the importance of understanding our own cognitive capabilities.
"found that there was the problem of time and space and general relativity but there were so many other problems of the same level of difficulty and importance that I could even if I were I stein it wa..."
Biological vs. Artificial Intelligence
In this segment, Poggio examines the relationship between biological and artificial neural networks. He discusses the progress in AI, particularly in reinforcement learning, and the influence of neuroscience on recent breakthroughs, suggesting that understanding the brain is crucial for future advancements.
"in a hypothetical way let me ask how far do you think we can get in creating intelligent systems without understanding the biological the understanding how the human brain creates intelligence put ano..."
The Role of Neuroscience in AI
Poggio elaborates on how neuroscience has shaped recent AI developments, particularly in deep learning. He highlights the historical context of these breakthroughs and expresses optimism about the continued influence of neuroscience on future AI innovations.
"or that is the best way to get there is to understand better the human brain yeah okay this is I think an educated bet that different people with different background will decide in different ways the..."
Challenges in Deep Learning
Poggio discusses the limitations of current deep learning techniques, particularly the reliance on large labeled datasets. He contrasts this with how biological systems learn from fewer examples, emphasizing the need for AI to evolve beyond its current data dependency.
"bet is that there is a good chance they continue to play a big role maybe not in all the future breakthroughs but in some of them at least in inspiration so at least in a new spirit absolutely yes so ..."
Nature vs. Nurture in Learning
In a deep exploration of learning, Poggio addresses the balance between genetic predispositions and experiential learning. He reflects on evolutionary mechanisms that shape intelligence and the potential for general learning capabilities in artificial systems.
"take care takers so how do you solve that you know I think there is this interesting challenge that today deep learning and related techniques are all about big data big data meaning a lot of examples..."
Face Recognition and Brain Development
Poggio examines the neural mechanisms behind face recognition, referencing research on baby monkeys raised without exposure to faces. He speculates on whether the brain's face recognition area is hardwired or learned, emphasizing the importance of early experiences in shaping this critical cognitive function.
"machinery and then had to give very weak priors like for instance let me take give a specific example which is recent to work by a member of our Center for brains minds and machines we know because of..."
The Plasticity of the Brain
This segment focuses on the brain's plasticity, particularly in the context of face recognition. Poggio discusses how early exposure to faces can influence neural development and the implications of this for understanding how the brain learns to recognize important stimuli in its environment.
"guess at some of these so of course there's a caveat before a lot of the stuff we talk about but what is your hunch is the face the part of the brain that that seems to be concentrated on face recogni..."
Understanding Brain Modules
Poggio addresses the modularity of the brain, explaining how different regions are specialized for various functions. He contrasts historical beliefs about brain equipotentiality with current understanding that specific brain areas are crucial for tasks like language and motor control.
"but this will require probably a lot of bits you had to specify a lot of connection of a lot of neurons instead that the command that commands from the gene is something like imprint memorized what yo..."
Functional Specialization in the Brain
In this segment, Poggio elaborates on the concept of functional specialization within the brain. He discusses how certain brain injuries can lead to specific deficits, reinforcing the idea that different brain regions are dedicated to distinct cognitive functions.
"contribute to intelligence do you see the brain as a bunch of different modules and they together come in the human brain to create intelligence or is it all one mush of the same kind of fundamental a..."
The Interplay of Brain Regions
Poggio explores the interplay between different brain regions, particularly the cortex, and how they contribute to intelligence. He discusses the similarities in structure across various cortical areas and the implications for understanding cognitive processes.
"there are specific modules of the answer that we know from this old work which was basically on based on lesions either on animals or very often there were a mine of well it there was a mine a very in..."
Cortex Functionality and Learning
This segment focuses on the functionality of the cortex, discussing how different regions are responsible for various modalities like vision and language. Poggio raises questions about the underlying mechanisms of learning and how they might relate to artificial neural networks.
"solidified in hardware that's a great question so there are parts of the brain like the cerebellum or they put campus that are quite different from each other they clearly have different Anatomy diffe..."
The Human Visual Cortex Explained
Poggio provides insights into the human visual cortex, discussing what is known and unknown about its functioning. He emphasizes the complexity of understanding vision and the challenges that remain in neuroscience, including fundamental questions like the purpose of sleep.
"instructive for it maybe artificial neural networks so you've done a lot of great work in vision and human vision computer vision and you mentioned the problem of human vision is really as difficult a..."
Levels of Understanding Intelligence
In this segment, Poggio discusses the different levels of understanding intelligence, comparing it to computer systems. He emphasizes the interconnectedness of hardware and software in the brain, suggesting that understanding intelligence requires a holistic approach that integrates various fields of expertise.
"you think so taking a step back on that so sleep for examples fascinating do you think that's a neuroscience question or if we talk about abstractions what do you think is an interesting way to study ..."
Compositionality in Neural Networks
Poggio introduces the concept of compositionality in neural networks, explaining how deep networks can learn complex functions more effectively than shallow networks. He discusses the significance of compositional structures in understanding cognitive processes and their implications for artificial intelligence.
"and have PowerPoint do you think it's digging in a little bit more how useful is it to understand the transistor in order to be able to understand PowerPoint and these higher-level very good intellige..."
Compositionality in Neural Networks
Poggio explains the concept of compositionality in deep neural networks, illustrating how they can learn complex functions more effectively than shallow networks. He draws parallels between interpreting images and reading sentences, emphasizing that deep networks leverage local computations to understand larger structures.
"these neural networks in artificial or biological sense learn anything you talk about compositionality see there's a sense that nature can be disentangled our purpura well all aspects of our cognition..."
Local Interactions and Compositional Structure
In a philosophical discussion, Poggio contrasts his views with physicist Max Tegmark regarding the origins of compositional structures in learning. He argues that the brain's deep network architecture is designed to solve problems with compositional structures, which may have evolved due to survival needs.
"structure is a structure of compositionality if the function is made up of functions of function so that you need to look on when you are interpreting an image classifying an image you don't need to l..."
Biological Constraints on Neural Connectivity
Poggio explores the biological limitations of neural connectivity in the brain, suggesting that the architecture of deep networks may stem from evolutionary constraints. He discusses how local connectivity in the brain facilitates problem-solving in compositional contexts, which are crucial for survival.
"differently he is conclusion is that for images for instance the compositional structure of this function that we have to learn or to solve these problems comes from physics comes from the fact that y..."
Stochastic Gradient Descent and Learning
Poggio delves into the mechanics of how artificial neural networks learn through stochastic gradient descent. He raises questions about the biological plausibility of this method and speculates on the existence of alternative optimization algorithms that could enhance learning in neural networks.
"evolutionary perspective as we've so the ones that didn't have the they weren't dealing with a compositional nature of reality died off yes it also could be may be the reason why we have this local co..."
The Mystery of Over-Parameterization
Discussing the phenomenon of over-parameterization in neural networks, Poggio highlights how having more parameters than data can lead to effective learning. He explains the implications of this for finding solutions in deep networks and the abundance of minima available for optimization.
"problems right mobile a driving is a compositional problem right so on the unlearning task i mean we don't know much about how the brain learns in terms of optimization but so the thing that's stochas..."
Universality and the Curse of Dimensionality
Poggio addresses the universal approximation theorem and its implications for neural networks. He discusses the challenges posed by the curse of dimensionality, emphasizing how deep architectures can mitigate these issues when dealing with complex functions.
"algorithms that can replace stochastic gradient descent and my my guess is yes but nobody has found yet a real answer I mean people are trying still trying and there are some interesting ideas the fac..."
Unsupervised Learning and GANs
Poggio shares his views on the challenges of unsupervised learning, particularly in relation to Generative Adversarial Networks (GANs). He expresses skepticism about their potential to significantly advance intelligence, while acknowledging their applications in computer graphics and other areas.
"been one of the puzzles about neural networks how can you get something that really works when you have so much freedom in its in from that Laura Derek in general right somehow right exactly do you th..."
The Future of Learning Algorithms
In a forward-looking discussion, Poggio contemplates the future of learning algorithms, particularly in the context of supervised and unsupervised learning. He emphasizes the importance of finding effective methods to reduce the reliance on labeled data while acknowledging the limitations inherent in current approaches.
"this is why this is because of the overall parameterization the organization sprinkles an entire space for solutions pretty good and so not so surprising right is like you know if you have a system of..."
Learning from Evolution
Poggio discusses how biological evolution has shaped learning processes in humans and other organisms. He emphasizes the importance of innate mechanisms for detecting motion and how these can inform AI learning strategies.
"n to infinity in supervised and to infinity in supervised learning so for example mobile I has very successfully I mean essentially annotated large amounts of data to be able to drive a car now one th..."
Understanding Scene Recognition
This segment explores the gap between visual recognition and scene understanding in AI. Poggio argues that while current algorithms excel at low-level vision tasks, they still fall short of true comprehension of scenes and actions.
"of making one-shot networks that learn much faster is often on the architecture side but how can we pick better examples with wish to learn do you have intuitions about that well that's part of the qu..."
The Existential Threat of AI
Poggio shares his thoughts on the existential risks posed by AI, advocating for early concern and proactive safety measures. He critiques sensational comparisons of AI dangers to nuclear weapons, emphasizing the need for a balanced perspective.
"know all insects from fruit flies other animals they have this even in the readiness of in the very peripheral part it's very conserved across species something that evolution discovered early it may ..."
Predictions for General Intelligence
In this segment, Poggio discusses predictions regarding the timeline for achieving artificial general intelligence (AGI). He reflects on varying opinions within the field and shares his belief that true AGI may still be a century away.
"otherwise it's really difficult because you have so much stuff so suppose you do this at the beginning first weeks then after that you can recognize the object now they're imprinted a number of even i..."
Understanding AI Systems
Poggio delves into the complexities of understanding AI systems, particularly deep learning models. He compares this understanding to other complex systems, suggesting that while we may grasp their functionality, the intricacies of their learning processes remain elusive.
"language of actions of people that is despite the claims that's I think very far or a little bit off so in popular culture and among many researchers some of which I've spoken with the sue Russell and..."
The Ethics of AI
Poggio discusses the challenges of instilling ethics in AI systems. He emphasizes the importance of understanding human ethics and the potential for neuroscience to inform the development of ethical AI.
"think about possible safety against what I find a bit misleading are things like that I've been said by people I know like Elon Musk and what is boström important notice first name a neck panic poster..."
Consciousness and AI
In this thought-provoking segment, Poggio explores the relationship between consciousness and intelligence in AI. He debates whether consciousness is necessary for true intelligence and discusses the philosophical implications of creating conscious machines.
"and what's the confidence interval or not you know it's so difficult to predict the future and even the presence of it's nothing it's pretty hard to predict a bit I'll be but as I said this is complet..."
The Future of Learning Machines
Poggio reflects on the future of learning machines, suggesting that while we may not fully understand their learning processes, we can still construct effective systems. He draws parallels to human learning and the complexities involved.
"all right so many years ago this was actually 40 41 years ago I wrote a paper with David Marr who was one of the founding father of computer vision of computational dish I wrote a paper about levels o..."
The Purpose of Neuroscience
Tomaso Poggio discusses the potential of neuroscience to illuminate ethical questions and its role in understanding consciousness. He highlights the ongoing debate among neuroscientists and philosophers about whether consciousness requires a biological substrate or if it can exist in silicon devices, setting the stage for deeper inquiries into the nature of intelligence.
"colleague of mine Rebecca Saxe and there is a other researchers doing similar work and I think you know this is the beginning but ideally at some point we'll have an understanding of how this works an..."
Consciousness and Intelligence
In this segment, Poggio explores the relationship between consciousness and artificial intelligence. He questions whether consciousness is necessary for creating intelligent systems, referencing the Turing test and differing opinions among experts. This discussion raises fundamental questions about self-awareness and the essence of intelligence.
"because it's very difficult to define consciousness and and there is the debate among neuroscientist and about whether consciousness and philosophers of course whether consciousness is something that ..."
Mortality's Role in Consciousness
Poggio reflects on the significance of mortality in shaping consciousness and intelligence. He references Steve Jobs' perspective on the importance of a finite life for achievement, pondering whether the awareness of death enhances the quality of life and consciousness, or if it's merely an evolutionary byproduct.
"to the interlinked you know most of the people working in artificial intelligence I think who'd answer we don't strictly need the consciousness to have an intelligent system that's sort of the easier ..."
The Next Breakthrough in AI
Discussing the future of artificial intelligence, Poggio speculates on the next major breakthroughs inspired by neuroscience. He mentions MIT's quest for intelligence and expresses excitement about ongoing projects aimed at understanding visual intelligence, emphasizing the importance of perception in navigating the world.
"you know I never thought about it is it a very interesting question I think Steve Jobs in his commencement speech at Stanford argued that you know having a finite life was important for for stimulatin..."
Understanding Visual Intelligence
Poggio elaborates on a moonshot project focused on visual intelligence, detailing an experiment conducted at Google X. He describes how participants experienced a sense of self-awareness while perceiving the world through a robot's eyes, linking this to the broader challenges of understanding consciousness and perception.
"intelligence you know and there's a few moon shots which in that spirit which ones are you excited about what which projects kind of well of course I'm excited about one of the moon shots with it whic..."
The Joy of Collaborative Discovery
In this segment, Poggio shares insights on the importance of collaboration in scientific discovery. He emphasizes that curiosity and fun, along with a supportive environment, are crucial for success in science and engineering. He reflects on his experiences as a mentor and the value of enthusiasm in fostering innovative ideas.
"what the world is right so yeah so we may have to solve the hard problem of consciousness on their way yes but it's quite quite quite a moonshot eyes so if you've been an adviser to some incredible mi..."
Debate and Discourse in Science
Poggio discusses the role of debate and critical discourse in scientific progress. He contrasts the American and German academic cultures regarding directness in discussions, advocating for a respectful yet forthright exchange of ideas. This segment highlights the importance of constructive criticism in advancing knowledge.
"discovering something yes so as you've led many incredible efforts here what's the secret to being a good advisor mentor leader in a research setting is that similar spirit or yeah what what advice co..."
Intelligence: Gift or Curse?
Poggio contemplates whether intelligence is a gift or a curse in the context of happiness and the meaning of life. He reflects on the complexities of intelligence and its relationship with personal fulfillment, concluding that the pursuit of understanding intelligence is intertwined with deeper existential questions.
"rules that's great so if you somehow found yourself in a position to ask one question of an Oracle like a genie maybe a god whoa and you're guaranteed to get a clear answer what kind of question would..."