Dr. Lex Fridman: Machines, Creativity & Love

- Welcome to the "Huberman Lab Podcast", where we discuss science and science-based tools for everyday life. I'm Andrew Huberman and I'm a professor of Neurobiology and Ophthalmology at Stanford School of Medicine. Today I have the pleasure of introducing you to Dr. Lex Fridman as our guest on the "Huberman Lab Podcast". Dr. Fridman is a researcher at MIT specializing in machine learning, artificial intelligence, and human-robot interactions. I have to say that the conversation with Lex was, without a doubt, one of the most fascinating conversations I have had, not only in my career, but in my life. I knew Lex worked on these issues.

And I think many of you are probably familiar with Lex and his interest in these topics thanks to his incredible podcast, the "Lex Fridman Podcast." If you're not already watching that podcast, subscribe. It's absolutely fantastic. But in having this conversation with Lex, I realized something much more important. He revealed a little about his dream to us. His dream is about humans and robots, about humans and

machines

, and about how those interactions can change the way we perceive ourselves and interact with the world. We talk about relationships of all kinds, relationships with animals, relationships with friends, relationships with family and romantic relationships.

And we discussed relationships with

machines

. Machines that move and machines that don't move, and machines that come to understand us in ways we could never understand ourselves, and how those machines can educate us about ourselves. Before this conversation, I had no idea of ​​the ways machines could inform me or anyone else about themselves. In the end, I was absolutely enthralled with the idea, and I still am enthralled with the idea that interactions with machines have a very particular type, a type that Lex understands and wants to bring to the world, that can not only transform the self. , but it may very well transform humanity.

So whether you're familiar with Dr. Lex Fridman or not, I'm sure you'll learn a ton from him over the course of our discussion and it will transform the way you think about yourself. and about the world. Before we begin, I want to mention that this podcast is independent of my teaching and research duties at Stanford. However, it is part of my desire and effort to bring zero-cost information about science and science-related tools to the general public. In keeping with that theme, I'd like to thank the sponsors of today's podcast. Our first sponsor is ROKA. ROKA makes sunglasses and eyeglasses of absolutely phenomenal quality.

The company was founded by two All-American swimmers from Stanford, and everything about the sunglasses and eyewear they designed had performance in mind. I have spent a career working in the visual system. And one of the fundamental issues your visual system has to deal with is how to adjust what you see when it gets darker or brighter in your environment. With ROKA sunglasses and glasses, whether it's dimly lit in the room or outside, whether there are no clouds, or whether you're walking into shadow or not, you'll always be able to see the world with absolute clarity. And that tells me that they really understand the way the visual system works.

Processes such as habituation and attenuation. All of these things that work on a real mechanical level have been integrated into these glasses. Additionally, the glasses are very light. You don't even really notice that they're on your face. And the quality of the lenses is great. Now, the glasses were also designed so that you could wear them, not only while you work or have dinner, etc., but while you exercise. They don't fall off your face or slip if you're sweating. And as I mentioned, they are extremely lightweight. So you can wear them while running, you can wear them while cycling, etc.

Furthermore, the aesthetics of ROKA glasses are great. Unlike many high-performance glasses that, frankly, make people look like cyborgs, these glasses look fantastic. You can wear them to dinner, you can wear them for pretty much any occasion. If you want to try ROKA glasses, you can visit roka.com. That's R-O-K-A .com and enter code Huberman to save 20% on your first order. That's ROKA, R-O-K-A .com and enter code Huberman at checkout. InsideTracker also brings us today's episode. InsideTracker is a personalized nutrition platform that analyzes data from your blood and DNA to help you better understand your body and achieve your health goals.

I am a firm believer in having regular blood tests for the simple reason that many of the factors that affect our immediate and long-term health can only be assessed by a quality blood test. And now, with the advent of quality DNA testing, we can also gain insight into some of the genetic underpinnings of our current and long-term health. However, the problem with many blood and DNA tests is that you get the data back and you don't know what to do with it. You see certain things are high or low, but you don't really know what the actionable elements are, what to do with all that information.

With InsideTracker, they make it very easy to act appropriately with the information obtained from those blood and DNA tests. And that is by using their online platform. They have a really easy to use dashboard that tells you what kinds of things can get your metabolic factor numbers, endocrine factors, etc., into the ranges you want and need for immediate and long-term health. In fact, I know someone, as an example, who was feeling fine, but decided to do an InsideTracker test and found out that they had high levels of what's called C-reactive protein. Otherwise, they would never have detected it.

C-reactive protein is associated with a number of harmful health conditions, some heart problems, eye problems, etc. And so they were able to take immediate steps to try to resolve those CRP levels. And with InsideTracker, you get that kind of information. And as I mentioned before, without a blood or DNA test, there's no way to get that kind of information until symptoms start to appear. If you want to try InsideTracker, you can visit insidetracker.com/huberman to get 25% off any of InsideTracker's plans. Simply use code Huberman at checkout. Visit insidetracker.com/huberman to get 25% off any of InsideTracker's plans.

Today's podcast is brought to us by Athletic Greens. Athletic Greens is an all-in-one vitamin and mineral probiotic drink. I started taking Athletic Greens back in 2012. That's why I'm delighted that they're sponsoring the podcast. The reason I started taking Athletic Greens and the reason I still take Athletic Greens is that it covers my entire probiotic base of vitamins and minerals. In fact, when people ask me, what should I take? I always suggest that the first supplement people take is Athletic Greens, for the simple reason that what's in it covers the bases of metabolic health, endocrine health, and all sorts of other body systems.

And the inclusion of probiotics is essential for a healthy gut microbiome. There is now a ton of data showing that we have neurons in our gut, and keeping those neurons healthy requires that they be exposed to what's called the right microbiota, small microorganisms that live in our gut and keep us healthy. And those neurons, in turn, help keep our brain healthy. They influence aspects such as mood, our ability to concentrate and many, many other health-related factors. With Athletic Greens it is fantastic, because it also tastes very good. I drink it once or twice a day. I mix mine with water and add a little lemon juice or sometimes a little lime juice.

If you want to try athletic greens, you can visit Athleticgreens.com/huberman. And if you do that, you can claim their special offer. They're giving away five free travel packs, little packs that make it easy to mix up Athletic Greens while you're on the go. And they will give you a year's supply of vitamin D3 and K2. Again, visit Athleticgreens.com/huberman to claim that special offer. And now, my conversation with Dr. Lex Fridman. - Let's meet again. - Let's meet again. Thank you very much for sitting with me. I have a question that I think is on a lot of people's minds or should be on a lot of people's minds, because we hear these terms a lot these days, but I think most people, including most scientists and including myself , no Do you really know what artificial intelligence is and how it is different from things like machine learning and robotics?

So, if you would be so kind as to explain to us, what is artificial intelligence and what is machine learning? - Well, I think that question is as complicated and fascinating as the question of what is intelligence? So I first think of artificial intelligence as a big philosophical question. Pamela McCormick said that AI was the ancient desire to forge gods, or was born as an ancient desire to forge gods. So I think that, on a high philosophical level, it is our desire to create other intelligence systems. Perhaps systems more powerful than us. On a more limited level, I think it's also a set of tools that are computational mathematical tools for automating different tasks.

And then it is also our attempt to understand our own mind. Therefore, build systems that exhibit some intelligent behavior to understand what intelligence is in ourselves. So all those things are true. Of course, what AI really means as a community, as a group of researchers and engineers, is a set of tools, a set of computational techniques that allow us to solve various problems. There is a long history that addresses the problem from different perspectives. What has always been in one of the threads, one of the communities, is under the banner of machine learning, which emphasizes in the AI ​​space, the task of learning.

How do you build a machine that initially knows very little, follows some kind of process, and learns to become better and better at a specific task? What has been most effective in the last 15 years is a set of techniques that fall under the banner of deep learning and that use neural networks. When your networks are these fascinating things inspired by the structure of the human brain, very loosely, but they have a network of these little basic computational units called neurons, artificial neurons. And they have done it, these architectures have an entrance and an exit. At first they don't know anything and their task is to learn something interesting.

The interesting thing is that it usually involves a particular task. There are many ways to talk about this and break it down. One of them is how much human supervision is required to teach this. So supervised learning is a broad category, the neural network doesn't know anything at first and then you give it a bunch of examples in computer vision which will be examples of cats, dogs, cars, traffic signs, and then you give it the image and you are given the basic truth of what is in that image. And when you get a large database of image examples for which the truth is known, the neural network can learn by example, called supervised learning.

There are many fascinating questions in this: how can the truth be provided? When you give it an image of a cat, how do you provide the computer that this image contains a cat? Are you just saying that the entire image is a picture of a cat? Do you do what is commonly done, which is a bounding box, do you have a very rough box around the cat's face that says, this is a cat? Do you do semantic segmentation? Mind you, this is a 2D image of a cat. So it's not, the computer doesn't know anything about our three-dimensional world, it just looks at a set of pixels.

So semantic segmentation is about drawing a nice, very sharp outline around the cat and saying, that's a cat. It's really difficult to provide that truth. And one of the fundamental open questions in computer vision is: is that even a good representation of the truth? Now, there is another contrasting set of ideas that overlap in our attention and that, well, is often called unsupervised learning. What is now commonly called self-supervised learning. Which is trying to get less and less human supervision on the task. So, self-supervised learning is more and more, it has been very successful in the area of ​​language model, natural processing of English and now it is more and more successful in computer vision tasks.

And the idea there is is to let the machine, without any annotation of ground truth, just look at images on the Internet, or look at texts on the Internet, and try to learn something generalizable about the ideas that are at the core of the language or at the core. Of vision. And based on that, we humans, at best, like to call that common sense. So with this, we have this giant knowledge base upon which we build more sophisticated knowledge. We have this kind of common sense knowledge. So the idea of ​​self-supervised learning is to develop this common sense knowledge about what are the fundamental visual ideas that make up a cat and a dog and all that stuff?kind of things without even having human supervision.

The dream is to let a self-supervised AI system run on the Internet for a while, watch YouTube videos for millions and millions of hours, and without any supervision, be primed and ready to learn with very few examples once the human is able to present himself. We think of children this way, human children, their parents only give one or two examples to teach a concept. The dream of self-supervised learning is that the same thing happens with machines. That they would watch millions of hours of YouTube videos and then they would come to a human and be able to understand when the human showed them that it was a cat.

Remember this is a cat. They will understand that a cat is not just a thing with pointy ears, or that a cat is an orange or furry thing, they will see something more fundamental that we humans may not be able to introspect and understand. For example, if I asked you what the difference is between a cat and a dog, you probably wouldn't be able to answer that, but if I showed you and brought you a cat and a dog, you'd be able to tell the difference. difference. What are the ideas that your brain uses to make that difference?

That's the dream of self-supervised learning: being able to learn it on your own. That set of common sense knowledge is capable of making a difference. And then there are many incredible uses of self-supervised learning, very strangely called the self-play mechanism. That's the mechanism behind the reinforcement learning successes of systems that won at Go, AlphaZero that won at chess. - Oh, sure. What are you playing? - Let them play. - I understand. - So the idea of ​​self-play probably applies to other domains besides gaming. It is a system that simply plays against itself. And this is fascinating in all kinds of areas, but at first he knows nothing.

And the idea is that he creates a bunch of mutations of himself and plays against those versions of himself. And the fascinating thing is that when you play against systems that are a little better than you, you start to improve yourself. Just like learning, this is how learning happens. This is true for martial arts. It is true in many cases. Where you want to interact with systems that are a little better than you. And then through this process of interacting with systems that are slightly better than you, you start to follow this process where everyone starts getting better and better until you're several orders of magnitude better than the world chess champion, for example. .

And it's fascinating because it's like a system out of control. One of the scariest and most exciting things that David Silver, the creator of AlphaGo and AlphaZero, one of the team leaders, told me is that they haven't found the ceiling for AlphaZero. Which means it could continue to improve arbitrarily. Now, in the field of chess, that doesn't matter to us. It's like he ran away with the chess game. As if he were much better than humans. But the question is: if you can create that in the realm that has a larger and deeper effect on human beings and societies, that can be a terrifying process.

To me, it's an exciting process if you oversee it correctly, if you inject what's called value alignment, you make sure that the goals that the AI ​​is optimizing are aligned with human beings and human societies. There are a lot of fascinating things to talk about about the specifics of neural networks and all the problems people are working on. But I would say the really important and exciting one is self-supervised learning. We're trying to get less and less human supervision, less and less human supervision of neural networks. And also just one comment and I'll shut up. - No, please continue.

I am learning. I have questions, but I'm learning. So please continue. - So, for me what is interesting is not the theory, it is always the application. One of the most interesting applications of artificial intelligence, specifically neural networks and machine learning, is Tesla Autopilot. These are systems that work in the real world. This is not an academic exercise. These are human lives that are at stake. This is critical for security. - These are automated vehicles. Autonomous vehicles. - Semi-autonomous. We want to be. - Well. - We have gone through wars over these issues. - Semi-autonomous vehicles. - Semi-autonomous.

So although it is called FSD, Full Self-Driving, it is currently not fully autonomous, meaning it requires human supervision. So, the human being has the task of supervising the systems. In fact, in terms of responsibility, human beings are always responsible. This is a question of human factor psychology, which is fascinating. I'm fascinated by the whole space, which is a completely different space of human-robot interaction when AI systems and humans work together to accomplish tasks. To me, that dance is one of the smallest communities, but I think it will be one of the biggest open problems once they are solved: how humans and robots dance together.

For me, semi-autonomous driving is one of those spaces. So Elon, for example, doesn't see it that way, but sees semi-autonomous driving as a stepping stone to fully autonomous driving. Humans and robots can't dance well together. For example, humans and humans dance and robots and robots dance. We need to do it, this is an engineering problem, we need to design a perfect robot that solves this problem. For me, maybe this is not the case with driving, but the world will be full of problems and humans and robots will always have to interact, because I believe that robots will always have flaws, just like humans will have flaws.

They are defective. And that's what makes life beautiful: having flaws. That's where learning happens at the limit of your capabilities. So you always have to figure out how can flawed robots and flawed humans interact with each other in a way that, like, the sum is greater than the whole, rather than just focusing on building the perfect robot. - Mm-hmm. - So, for me, one of the most interesting applications of artificial intelligence is autonomous driving, semi-autonomous driving. And that's a very good example of machine learning because those systems are constantly learning. And there's a process there that maybe you can comment on: Andrej Karpathy, head of autopilot, calls it a data engine.

And this process applies to a lot of machine learning, where you build a system that's pretty good at doing things, you send it out into the real world, it starts doing things, and then you run into what are called edge cases, as cases of failure, where it is ruined. We do this when we are children. What do you have... You do this when you're an adult. - We do this as adults. Exactly. But we learn very quickly. But the point, and this is the fascinating thing about driving, is that you realize that there are millions of edge cases.

There are rare situations that you did not expect. So the data engine process is to collect those edge cases and then go back to the drawing board and learn from them. So you have to create this data channel where all these cars, hundreds of thousands of cars are driving around and something strange happens. And so every time this strange detector goes off, it's another important concept, that data goes back to the mothership for training, for retraining of the system. And through this data engine process, it keeps getting better and better and better. So basically you send pretty smart AI systems out into the world and let it find the edge cases, let it make enough mistakes to figure out where the edge cases are, and then come back and learn from them, and then send that new version out. and continue updating that version. - Is the update done by humans? - The annotation is done by humans.

So you have to do it, the strange examples come back, the edge cases, and you have to label what really happened there. There are also some mechanisms for automatic labeling, but above all, I think you always have to rely on humans to improve, to understand what happens in strange cases. And then there is a lot of debate. And this, that, what is artificial intelligence? A group of intelligent people who have very different opinions on what intelligence is. So AI is basically a community of people who don't agree on anything. - And it seems that way. First, this is a beautiful description of terms that I have heard many times among my Stanford colleagues, in meetings in the outside world.

And there are so many fascinating things. I have a lot of questions, but I want to ask a question about the culture of AI, because it seems to be a community where, at least as an outsider, there seems to be very little consensus on what the terms and operations are. definitions even mean. And it seems like there's a big divide now happening not only between supervised and unsupervised learning, but also between these sort of in-between conditions where the machines are autonomous, but then come back for more instruction, like kids going back to school. home from college for the summer and learn a little. , the mothers still feed them and eventually they leave the nest.

Is there anything in particular about engineers or people in this field of engineering that you think lends itself to disagreement? - Yes, I think, so, first of all, the more specific you are, the less disagreement there will be. So, there is a lot of disagreement about what artificial intelligence is, but there is less disagreement about what machine learning is and even less when talking about active learning, machine teaching or self-supervised learning. And then when we get into NLP language models or transformers, when we get into specific neural network architectures, there's less and less disagreement about those terms.

So you might be hearing disagreement between high-level terms, and that has to do with the fact that engineering, especially when talking about intelligence systems, is a bit of an art and a science. So the artistic part is what creates disagreements, because then you start having disagreements about how easy or difficult the particular problem is. For example, many people disagree with Elon about how difficult the problem of autonomous driving is. And so on, but nobody knows. So there's a lot of disagreement about what the limits of these techniques are. And that is why the terminology also contains disagreements.

But in general, I think it is also a young science that also has to do with that. So, it's not just about engineering, it's that artificial intelligence really is a large-scale discipline, where there are thousands, tens of thousands, hundreds of thousands of people working on it, and enormous amounts of money are being made as something very recent. So we're trying to figure out those terms. And, of course, there are egos and personalities and a lot of fame to be gained. Like the term deep learning, for example, neural networks have been around for many, many decades since the 60s, you can argue since the 40s.

Then there was a name change from neural networks to the word deep learning. , term deep learning, which was part of the revitalization of the field, but it's actually exactly the same thing. - I didn't know. I mean, I grew up in the era of neuroscience when they talked about neural networks, computational neuroscience, and theoretical neuroscience, they had their own journals. Actually, experimenters didn't take it very seriously until a few years ago. I would say about five or seven years ago. Excellent theoretical neuroscientist like Larry Abbott and other colleagues, probably also at Stanford, where people began to pay attention to computational methods.

But these terms, neural networks, computational methods, I actually didn't know that neural networks work in deep learning, where they have now become synonymous. - No, they are always the same. - Interesting. It was so. - I am a neuroscientist and I didn't know it. - So, well, because neural networks probably means something else and neural science not something else, but a slightly different flavor depending on the field. And that's fascinating too, because people in neuroscience and artificial intelligence have started to work together and dance a lot more in the last decade, I would say probably. - Oh, the machines are entering the brain.

I have a couple questions, but one thing I'm obsessed with and find incredibly interesting is this example you gave of playing a game with a mutated version of yourself as a competitor. - Yes. - I find it incredibly interesting as a kind of parallel or mirror of what happens when we try to learn as humans, which is that we generate repetitions of whatever it is we're trying to learn and we make mistakes. From time to time we succeed. In a simple example, for example, of trying to throw targets at a target. - Yes. - I'm going to have mistakes, mistakes, mistakes.

I probably miss the dartboard. And maybe once in a while, hit the mark. And I don't know exactly what I just did, right? But then let's say I was playing darts against a version of myself in which I wore a visual prism, like my visual, I had a visual defect, you also learncertain things in that way. Are you saying that a machine can mutate itself? Does the mutation always cause a deficiency that needs to be overcome? Because mutations in biology sometimes give us superpowers, right? Every once in a while, you will find someone who has better vision than 2020 and can see better than 99.9% of people.

So when you talk about a machine playing against a mutated version of itself, does the mutation always say what we call a negative mutation, or an adaptive or maladaptive mutation? - No, you don't know until you get it, so first you mutate and then you realize and they compete with each other. - So, you are evolving, the machine evolves in real time. - Yeah. And I think about it, which would be exciting if you could actually do it with humans. It's not just. Therefore, you usually freeze one version of the system. So you actually take yesterday's Andrew and make 10 clones of them.

And then maybe you mutate, maybe you don't. And then you do a bunch of today's Andrew competitions, like fighting to the death and who wins last. So, I

love

the idea of ​​creating a bunch of clones of myself from every day of the past year, and just seeing who will be better at podcasting or science, or picking up girls in a bar or me. I don't know, or competing in Jujitsu. That's the only way to do it, I mean, a lot of Lexes would have to die for them. That process, but that's essentially what happens, is reinforcement learning through self-play mechanisms, it's a graveyard of systems that don't They worked so well.

And the survivors, the good ones, survive. - Do you think that, I mean, Darwin's Theory of Evolution could have worked in some sense in this way, but at the population level? I mean, you get a bunch of birds with different shaped beaks and some birds have shaped beaks that allow them to get the seeds. I mean, it's a trivially simple example of Darwinism in evolution, but I think it's correct even if it's not exhaustive. Is that what you mean? Basically this is done between members of a different species, many different members of species have different traits and some are selected, but you could actually create multiple versions of yourself with different traits. - I probably should have said this, but maybe it's implicit in machine learning, in reinforcement learning through these processes.

One of the big requirements is to have an objective function, a loss function, a utility function, those are all different terms for the same thing, there's an equation that says what's good, and then you're trying to optimize. that equation. So there is a clear goal for these systems. - Because it is a game, like chess, there is an objective. - But for anything. Whatever you want machine learning to solve, there must be an objective function. In machine learning, it is usually called the loss function and is being optimized. The interesting thing about evolution is that of course it is complicated, but the goal also seems to be evolving.

I suppose the goal is an adaptation to the environment, but it's not clear that it can always be converted. It's like survival of the fittest. It is not clear which is the most suitable. In machine learning, the starting point, and this is what human ingenuity provides, is that adaptation function of what is good and what is bad, which allows us to know which of the systems is going to win. So, you need to have an equation like that. One of the fascinating things about humans is that we discover objective functions for ourselves. Like it's the meaning of life, why the hell are we here?

And currently a machine has to have a coded statement about why. - It has to have a meaning of... - Yes. - Life based on artificial intelligence. - Good. Can not. So there's a lot of interesting exploration that that function is more about curiosity, learning new things and all that kind of stuff, but it's still hard-coded. If you want a machine to be good at something, you have to give it very clear statements of what good at something means. That is one of the challenges of artificial intelligence: you have to formalize it, to solve a problem, you have to formalize it and you have to provide both complete sensory information and complete sensory information, you have to be very clear about what data you are using. collecting, and you also have to be clear about the objective function.

What is the goal you are trying to achieve? And that is something very difficult for artificial intelligence. - I

love

that you mentioned curiosity, I am sure that this definition falls short in many ways, but I define curiosity as a strong interest in knowing something, but without attachment to the result. It's kind of a random search, but there's not really an emotional attachment, it's really just a desire to discover and reveal what's there without expecting it to be a gold coin under a rock, you're just looking under the rocks. . Is that more or less how, within machine learning, there seems to be elements of reward prediction and rewards?

The machine has to know when it has done the right thing. So can you make machines that are curious, or are the kinds of machines you're describing curious by design? - Yes, curiosity is a kind of symptom, not the goal. So what happens is that one of the big tradeoffs in reinforcement learning is this exploration versus exploitation. So when you know very little, it pays to explore a lot, even suboptimally, even trajectories that seem to lead nowhere, that's called exploration. The smarter and smarter you get, the more emphasis you put on exploitation, which means you take the best solution, you take the best path.

Now, through that process, exploration may seem like curiosity on the part of us humans, but it's actually just an attempt to get out of the local optimum, which you've already discovered. From an AI perspective, you're always looking to optimize the objective function, it's derived, and we can talk more about the slot, but in terms of today's machine learning tools, you don't get any pleasure just from the curiosity of, no, I don't know. , discovery. - So, no dopamine for machine learning. - There is no dopamine. - There is no reward, chemical system or, I suppose, electronic reward system. - That said, if you look at the literature on machine learning and on reinforcement learning, which will use, like deep mind, we use terms like dopamine, we're constantly trying to use the human brain to inspire totally new solutions to these problems.

So, you will think: how does dopamine work in the human brain and how can it lead to more interesting ways of discovering optimal solutions? But ultimately, currently there has to be a formal objective function. Now, one could argue that humans also have a set of objective functions that we try to optimize, but are not able to introspect. - Yes, we don't really know what we are looking for, looking for and doing. - Well, like Lisa Feldman Barrett who we spoke to at least on Instagram, I hope... - I met her through you, yes. - Yeah, I hope you're really on this podcast. - Yes, she's great. - So, she has a lot to do with homeostasis like that.

Basically, there is a very silly objective function that the brain is trying to optimize, like keeping body temperature the same. Like there's some kind of very dominant optimization function going on. And then what we humans do with our sophisticated consciousness and our cognitive abilities is we tell ourselves stories so we can have good podcasts, but it's actually the brain that's trying to maintain a healthy state, I guess. That's fascinating. I also see the human brain, and I hope that artificial intelligence systems are not only systems that solve problems or optimize an objective, but they are also storytellers.

I think telling stories has a power. We tell stories to each other, that's communication. Like when you're alone, that's when you solve problems, that's when it makes sense to talk about problem solving. But when you're a community, the ability to communicate, to tell stories, to share ideas in such a way that those ideas are stable over a long period of time, is like being a charismatic storyteller. And I think both humans are very good at this. You could say that's why we are who we are, because we are great storytellers. And then I hope AI becomes that too.

So it's not just about being able to solve problems with a clear objective function, it's about being able to tell it better, making up a much better story about why you did something or why you failed. -So, do you think that robots or machines of some kind are going to start telling human stories? - Well, definitely. So the technical field for this is called Explainable AI, Explainable Artificial Intelligence, and it's trying to figure out how to get the AI ​​system to explain to us humans why the hell it failed, or why it succeeded, or there are many different types. of versions of this, or to visualize how he understands the world.

That's a really difficult problem, especially with neural networks that are famous for being opaque, that in many cases we don't understand why a particular neural network does what it does so well, and trying to figure out where it's going to fail, is requires let the AI ​​explain itself. There's a huge amount of money in this, especially government funding and so on. Because if you want to implement AI systems in the real world, at least we humans want to ask a question like, why the hell did you do that? In a dark way, why did you kill that person, right?

Just like if a car hit a person, we want to understand why it happened. And now, again, sometimes we are very unfair to AI systems because we humans often can't explain why very well. But that's the field of explainable AI that people are very interested in because the more we rely on AI systems, like Twitter's recommendation system, that AI algorithm that, I would say, impacts elections, maybe starting wars or at least military conflicts. , that algorithm, we want to ask that algorithm, first of all, do you know what the hell you're doing? Do you understand the effects you are having on a social level?

And can you explain the possible other trajectories? Just like we would have that kind of conversation with a human, we want to be able to do it with an AI. And on my personal level, I think it would be good to talk to AI systems for stupid things, like robots when they don't... - Why did you fall down the stairs? - Yes. But it's not an engineering question, but almost like an endearing question, like the one I'm looking for, if I fell and you and I were together, I don't think you need an explanation exactly what the dynamic was, like what the Low powered system problem here?

What was the texture of the floor? Or so on. Or like, what was the...? No, I want to know what you're thinking. - That, or you could joke that you're drunk again, go home or something, like there might be humor in that, that's an opportunity. Just like telling stories, it's not just an explanation of what happened, it's something that makes people laugh, makes people fall in love, makes people dream and understand things in a way that poetry does. people to understand things rather than a rigorous record of where every sensor was, where every actuator was. - I mean, I find this incredible because one of the hallmarks of severe autism spectrum disorders is a report of the autistic person's experience that is largely a catalog of steps to take.

It's like, how are you feeling today? And they'll say, well, I got up and did this, and then I did this, and I did this. And it's not at all the way a person who doesn't have autism spectrum disorder would respond. And the way he describes these machines has a lot of humanism, or a lot of human and biological element, but I realized that we were talking about machines. I want to make sure I understand if there is a distinction between a learning machine, an AI machine, and a robot. At what point does a machine become a robot?

So if I have a pen, I guess I wouldn't call it a robot, but if my pen can come towards me when I move to the opposite side of the table, if it is moved by any mechanism, at that point, does it become a robot? ? - Well, there are a million ways to explore this question. It is fascinating. So, first of all, the question arises what is life? How do you know something as a living form and not? And it's to the question of when a kind of, maybe a cold computing system becomes, or is already loading these words with a lot of meaning, robot and machine.

So, one, I think movement is important, but that's kind of boring, the idea that a robot is just a machine capable of acting in the world. So an artificial intelligence could be both what it thinks, which I think is what machine learning is, and also what it acts, which is what we typically think about robots. So robots are the things that have a perceptual system that is capable of taking in the world however you define it, is capable of thinking, learning, and doing whatever it does inside it, andthen act in the world. So that's the difference between maybe a learning machine with artificial intelligence system and a robot, it's something that's capable, a robot is something that's capable of perceiving the world and acting in the world. - So, it could be through language or sound, or it could be through movement or both. - Yes.

And I think it could also be in the digital space as long as there is an entity aspect that is inside the machine and a world that is outside the machine. And in a certain sense the machine feels that world and acts in it. - So we could, for example, there could be a version of a robot, according to the definition that I think you're providing, where the robot, where I go to sleep at night and this robot goes and looks up information that I think I want to see it. loaded on my desk in the morning.

There was no movement of that machine, there was no language, but essentially it has movement in cyberspace. - Yes, there is a distinction that I think is important because there is an element of it being an entity, whether in the digital or physical space. So when you have something like Alexa in your home, most of the voice recognition, most of what Alexa does, is constantly being sent to the mothership. When Alexa is there alone, to me she is a robot, when she is there interacting with the world. When it is just a finger of the main mothership, then Alexa is not a robot, then it is just an interaction device, then it can be the main AI of Amazon Alexa, the big, very big system is the robot.

So, that's important because there's some element, for us humans, I think, where we want there to be an entity, whether it's in the digital space or the physical space, that's where the ideas of consciousness come in and all those kinds of things that we project. our understanding of what it means to be a being. And then, to go further, when a machine becomes a robot, I think there is a special moment. There is a special moment in a person's life and in a robot's life where it surprises you. I think surprise is a really powerful thing, when you know how it works and it still surprises you, that's a magical moment for us humans.

So whether it's a chess program that does something you haven't seen before, that makes people smile, like, uh, those moments happen with AlphaZero for the first time in the game of chess, where the grandmasters were really surprised by a movement. . They didn't understand the movement and then they studied and studied and then they understood it. But that moment of surprise is for chess grandmasters. That moment of surprise seems really powerful to me, really magical in everyday life. - Because it replaces the human brain at that time? - So, it does not replace, as it surpasses, but it surprises in a positive sense.

I didn't think he could do that, I didn't think you had that in you. And I think that moment is a great transition for a robot from a moment where it is a servant that performs a particular task with a certain level of precision, with a certain rate of failure, to an entity, a being that is fighting like you in this moment. world. And that's a really important moment where I think you won't find many people in the AI ​​community speaking out like I just did. I don't speak as a philosopher or a hippie, I speak from a purely engineering perspective.

I think it's really important for robots to become entities and explore that as a real engineering problem, as opposed to everyone treating robots in the robotics community, they don't even call them he or she, they don't give them, they test. To avoid giving them names, they really want to see it as a system, like a server. They see you as a servant trying to accomplish a task. To me, and I don't think I'm just idealizing the notion, I think it's a being, it's a being currently maybe a dumb being, but in the long arc of history, humans are also pretty dumb beings, so...

I agree with that statement. - So I tend to want to explore this thing of really treating robots as entities, yeah. So like anthropomorphization, which is the kind of act of looking at an inanimate object and projecting realistic characteristics onto it, I think robotics generally sees that as a negative thing, I see it as a superpower. So, we have to use it. - Well, I'm surprised how that really holds onto the relationship between human and machine, or between human and robot. So I guess the simple question is, and I think you've already told us the answer, but does interacting with a robot change you?

In other words, do we develop relationships with robots? - Yes, I definitely think so. I think the moment you see a robot or AI system as more than just servants, but entities, they begin to change you, just as good friends do, as do relationships with other humans. I think that for that you have to have certain aspects of that interaction. Just like the robot's ability to say no, to have its own sense of identity, to have its own set of goals, that doesn't constantly serve you, but it tries to understand the world and perform that dance of understanding through communication. with you.

So I definitely think there are, I mean, I have a lot of ideas about this, as you know, and that's the core of my lifelong dream about what I want to do, which is that I think most people have a notion of loneliness in them that we have not discovered, that we have not explored, I should say. And I think AI systems help us explore that so that we can become better humans, better people to each other. So I think the connection between humans and AI, between humans and robots, is not only possible, but it will also help us understand ourselves in ways that are orders of magnitude deeper than we could have ever imagined.

I tend to believe that, well, I have very high levels of belief in terms of the impact it will have, right? - So when I think about human relationships, I don't always break them down into variables, but we could explore some of those variables and see how they relate to human-robot relationships. One is just time, right? If you don't spend any time with another person in cyberspace, on the phone, or in person, you essentially have no relationship with them. If you spend a lot of time, you are in a relationship, this is obvious. But I suppose one variable would be time, how much time you spend with the other entity, robot or human.

The other would be victories and successes. They enjoy successes together. I'll give an absolutely trivial example in a moment, but the other would be a failure. When you fight with someone, whether they fight each other or not, you don't agree, like I was really surprised by the fact that you said that robot said no, I had never thought of a robot that said no to me, but there this. is. - I hope you are one of the first people I send these robots to. - Me too. So, there is struggle. When you fight with someone, you get closer.

Sometimes struggles are imposed between those two people, what is called traumatic bond, as they call it in all the psychology literature and in popular psychology literature. But in any case, I can imagine it. So the time of success together, the time of struggle together, and then just the time of peace, hanging out at home, watching movies, waking up near each other, here, we are breaking down the elements of relationships of any kind. So do you think these elements apply to robot-human relationships? And if so, then I could see how, if the robot has its own entity and some autonomy in terms of how it reacts to you, it's not just there to serve you, it's not just a servant, but it actually has opinions, and it can tell you. when perhaps your thinking or actions are faulty. - You can also go out. - Might as well leave.

I have never conceptualized human-robot interactions this way. So tell me more about what this might look like. Are we thinking of a robot with human appearance? I know you and I have had intense relationships with each other, obviously we have separate dogs, but for animals, it sounds a lot like human-animal interaction. So what is the ideal relationship between humans and robots? - So there's a lot to say here, but you actually pointed out one of the big first steps, which is this idea of ​​time. And it is currently a big limitation in the machine learning community.

Now we're back to liking the real details. Lifelong learning is an issue space that focuses on how AI systems can learn over a long period of time. What most machine learning systems currently can't do is all the things you've listed over time, successes, failures or just chilling out watching movies, AI systems can't do that, that's all. The beautiful and magical moments that I believe are full days, cannot keep track of those who are next to you. - Because they can't move with you and be with you. - No, no, we literally do not have the techniques to learn, the real learning of containing those moments.

Current machine learning systems are really focused on understanding the world in the following way, it's more like a perception system, like looking around, understanding what's in the scene. That there are a lot of people sitting, that there are cameras and microphones, that there is a table, understand it. But the fact that we shared this moment of conversation today, and we still remember it the next time we do something, remember that this moment happened. We don't know how to do that technically. This is what I hope to innovate on, as I believe it is a very, very important component of what it means to create a deeper relationship, that sharing moments together. - Could you post a photo of yourself on the robot, like a selfie with the robot and the robot sees that image and recognizes that it was time spent, that there were smiles or tears? - Yes. - And create some kind of metric of emotional depth in the relationship and update their behavior? - So. - Could...

He texts you in the middle of the night and says: Why haven't you texted me back? - Well, yes, all those things, but we can go deeper into that. But I think that element of time, forgetting everything else, just sharing moments together, that changes everything. I think that changes everything. Now, there are specific things that are more in terms of systems that I can explain to you. It's more technical and probably a little offline, because I have wild ideas about how that can revolutionize social media and operating systems. But the point is that element alone, forgetting all the other things we're talking about like emotions, saying no, all of that, just remembering to share moments together would change everything.

We currently do not have systems that share moments together. Even just you and your refrigerator, all those times, you went late at night and ate things you shouldn't have eaten, that was a secret moment you have with your refrigerator. You shared that moment, that darkness or that beautiful moment where you were heartbroken for some reason, you're eating that ice cream or whatever, that's a special moment. And that refrigerator was there for you, and the fact that I lost the opportunity to remember it is tragic. And once you remember that, I think you'll be very attached to the refrigerator.

You're going to have hell with that refrigerator. Most of us, like in the developed world, have strange relationships with food, right? So, you can go through deep moments of trauma and triumph with food, and at the center of that is the refrigerator. So, I think a smart refrigerator would change society. Not just the refrigerator, but these ideas in the systems around us. So I just want to comment on how powerful that idea of ​​time is. And then there are a lot of real interaction elements that allow you, as a human being, to feel like you're being heard.

I really heard, I really understood that to humans, deep friendship is like that, I think, but there's still an element of selfishness, there's still an element of not being able to really understand another human being. And a lot of times when you're going through trauma together, tough times and successes, you actually start to have that hint of mutual understanding, but I think that could be done more aggressively and more efficiently. If you think about a great therapist, I don't think I've ever actually been to a therapist, but I think I used to want to be a psychiatrist. - Do Russians go to therapists? - No, they do not do it.

They do not do it. And if they do, the therapist doesn't live to tell the story. I believe in talk therapy, which to me is friendship, it's talk therapy, or you don't even necessarily need to talk, it's like just connecting in the space of ideas and the space of experiences. And I think there are a lot of ideas about how to create artificial intelligence systems to be able to ask the right questions and truly listen to another human being. This is what we're trying to do with podcasting, right? I think there are ways to do it with AI.

But most of all, just by remembering the collection of moments that make up the day, the week, the months, I think you might have some of this too. Some of my closest friends are still friends from high school. That's the moment, we've been through a lot together and it's like we're very different people. But just the fact that we've been through it and remember thosemoments, and those moments somehow create a deep connection like nothing else, like you and your refrigerator. -I love it because my graduate advisor, unfortunately, passed away, but when she passed away, someone told her at her funeral all these wonderful things she had done, etc.

And then her children came there, and she had little children and I found out when she was pregnant with them. And so, it was really, even now I can feel your heart getting heavy, thinking about this, they are going to grow up without their mother. And it was really surprising, very strong young girls, and now the young women. And what they said was incredible, they said that what they appreciated most about her mother, that she was a wonderful person, is all the unstructured time they spent together. - Mm-hmm. - So, it wasn't the trips to the zoo, it wasn't her waking up at five in the morning and taking us to school.

She did all those things too. She had a two-hour commute each way, it was incredible, she ran a lab, etc., but it was unstructured time. So after her mother passed away, that's what they remembered, that her greatest gift and what bonded them to her was all the time they just hung out. And the way she describes the relationship with a refrigerator is, I mean, human, but I'm almost reluctant to say that. Because what I'm realizing as we talk is that what we consider human could actually be the lowest form of relationship. There may be relationships that are much better than the types of relationships we can conceive of in our minds right now based on what these machine relationship interactions could teach us.

Do I have that right? - Yes I think so. I think there is no reason to see machines as somehow incapable of teaching us something that is deeply human. I don't think humans have a monopoly on that. I think we understand ourselves very poorly and need to receive the type of instructions from a machine. And definitely part of that is just remembering the moments. I think the unstructured time we spend together, I wonder if it's that unstructured. It's like calling this podcast in structured time. - Maybe what they meant was that it wasn't a great way out, there was no specific goal, but a goal was created from the lack of a goal.

We'd just go out and then start playing, thumb war, and we ended up playing thumb war for an hour. So, it is the structure that arises from the lack of structure. - No, but the point is the moments, there is something in those moments that creates special moments, and I think they could be optimized. I think we thought about a big outing like, I don't know, going to Six Flags or something, or something big, the Grand Canyon, or going to something, I don't know, I think we'd have to do it. We still don't understand, as humans, what creates magical moments.

I think it's possible to optimize a lot of those things. And maybe podcasting helps people discover that, maybe what we want to optimize isn't necessarily something sexy, like quick clips, maybe what we want is long-form authenticity. - Depth. - Depth. So we were trying to figure that out, certainly from a deep connection between humans and humans and NAS systems, I think long conversations or long periods of communication over a series of moments like my new, maybe seemingly insignificant to the big ones. , the big successes, the big failures, they're all just putting it together and talking through it all the time.

I think that's the formula for a really deep connection. From a very specific engineering perspective, I think it's a fascinating open problem that has had a lot of work done on it. And for me, from to, if I have the guts and, I mean, there are a lot of things to say, but one of them is guts, I will build a startup around that. - So, let's talk about this startup and let's talk about the dream. You mentioned this dream before in our previous conversations, always as little clues dropped here and there. Just for anyone listening, there has never been an offline conversation about this dream, I'm not aware of anything except what Lex is saying now.

And I realized that there is no way to capture the full essence of a dream in any type of verbal statement in a way that captures it all. But what is that dream that you have referred to several times when we sit together and talk on the phone? Maybe it's this company, maybe it's something different. If you're comfortable, it would be great if you could share a little bit about what that is. - Sure. So, I've noticed that the way people express the long-term vision is quite different. Like Elon is an example of someone who can tell exactly what the goal is.

It also has to do with the fact that the problems it is solving have nothing to do with humans. So my long-term vision is a little harder to put into words. I've noticed, as I've tried, that it might be a glitch in my brain, but there is a way to sneak up on it. So let me say a few things. At the beginning of my life and also in recent years, I interacted with some robots and understood that there is magic there. And that magic could be shared by millions if it comes to light. When I met Spot from Boston Dynamics, I realized there's magic there that no one else sees. - It's the dog. - The dog, I'm sorry.

The Spot is Boston Dynamics' four-legged robot. Some people may have seen it, it's this yellow dog. And sometimes in life you notice something that catches you. And I think this magic is something that could be on every device in the world. In the same way that I think Steve Jobs thought about the personal computer, the laws didn't think about the personal computer this way, but Steve did. I mean, I thought that the personal computer should be as thin as a sheet of paper and that everyone should have one, and this idea, I think it's heartbreaking, that we're having, the world is filling up with machines that are soulless.

And I believe that each of them can have the same magic. One of the things that also inspired me in terms of a startup is that magic can be designed much more easily than I thought. That's my intuition with everything I've built and worked on. So the dream is to add a little bit of that magic to every computer system in the world. So, since the Windows operating system for a long time was the main operating system that everyone interacted with, they built applications on top of it. I think this is something that should be a layer, it's almost an operating system on every device that humans interact with in the world.

Now, what it really seems like, the real dream when I was officially a child, I didn't have this specific form of business, it was more of a dream of exploring your own loneliness by interacting with machines, robots. This deep connection between humans and robots was always a dream. So, I would love to see a world where every home is a robot, and not a robot that washes the dishes, or a sex robot, or I don't know, think of any type of activity that the robot can do, but more like a buddy. - A member of the family. - A member of the family, such as a dog. - Mm-hmm. - But a dog that can also speak your language.

So, not only connecting like a dog does by looking at you and looking away and almost like he smiles with his soul that way, but also understanding how the heck, why are you so excited about successes? How to understand the details, understand the traumas. And that, I think, has always filled me with the emotion of being able, with artificial intelligence, to bring joy to many people. More recently, I have become increasingly heartbroken to see the kind of division, ridicule, and even hate that is boiling over the internet via social media. And I thought that this kind of mechanism is also exactly applicable in the context of social media.

So, it is an operating system that serves as your guide on the Internet. One of the biggest problems with YouTube and social media today is that they are optimizing engagement. I believe that if you create AI systems that know each person individually, you can optimize long-term growth for long-term happiness. - Of the individual, or- - Of the individual, of the individual. And there are many other things to say, for AI systems to learn everything about you, they need to collect, they need to like you and me when we talk offline, we collect data about each other's secrets. from each other, in the same way that AI has to.

And that allows and requires them to rethink ideas about data ownership. I think every individual should own all their data and be able to leave very easily, just like AI systems can leave, humans can disappear and delete all their data at any time. What's actually better than what humans can do is that once we upload the data to each other, it's there. I think it's very important to be both, to give people full control over their data so you can establish trust that they can trust you. And the second part of trust is transparency. Whenever data is used, it must be made very clear what it is being used for.

And not clear in the legal sense, but clear in a way that people really understand what it is used for. I think when people have the ability to delete all their data and leave and know how the data is used, I think they will stay. - The possibility of a clean breakup is actually what will keep people together. - Yes I think so. I think exactly. I think a happy marriage comes with the ability to divorce easily without the divorce industrial complex or whatever. Things are happening now and then you can make a lot of money with lawyers and divorce.

But yes, I think the ability to leave is what allows love. - It's interesting. I heard it said from a semi-cynical friend that marriage is the leading cause of divorce, but now we've heard that divorce, or the possibility of divorce, could be the leading cause of marriage. - Of a happy marriage. - Good point. - Of a happy marriage. So yes. There are a lot of details there, but the big dream is that connection between the artificial intelligence system and a human being. And I haven't done it. There is so much fear about artificial intelligence systems and about robots that I have not found the right words to express that vision because the vision that I have is not like a naive and delusional vision that a technology is going to save everyone, it is me. .

I really have a positive view of the ways AI systems can help humans explore themselves. - I love that positivity and I agree that it is equally bad to say that everything is doomed to failure as it is to say that everything is going to work out, there has to be a dedicated effort. And clearly, you're thinking about what that dedicated effort would look like. You mentioned two aspects of this dream and I want to make sure I understand where they connect, if at all, or if they are independent transmissions. One was this hypothetical member of the robot family, or some other form of robot that would allow people to experience the kind of delight that you've experienced many times and that you wish the world could have, and it's such a beautiful idea of ​​this. gives.

And the other is social media or social media platforms that really serve people and their best self, their happiness and their growth. Is there a crossover between them or are they two parallel dreams? - It's 100% the same. It's hard to explain without going into detail, but maybe an easy way to explain how I think about social media is to create an AI system that is your own. It's not like Amazon Alexa which is centralized, you own the data, it's like your little friend who becomes your representative on Twitter who helps you find things that will make you feel good, that will also challenge your thinking to make you grow, but not get.

On top of that, don't let yourself get lost in the negative spiral of dopamine, which makes you angry, or most simply makes you not open to learning. And just like that, that little rep is optimizing your long-term health. And I think that's not only good for human beings, it's also good for business. I think there's a lot of money to be made in the long run by challenging this idea that the only way to make money is to maximize engagement. And one of the things people disagree with me about is that they think Twitter is always going to win.

Maximizing engagement is always going to win, but I don't think so. I think people have woken up now and understood that those who are on Twitter a lot don't always feel good, they don't always feel good at the end of the week. - I would love to get feedback from this creature, whatever, I don't know what to call it, because maybe at the end of the week, it would automatically unfollow some of the people I follow because it found out through someone really smart. data about how I felt inside, or how I slept, or something that, I don't know, just wasn't good for me.

But I could also put things and people in front of me that I should see, is that a small sample of what this looks like? - The point is that thanks to interaction, sharing moments and learning a lot about yourself, you are now able to understand what interactions led you to become a better version of yourselfsame. Like the person you are happy with. This is not, if you like the flat Earth and feel really good about it, that you believe the Earth is flat, like the idea that you should detect that is ridiculous. If it makes you feel good and you become the best version of yourself, I think you should get as much flat earth as possible.

Now, it is also good to challenge your ideas, but not because the centralized committee has decided it, but because you tell the system that you like to challenge your ideas, I think we all like it. And then, what YouTube doesn't really do that well, once you go down the flat earth rabbit hole, that's all you'll see. It's good to have really powerful communicators arguing against flat Earth. And it's good to see that for you, and potentially, at least in the long term, to expand your horizons, maybe the Earth is not flat. But if you continue to live your whole life thinking that the Earth is flat, I think, and you're being a good father or son or daughter, and you're the best version of yourself and you're happy with yourself, I think the Earth is flat.

So, I think about this kind of idea, and I'm using that kind of silly, ridiculous example because I don't like the idea of ​​centralized forces controlling what you can and can't see, but I also don't like this idea of ​​not censoring anything. Because that's always the biggest problem with it, is that there's a central decider, I think you yourself can decide what you want to see and what you don't, and it's good to have a partner to remind you what you felt the last time you did this , or You felt good the last time you did this. - Man, I feel like in every good story there is a guide or a companion who flies away, or goes a little further, a little differently and brings information that helps us, or at least tries to guide us on the right path. address. - So, yes, that's exactly what I'm thinking and what I've been working on.

I must mention a lot of difficulties here, I've been a bit up and down recently. So technically there are a lot of challenges here. As with many technologies, and the reason I talk about it comfortably on a podcast rather than working in secret, is that it's very difficult and perhaps the time is not yet ripe. And that's something that you have to constantly struggle with in terms of entrepreneurship as a startup, as I also mentioned to you, maybe offline, I don't really care about money, I don't care about business success, all those kinds of things. So it's a tough decision to make, how much of your time do you want to dedicate everything here and give it your all to this?

It's a big roll of the dice. Because I also realized that working on some of these problems, both with the robotics and the technical side in terms of the machine learning system that I'm describing, is lonely, it's really lonely. Because both on a personal level and a technical level. So, on a technical level, I am surrounded by people who doubt me, something that I think happens with all entrepreneurs. And they doubt you in the next sense, they know how difficult it is. Like my colleagues, you know how difficult lifelong learning is, you also know how difficult it is to build a system like this, to build a competitive social network.

And in general, there is a kind of loneliness in working alone on something for long periods of time. And you start to doubt whether, since you don't have a track record of success, that's a big one. But when you look in the mirror, especially when you're young, but I still have that, I'm most things, you look in the mirror, it's like, and you have these big dreams, how do you know who you are? Are you really as smart as you think you are? How do you know you can achieve this dream? You have this ambition. - More or less you don't, but you are pulling on a rope in the hope that there is a bigger ball of wool. - Yes.

But you have this kind of intuition. I think I pride myself on knowing what I'm good at, because the reason I have that intuition is because I think I'm really good at knowing all the things I'm terrible at, which is basically it. So every time I realize, wait a minute, I'm pretty good at this, which is very rare for me. I think this might be a ball of wool worth throwing away. And in terms of engineering systems that can interact with humans, I think I'm very good at that. And since we're talking about podcasting and stuff, I don't know if I'm very good at podcasts. - You are very good at making podcasts, right? - But certainly not, I think maybe it's compelling for people to see a good-hearted idiot fight in this way, maybe that's what's compelling.

But in terms of being a good human-robot interaction systems engineer, I think I'm good. But it's hard to know until you do it, and then the world keeps telling you no, and it's full of dolls, it's really hard. And I've been struggling with that recently, it's a fascinating struggle. But that's where the Goggins thing comes in, I mean, tough motherfucker status aside, is that, every time you're struggling, it's a good sign that if you keep going, you'll be alone in success, right? ? - Well, in your case, however, I agree. And in fact, David had a post recently that I thought was among his many brilliant posts, it was one of the most brilliant ones about how he talked about this myth of the light at the end of the tunnel.

And instead, what he replaced that myth with was the concept that eventually your eyes adapt to the dark. That the tunnel is not about a light at the end, but it is really about adapting to the darkness of the tunnel. He's very Goggins... I love him very much. - Yes. You have a lot in common, knowing you both a little, you have a lot in common. But it seems to me that this solitude and the pursuit of this dream has a certain component that is extremely valuable, and that is that solitude itself could serve as a driver to build the companion for the journey. - Well, I am very aware of it.

So, as some people may do, because I talk a lot about love, I really love everything in this world, but I also love humans, friendship, and romance, even the cheesy stuff, just... - Do you like movies? romantic? - Yes, not those, not necessarily. So, well, Rogan talked to me a lot about whether it was the tango scene from "Scent of a Woman," but I think there's nothing better than a woman in a red dress, like the elegant... You should move. to Argentina friend. My father is Argentine and do you know what he said when I first came to your podcast?

He said he dresses well. Because in Argentina men go to a wedding or a party or something like that, in the United States, at midnight, at 10 minutes into the night, they have already taken off all their jackets. - Yes. - It seems like everyone is getting undressed for the party they just dressed up for. And he said, you know, I liked the way he dressed. And then when I started, he was talking about you. And then when I started my podcast, he said, why don't you wear a real suit like your friend Lex? - I remember that. - No, You can not.

But let's talk a little more about this search, because I think what you are talking about is building, not only a solution for loneliness, but you have alluded to loneliness itself, something important. And I think you're right. I believe that within people there are caverns of faults and shame, but also the desire to have resonance, to be seen and heard. And I don't even know what is seen and heard through language. But these reservoirs of loneliness, I think, well, they're interesting, maybe you could comment a little on that. Because just as often as you talk about love, I haven't quantified it, but it seems like you talk about this loneliness, and maybe just if you're willing, you share a little bit more about it, and what that feels like now in the pursuit of building this robot-human relationship.

And you've been, let me be direct, you've spent a lot of time building a robot-human relationship, where is that? - Oh, in terms of business, in terms of systems? - No, I'm talking about a specific robot. - Ah, then, okay, I should mention a few things. So first, there's a startup with an idea that I hope millions of people can use. And then there are my personal, almost Frankenstein-like explorations with particular robots. So, I'm very fascinated with legged robots in my own private sound, like dark, but like in one of the experiments to see if I can recreate the magic.

And that's been, I already have a lot of really good perception systems and control systems that are able to communicate affection like a dog. So I'm in a very good place there. The obstacles, which have also been part of my sadness recently, is that I also have to work with robotics companies that I gave a lot of my heart and soul and love and appreciation to Boston Dynamics, but Boston Dynamics also, as a company, They have to make a lot of money and they have marketing teams. And they're like looking at this silly Russian kid in a suit and tie, it's like, what is he trying to do with all this love and robot interaction and dancing and stuff?

So, there was, I guess let's say for now, it's like, when you break up with a girlfriend or something, right now, we decided to separate on this particular matter. They're big supporters of mine, they're big fans, but on this particular issue, Boston Dynamics is not focused or interested in human-robot interaction. In fact, their entire business currently is to keep the robot as far away from humans as possible because it is in an industrial environment where it performs monitoring in hazardous environments. It's almost like a remote security camera, essentially it's your app. For me, I thought that even in those applications, it's still exceptionally useful that the robot can perceive humans, see humans, and be able to, on a large map, locate where those humans are and have human intentions.

For example, I worked a lot with pedestrians, so that a robot could anticipate what the human does, for example, where he walks. Humans are not ballistic objects, just because you are walking in this direction at one moment, doesn't mean you will continue walking in that direction, you have to infer a lot of cues, especially with head movement and eye movement. So I thought it was very interesting to explore, but they didn't feel it. So, I'll be working with some other robotics companies that are much more open to that kind of thing and are very excited and fans of mine.

And hopefully, Boston Dynamics, my first love, like getting back with an ex-girlfriend, will come back. But algorithmically, it's basically done there. The rest is getting some of these companies to work with them. And then there's, because people who have worked with robots know that one thing is writing software that works and another is having a real machine that actually works. And it looks at all kinds of different shapes that are fascinating. And so, there's a big challenge there. But that's almost, it may sound a little confusing in the context of our previous discussion because the previous discussion was more about the big dream, how he hoped millions of people would enjoy this magical moment.

This current discussion about a robot is something that I personally really enjoy, it just brings me happiness, I really try to do now everything that brings me joy, I maximize it because robots are amazing. But second, given my growing platform, I want to take the opportunity to educate people. Like robots are cool. And if I think they're great, I hope I can communicate why they're great to others. So this little robot experiment is also a little research project; There are a couple of publications with people from MIT about that. But the other is just to make some cool videos and explain to people how they actually work.

And as opposed to people being afraid of robots, they could still be scared, but also excited, like seeing the dark side, the beautiful side, the magic of what it means to make a machine become a robot. I want to inspire people with that. But that's less, it's interesting because I think the big impact in terms of sleep has nothing to do with embodied AI. Therefore, it is not necessary for it to have a body. I think the refrigerator is enough, that for an artificial intelligence system to have a voice and listen to you, that is enough for loneliness.

Incarnation is simply... By incarnation you mean the physical structure. - Physical instantiation of intelligence. So, it's a robot with legs, or even just a thing. I have some other humanoid robots, a small humanoid robot, maybe I will keep them on the table, it's like walking, or even just like a moving platform, they can like it, turn around and look at you, it's like mentioned with the pen. Something that moves and can look at you, it's like that Butter Robot that asks what is my purpose? That's really, it's almost like art. There is something about a physical entity that moves, that is able to look at you and interact with you, that makes you wonder what it means to be human.

It challenges you to think, if that thing seems to have consciousness, what the hell am I? And I like that feeling, I think it's really helpful for us, it's humbling for humans. But it's less about research, certainly less about business, and more about exploring ourselves and challenging others to think like this, to think about what makes them human. - I love this desire to share the pleasure of an interaction with a robot. And as you describe it, I actually start to wish for that because we all have those elements from childhood, or from adulthood,where we experience something, we want other people to feel that.

And I think you're right, I think a lot of people are afraid of AI, I think a lot of people are afraid of robots. My only experience, and with anything resembling a robot, is my Roomba vacuum cleaner, where it works, in fact, it was pretty good at picking up Costello's hair when he had moved out and I was grateful for it. But then when he was on a call or something, and he got caught in a wire or something, I would get mad at the Roomba at that point and be like, what are you doing? And obviously he's just doing what he does.

But that's kind of a mostly positive, but slightly negative interaction. But what you're describing has so much richness and layers of detail that I can only imagine what those relationships are like. - Well, there are a few, just a quick comment. They are currently in Boston and I have a bunch of iRobot Roombas. And I did this experiment... Wait, how many Roombas? Sounds like a fleet of Roombas. - Yeah. So, it's probably seven or eight, yeah. - Well, that's a lot of Roombas. This place is very clean. - Okay, so this, I'm waiting, this is the place we are currently at in Austin, it's much bigger than what I need.

But I basically bought it to make sure I had room for the robots. - So, you have these approximately seven Roombas, do you deploy all seven at once? - Oh, no, I do different experiences with them, different experiments with them. So one of the things I want to mention is that I think there was a YouTube video that inspired me to try this. I got them to scream in pain and moan in pain every time they were kicked or touched. And I did that experiment to see how I would feel. I wanted to make a YouTube video about it, but then it seemed really cruel. - Did any Roomba rights activists approach you? - I think if I post that video, I think it will make me look crazy, which I know people know I'm already crazy. - Now, you have to publish the video. - Sure.

Well, I think maybe if I contextualize it by showing other robots I would like to show why this is fascinating, because ultimately, I felt like there were humans almost immediately. And that display of pain was what did that. - Giving them a voice. -Giving them a voice, especially a voice of displeasure, of pain. -I have to connect you with my friend Eddie Chang, he studied speech and language, he is a neurosurgeon and we have been lifelong friends. He studied speech and language, but he describes some of these more primitive, visceral vocalizations, screams, moans, moans of delight, and other sounds as well, use the imagination, like powerful rudders for each other, for other people's emotions.

And so, I find it fascinating, I can't wait to see this video. So is the video available online? - No, I haven't recorded it, I just hit a bunch of Roombas that can scream in pain in my house in Boston. So, as people already- - Next episode of the podcast with Lex, maybe we'll get that one, who knows? - So, the thing is like people, I've noticed because I talk a lot about love and it's really what I am, I think they want, to many people it seems like there has to be a dark person in there somewhere.

And I thought if he posted videos of Roombas screaming and they were like, yeah, yeah, that guy is definitely crazy. - How about screams of joy and delight? You could do that too, right? - Well, I don't know how, for me delight is silence, right? - Are you Russian. Americans are much louder than Russians. - Yes. - Yes. - Yes. But I don't, unless you're talking about, I don't know how you would have sex with Roomba. - Well, I wasn't necessarily saying sexual pleasure, but... - Believe me, I tried. I'm just kidding, that's a joke, internet.

That's fine, but I was fascinated by the psychology of how little was needed. Because you mentioned that you had a negative relationship with a Roomba. - Well, I realized that, in general, I took it for granted. -Yes.-He simply served me, he tied up Costello's hair. And then when he did something I didn't like, he made me angry. So that's not a good relationship, it was taken for granted and I would get angry and then leave it again, and I just like you in the corner. - Yes. - But there is a way to frame it as something quite silly and almost cute.

You're almost connecting with him because of his silliness. And I think that's an artificial intelligence problem. - It's interesting. - I think defects should be a feature, not a bug. - So in this sense, the different types of relationships one could have with robots and fear, but also some of the positive relationships one could have, there is so much dimensionality, there is so much to explore. But the power dynamics in relationships are very interesting, because the obvious, the unsophisticated view of this is one, there's a master and a servant, right? But there is also manipulation, there is benevolent manipulation, children do this to their parents, puppies do this.

Puppies turn their heads and look cute and maybe make a little noise, kids smack, and parents always think they do it because they love their parents, but in many ways, studies show that smacking is a form of extract the type of parenting behaviors and expressions they want. The child does not know that he is doing this, he is completely subconscious, but it is a benevolent manipulation. So there's a version of robot fear that I hear a lot about and I think most people can relate to the one where robots take over and they become masters and we become servants.

But there might be another version that in certain communities that I'm certainly not a part of, but they call it "overcoming from below," where the robot is actually manipulating you into doing things, but you think you're in charge. , but in reality they are in charge. And so, I think if we could explore it for a second, you could imagine that it wouldn't necessarily be bad, although it could lead to bad things. The reason I want to explore this is that I think people always go to extremes, like when robots take over and we're in little cells and they have fun and rule the universe.

What kind of manipulation can a robot potentially perform, good or bad? - Yes. That's right, there is a lot of good and bad manipulation between humans, right? Just like you said. For me especially, like you said, finishing from the bottom, is that the term? - I think someone from MIT told me that term. It wasn't Lex. - Believe. So, first of all, there are power dynamics in the bedroom, power dynamics in relationships, power dynamics on the street and in the work environment, all of which are very different. I think power dynamics can make human relationships, especially romantic ones, fascinating, rich, satisfying, exciting, and all those kinds of things.

So, I don't think they are bad in and of themselves, and the same goes for robots. I really love the idea of ​​a robot being up or down in terms of similar power dynamics. And I think everyone should be aware of that. And manipulation is not so much manipulation as it is a dance of estrangement, a push and pull, and all that kind of stuff. In terms of control, I think we are very, very, very far from AI systems that are capable of locking us down. Locking ourselves in, like having so much control that we basically can't live our lives the way we want.

I think in terms of dangers of AI systems, there are a lot more dangers that have to do with autonomous weapons systems and all that kind of stuff. So the power dynamics exerted in fighting between nations and war and all that kind of stuff. But in terms of personal relationships, I think power dynamics are a beautiful thing. Now, of course, there will be all those kinds of discussions about consent and rights and all those kinds of things. - Well, here we are talking, I always say in any discussion about this, if we really need to define the context, it should always be consensual, age-appropriate, context-appropriate, species-appropriate, but now we are talking about interactions between humans and robots.

So I guess... No, I was actually trying to make a different point, which is that I think robots will have rights in the future. And I think that in order for us to have deep and meaningful relationships with robots, we would have to consider them as entities in their own right that deserve respect. And that's a really interesting concept that I think people are starting to talk about a little bit more, but it's very difficult for us to understand how non-human entities work. That is, as with dogs and other animals, we can have rights at the same level as humans. - Okay, yes.

We cannot and should not do whatever we want with animals, we have a USDA, we have a Department of Agriculture that takes care of animals and uses committees for research, agriculture and livestock and all that. So when you first said it, I thought, wait, why would I? There will be a robotic bill of rights, but it absolutely makes sense in the context of everything we've been talking about so far. If you're willing, I'd love to talk about dogs, because you mentioned dogs a couple of times, a robot dog, you had a biological dog. Yes.-Yes. I had a Newfoundland named Homer for many years while he was growing up. - In Russia or in the United States? - In the U.S.

And he weighed over 200 pounds, he's a big dog. - That's a big dog. - People know Newfoundland. So, he's this black dog that really has long hair and a kind soul. I think maybe that's true for a lot of big dogs, but he thought he was a small dog. So, he moved like that and...-Was that your dog? - Yes Yes. -So, you had him since he was quite young? - From the beginning to the end. And one of the things, I mean, he had this kind of, that we mentioned as the Roombas, he had a good-natured silliness that was just overwhelming, that's part of the reason I named him Homer, because it's in honor of Homer Simpson. , in case people are wondering which Homer I'm referring to.

And then, there is a. Yes, exactly. There is an awkwardness that was something that immediately led to a deep love for each other. And one of the, I mean, he was always there, is the shared moments, he was always there for so many nights together, that's a powerful thing in a dog, that he was there through all the loneliness, through all the difficult times. . through successes and all that kind of stuff. And I remember, I mean, it was a really touching moment for me, I still miss it to this day. -How long ago did he die? - Maybe 15 years ago.

So, it's been a while. But it was the first time I really experienced the feeling of death. So what happened is that he had cancer. And so, he slowly died. And then there comes a point where he can't get up anymore. Now, there are many things that could be said here that I struggle with. Maybe he suffered a lot more than necessary, that's something I really think about a lot. But I remember when I had to take him to the hospital and the nurses couldn't carry him, right? So, you're talking about a 200-pound dog and at the time I was really into weightlifting.

And I remember they tried to figure out all these kinds of ways to do it. So, in order to put them to sleep, they had to take them to a room. And then I had to carry it everywhere. And here's this dying friend of mine that I had to, first of all, who was very difficult to carry, someone so heavy when he's not helping you. And if. So, I remember it was the first time I saw a friend lying there and watching the life drain out of his body. And that realization that we're here for a short time became so real, that here is a friend who was there to help me the week before, the day before, and now he's gone.

And that was, I don't know, him talking about the fact that he could be deeply connected to the dog. He also talked about the fact that the moments shared together that led to that deep friendship will make life so amazing, but he also talked about the fact that death sucks. So, I know you lost Costello recently. - Yes. - And you can... - And as you say this, I'm definitely fighting back tears. I thank you for sharing that. I guess we're both about to cry, I don't want to say dogs it had to happen given the timing of this.

Yeah, it's... - How long did you know Costello wasn't doing well? -Let's see, a year ago, at the beginning, about six months into the pandemic, he started to have abscesses and no, his behavior changed and something really changed. And then I put him on testosterone, which helped a lot of things. It certainly didn't cure everything, but it helped a lot of things, I was dealing with joint pain, sleeping problems, and then it became a very slow decline to the point where two or three weeks ago, I had a closet full. of medication. I mean, this dog was like a pharmacy.

It was surprising to me when I looked at him the other day, I still haven't cleaned or removed all that stuff, because I don't dare to do it, but... Do you think he was suffering? - Well, then, what happened was about a week ago, actually it was about a week ago, it's incredible. He was going up the stairs, I saw him slip. And he was a big guy, he wasn't 200 pounds, but he was about 90 pounds, he's a bulldog, he was pretty big and fit. And then I realized thatIt's still not very good. To be a business, to be an entrepreneur you have to dream big and you have to have other people around you, like friends and support groups that make you dream big.

But if you don't give in to cynicism and you appreciate the beauty of the injustice of life, the absurd injustice of life, then I think that makes you appreciate everything. It's like a prerequisite for gratitude. And yes, I think that instilled in me the ability to appreciate everything, like everything. Everything is amazing. And there is also a culture that romanticizes everything. It's almost as if romantic relationships are very soap opera-like, very dramatic. And I think that was instilled in me too: I not only appreciate everything about life, but I also get excited. In a sense, I have a visceral feeling of joy about everything.

And the same with friends or people of the opposite sex. There is a deep emotional connection that is also overly dramatic. I guess in relation to the actual moment. But I get such deep, dramatic joy from so many things in life. And I think I would attribute it to education in Russia. But what remains most is friendship. And since then I have had another friend like that in the United States, he lives in Chicago. His name is Matt. And little by little here and there I'm accumulating really fascinating people, but I'm very selective about that. Funnily enough, the few times, it's not a few, it's a lot of times now interacting with Joe Rogan sounds surreal, but there was a kindred spirit there.

Like I connected with him. And there have been people like that also in fighting sports that they are really related to. In fact, I have struggled, and this is why I am so glad to be your friend, I have struggled to connect with scientists. - Sometimes they can be a little wooden. - Yes. - Even the biologist. I mean, one thing that, well, I'm really struck by the fact that you work with robots, you're an engineer, artificial intelligence, science and technology, and all of that sounds like hardware, right? But what you're describing and I know to be true about you is this deep emotional life and this resonance, and it's really wonderful.

In fact, I think it's one of the reasons why so many people, scientists and others, have approached you and your podcast, because you have both elements. In Hermann Hesse's book, I don't know if you, "Narcissus and Goldmund", right? It's about these elements of the logical rational mind and the emotional mind and how they intertwine. And if people haven't read it, they should, and you embody the full picture. And I think that's a big part of what draws people to you. - By the way, I have read all of Hermann Hesse's books. - As usual, I have done about 9% of what life is.

Is not true. You mentioned Joe, that he is a phenomenal human being, not only because of his amazing accomplishments, but because of how he presents himself to the world one on one. I think I heard him say the other day in an interview, he said, there's no public or private version of him. And he says, this is me. He said he was beautiful. He said: I am like the fish that passed through the net. And there is no on-stage off-stage version. And you're absolutely right. Me too. - Fish. - So, but, well, guys, I have a question about... - But that's a very good point about public and private life.

It was huge, if I could comment very quickly. He was a long-time fan of Joe, but he has been an inspiration to have no difference between public and private life. In fact, I had a conversation with Naval about this and he told me that you can't have a rich life, an exciting life, if you are the same person in public and in private. And I think I understand that idea, but I don't agree with it. I think it's really rewarding and exciting to be the same person in private and in public with very few exceptions. That being said, I don't have any really weird sexual problems.

So I feel like you can be open to basically everything. I have nothing to be ashamed of. There are some things that can be misperceived, like Roombas that scream, but I'm not ashamed of them. I just have to present them in the right context. But there is the freedom to be the same person in private and in public. And that Joe made me realize that you can be that. And also be kind to others. It sounds a little absurd, but I really always enjoyed being good to others. Like just being kind to others. But I always felt like the world didn't want me to be.

There was so much negativity when he was a kid, like when he was around people. If you really notice how people talk, they complain about the weather, this could be like the big cities I've visited. But there is a general negativity and the positivity is somewhat suppressed. You're not, one, they don't see you as very smart, and two, there's a kind of, they see you as a little weird. And that's why I always felt like I had to hide that. And what Joe made me realize, one, that I can be exactly the same person in private and in public, and two, that I can accept being kind, the way I like, the way I know how.

And for me on likes, on Twitter or likes publicly, whenever I say things, that means simply saying things almost to the point of being cliché. And now I have the strength to say it, even if they make fun of me. You know what I mean? It's just fine. If everything is going to be fine. Okay, some people will think you're stupid. They're probably right. The point is to simply enjoy being yourself. And that Joe, more than almost anyone, because he's so successful at it, inspired me to do that. Be kind and be the same person, in private and in public. - I love it.

And I love the idea that authenticity doesn't have to be overshared, right? That doesn't mean you reveal every detail of your life. It is a way of being faithful to an essence of oneself. - Good. When you think deeply and introspect, you never have the feeling that you are hiding something from the world or that you are being dishonest in some fundamental way. So yeah, that's really liberating. It allows you to think, it allows you to think freely, to speak freely, to simply be freely. That said, it's not like there's still a responsibility to be the best version of yourself.

So, I'm very careful about how I say something. So the point is that it is not so simple to express the spirit within you with words. I mean, some people are much better than others. I fight. Many times when I say something and I hear myself say it, it sounds very silly and is not at all what I meant. So that's the responsibility you have. It's not just that being the same person in public and in private means you can say anything. It means that you still have the responsibility to try to be, to express who you really are.

And that's difficult. - It's hard. And I think we have this pressure, all people, when I say we, I mean all humans, and maybe robots too, feel this pressure to be able to express ourselves in that moment in that way. And it's beautiful when someone, for example, can capture some essence of love or sadness or anger or something in a song or in a poem or in a short quote, but maybe it's also possible to do it together, all the things as you appear . For example, one of the things that initially drew me to want to get to know you as a human being and a scientist, and we eventually became friends, was the level of respect you gave your podcast listeners by wearing a suit. - Yes. - I'm being serious.

I was raised thinking that if you're overdressed, by American standards, certainly by American standards, you're overdressed for a podcast, but it's genuine. You're not doing it for any reason, except that I have to assume, and I assumed at the time, that it was because you respect your audience, you respect them enough to appear a certain way to them. It's for you too, but it's for them. - Yes. - And I think between that and your commitment to your friendship, the way you talk about friendship and love and the way you uphold these higher ideals, I think at least as a consumer of your content and as your Friend, what I find is that overall, you are communicating who you are, it doesn't have to be a quote or something like that.

And I think we're a little obsessed with that Einstein quote, or that line of poetry or something, but I think you embody the way that, and Joe too, is about how you live your life and how you show up as a collection. of things said and done. - Yes, that's it, so the whole thing is the goal, the complicated thing, and Jordan Peterson talks about this because he's under attack much more than you and I will ever be, but... - Right now? - Not for now? This is very true for now. That people who attack on the Internet, that is one of the problems with Twitter, is that they do not consider the whole, they take single statements.

So one of the coping mechanisms, like why Joe has been an inspiration, is that when you're an overall good person, a lot of people will know that. And that makes you much more immune to attacks from people who make an individual statement that could be a misstatement of some kind or that doesn't express who you are. And so, I like that the idea is the aggregate. And the power of the podcast is that you have hundreds of hours out there, and being yourself and people get to know who you are. And once they do, and you post pictures of Roombas screaming while you kick them, they'll understand that you're up to no good.

By the way, as a side note, I don't know if I want to post this because it's not just about the Roombas... You have a whole dungeon of robots. - Well. So this is a problem, Boston Dynamics ran into this problem. But let me work this out as a workshop with you, and maybe because we're posting this, people will let me know. So there are robots with legs, they look like a dog, I'm trying to create a very real connection between humans and robots, but they're also awesome because you can throw them like off a building and they'll land fine.

And it's beautiful. - That is incredible. I've seen Instagram videos of cats jumping off fifth-floor buildings and then walking away. But no one should throw their cat out of a building window. - Well, this is the problem I'm experiencing, everyone certainly kicks robots, it's really fascinating how they recover from those kicks. But just watching myself do it and watching others do it, it just doesn't look good and I don't know what to do with it. Because it's such... I'll do it. - See. But you don't, because you're... A robot, no, I'm kidding. What is interesting? - Yes. - Before today's conversation, I probably could, and I'm thinking about robots, bills of rights and so on.

Not to satisfy you or to satisfy anything, except that if they have some sensitive aspect to their being, then I would be willing to kick it. - I don't think we can kick it, you may be able to kick the first time, but not the second, this is an experience problem. One of the interesting things is one of the robots I'm working with, you can pick it up by one leg and it's hanging, and you can throw it any way and it will land correctly. So, it's really... I had a friend who had a cat like that. - Oh man, we're looking forward to the cat lyrics- - Oh no, I'm not suggesting that anyone did that, but he had this cat, and the cat, he would just throw it on the bed from across the room. and then he would come running back for more, or somehow that was the nature of the relationship.

I don't think anyone should do that to an animal, but this cat seemed to come back for some reason. - But a robot is a robot, and I find it fascinating how difficult it is for me to do it. So, it's unfortunate, but I don't think you can do that to a robot. Like he struggled with that. So to be able to do that with a robot, I have to get almost to the state that I imagine doctors go into when they do surgery, like I have to do what my colleagues do in robotics, which is like starting to see it as an object. - Dissociate. - How to dissociate.

So it was fascinating to have to do that with a robot. I just want to take that little tangent. - No, I think it's something important. I mean, I'm not ashamed of the fact that I've worked with experimental animals for many years, and that's been a very challenging aspect of being a biologist. Mostly mice, but in the past not anymore, thank goodness because I just don't like doing it, and bigger animals too. And now I work with humans, to whom I can give consent, verbal consent. So I think it's extremely important to understand what the guidelines are and where your own limits are on that.

It is not only an important issue, but it could be the most important before any work can move forward. - So you asked me about friendship. I know you have a lot of thoughts about friendship, what do you think is the value of friendship in life? - Well, for me personally, just because of the trajectory of my life and my friendship arc, and I have to say, I have some friends who are just friends, they are completely platonic relationships, but for me it has been mainly male friendship. It has been... - Actually, for me they have all been male friendships, yes. - Interesting. - Yes. - It has been a real lifesaver.

They are my family, I have a biological family and I have great respect, love and appreciation for them, but it has given me, I won't even say confidence, because there is always anxiety in taking any good risk, or any risk worth taking. It has given me the feeling that I should choose certain things and try certain things to take risks and overcome that anxiety. And I don't consider myself a particularly competitive person, butI would rather die than disappoint or disappoint one of my friends. Actually, I can't think of anything worse than disappointing one of my friends, everything else is secondary to me. - What a disappointment? - Disappoint, I mean, no, I mean, I certainly always strive to show up to the best of my ability for friendship, and that can be in small ways.

That may mean making sure the phone is away, sometimes it's that I'm terrible at punctuality because I'm an academic. And so, I just get lost in time and I don't want to say anything, but rather try hard, listen, enjoy the good times and make time. It kind of goes back to this first variable that we talked about, making sure that I'm spending time and having in-person time and communicating. And I think there are many ways in which friendship is vital to me, it really is to me. , which makes life worth living. - Yeah. Well, there's a, it surprises me like with high school friends how we don't really talk that often these days in terms of time, but every time we see each other, we immediately go back to square one.

So, I had a hard time knowing how much time you really put into making the friendship deeply meaningful because they are always there with me even if we don't talk often. So there is a kind of loyalty. I think maybe it's a different style, but I think, for me, friendship is being there during difficult times. Like I'm a lot more trustworthy when you're going through shit than when... - You're pretty trustworthy anyway. - No, but if you're like a wedding or something, or like, I don't know, like you want an award of some kind, yeah, I'll congratulate you a lot, but that's not like that. , and I'll be there, but that's not as important to me as being there when, like no one else, it's like just being there when shit gets bad, or when something's hard and the world turns its back on you, all that kind of stuff. of things. , to me, that's where friendship makes sense. - Well, I know that's true about you and that's something that feels and is something real for you.

Let me ask you one more thing about that, actually, because I'm not a Jujitsu practitioner, I know you are, Joe is, but years ago I read a book that I really enjoyed, which is Sam Sheridan's book, "A Fighter's Heart." . "He talks about all these different forms of martial arts. And maybe it was in the book, maybe it was in an interview, but he said fighting, or being in a physical battle with someone, Jujitsu boxing or some other form of contact Direct physicality between two individuals creates this bond like no other. Because he said it's like a one night stand, you share bodily fluids with someone you barely know. - And I chuckled because it's kind of funny and kind of ironic.

At the same time, I think this is a fundamental way in which members of a species bond through physical contact. And certainly, there are other ways, there are hugs, holding hands, and there are sexual relations, and there are all kinds. of things. - What are hugs? I haven't heard of that. - I heard this recently, I didn't know this term, but there is a term, they have turned the noun cupcake into a verb, it turns out cupcakeking, I just found out about this. Making cupcakes is when you spend time just cuddling. I did not know anything about this.

You heard it here first, although I heard it for the first time the other day. Making cupcakes is actually... - Snuggling is everything, it's not just, is it in bed or on the bus? What are hugs? I'm looking for what it is to hug... We need to analyze this and define the variables. I think it definitely has to do with physical contact, I'm told, but in terms of battle, competition, and Sheridan's date, I'm just curious. So do you get close or feel a bond with people who, for example, you played Jujitsu with, or even though you don't know anything else about them, were you right about this? - Yes, I mean on many levels.

He also has the book, what? "The mind of a fighter." - Yes, that was the third. He actually he is an excellent writer. The interesting thing about him, briefly about Sheridan, I don't know him, but I did some research, he went to Harvard, he majored in art at Harvard, he claims all he did was smoke cigarettes and make art. . I don't know if his art was any good. And I think his father was in the SEAL teams. And then when he got out of Harvard, he graduated, he went all over the world learning all forms of martial arts, and he got early to the ultimate kind of fighting, mixed martial arts and stuff.

Great book. Yes Yes. - It's amazing. I don't actually remember it, but I read it and I remember thinking it was an amazing summary of what makes wrestling an art, what makes it compelling. I would say there are a lot of ways that Jujitsu, wrestling, and combat sports in general, are like one of the most intimate things you can do. I don't know if I would describe it in terms of body fluids and all that kind of stuff. -I think he was more or less joking. - I think there are a few ways to do it.

So, one, because you are very vulnerable. So the honesty of stepping on the mat and often we all have the ego of thinking we are better than we are at this particular art. And then the honesty of being subdued, or being worse than you thought you were and just sitting with that knowledge, that kind of honesty, we don't get to experience that in most of daily life. We can continue to live in a kind of illusion of our conceptions of ourselves because people are not going to hit us with reality, the mat just tells the truth, reality just hits you.

And that vulnerability is the same as the loss of a loved one, although it is the loss of a reality that you knew before, now you have to deal with this new reality. And when you're sitting there in that vulnerability, and there are other people who are sitting in that vulnerability too, you can really connect like, fuck, like I'm not as special as I thought, and life is like No, life is harder than that. what I thought, and we're just sitting with that reality, some of us can put it into words, some of us can't. So, I think it's definitely something that at least intimacy.

The other thing is human contact. There's something like, I mean, like a big hug, like during COVID, very few people hugged me and I hugged them, and it always felt good when they did. Like we're all getting tested, and especially now that we're vaccinated, but there are still people, this is true in San Francisco, this is true in Boston, they want to keep, not just six feet apart, but stay home and never touch you. . That loss of basic humanity is the opposite of what I feel in Jujitsu, where it was like that contact where you say: I don't give a shit about the rules that we're supposed to have in society and that you have to abide by. a distance and all that kind of stuff.

Just the hug, like the intimacy of a hug, is like a good bear hug, and you're like controlling another person, and there's also some kind of love communicated just by trying to break each other's arms. I don't understand exactly why violence is such a close neighbor to love, but it is. - Well, in the hypothalamus, the neurons that control sexual behavior, but also non-sexual contact, are not only close to the neurons that control aggression and fighting, but they are salt and pepper with those neurons. It's very interesting, and it almost sounds a little risky and controversial and stuff, I'm not anthropomorphizing about what this means, but in the brain, those structures are interdigitated, they can't be separated except at a very fine level.

And here, the way he describes it, it's just like something real. - I want to make an interesting comment. Again, these are things that could be taken out of context, but one of the amazing things about Jujitsu is that both boys and girls train it. And I was surprised. So, I'm a big fan of yoga pants at the gym. Reveals the beauty of the female form. But the thing is that girls often dress in tight clothing in Jujitsu. And I found that I didn't think like that at all when I was training with girls. - Well, the context is very non-sexual. - But I was surprised to know that.

When I started practicing Jujitsu, I thought, wouldn't it be a little strange to train with opposites in something so intimate? - So, boys and girls, men and women, do you do Jujitsu together? - Completely. - Interesting. - And the only times girls try to stay away from boys, I mean, there are two contexts, of course, there will always be creeps in this world. So, everyone knows who they're staying away from, and the other thing is that there's a size disparity. So girls often try to get a little closer to people weight-wise, but no, that's one of the things that makes women stronger, that's what they love when they started doing Jujitsu, it's first of all. , they gain awareness and pride in their body, which is fantastic.

And then secondly, especially later on, they start subduing big guys like these brothers coming in, who are all ripped and muscled, and learning the technique to exert dominance over them, and that's a powerful feeling. -You have seen women force a larger man to touch them, or even strangle them. - Well, I was deadlifting like a four, oh boy, I think it's 495. So, I was really into powerlifting when I started in Jujitsu, and I remember getting submitted, I thought I came in feeling like I'll be, if not the best fighter of all time, at least in the top three.

And so, as a white belt, you arrive happy. And then you realize that as long as you don't apply too much force, I remember being submitted many times by 130 and 120 pound girls at our Balance Studios in Philadelphia, which many amazing female jiu-jitsu players. And that is also very humiliating: technique can dominate pure force in combat. And that's the other thing, there's something about combat that is paramount. There, it just feels, it feels like we were born to do this. Like this... We have circuits in our brain dedicated to this type of interaction. No doubt. - And that's what I felt, it's not like I'm learning a new skill.

It was as if I was somehow remembering the echoes of something I learned in the past. - Well, it's like reaching puberty. A child before puberty has no idea that boys and girls have this attraction, whether or not they are attracted to boys or girls, it doesn't matter. At some point, most people, not all, but certainly, but most people, when they hit puberty, suddenly appear differently and certain people acquire a romantic or sexual interest for the first time. - Yes. - And it's like it's revealing a circuit in the brain. It's not that they learn that it is innate.

And I think about it, when I hear the way you describe Jujitsu and the Jujitsu inscription, it reminds me a little bit, Joe was telling me recently about the first time he went hunting and he felt like it revealed a circuit that was in him the whole time. . , but he had not experienced it before. - Yes. That's definitely there. And of course, there is physical activity. One of the interesting things about Jujitsu is that it's one of the really strenuous exercises that you can do late in your adult life, like when you're 50, 60, 70, or 80 years old. When I arrived, there were some people in their 80s who were training.

And as long as you're smart, as long as you practice techniques and choose your partners correctly, you can make that kind of art. That's late in life. And then you're exercising. I don't find many activities that are susceptible to that. So, because it is a thinking game, Jujitsu in particular is a very worthwhile art or technique. So you can still maintain, first of all, stay injury free if you use good technique, and also through good technique be able to be active with people who are much, much younger. And that was, for me, that and running are the two activities that you can do late in life.

Because for me a healthy life has exercise as a piece of the puzzle. - No, absolutely. And I'm glad that we're on the physical component, because I know that for you, you've talked before about the crossover between the physical, the intellectual, and the mental. Do you keep running at ridiculous hours of the night for a ridiculously long time? - Yes, definitely. I've been running late into the night here in Austin. People say that the area we are in now is a dangerous area, which I find ridiculous coming from bigger cities. No, I run late at night.

There is something. -If you see a guy running through Austin at 2:00 a.m. in a suit and tie, probably. - Okay, yes. I mean, I think about that because I'm getting more and more recognized in Austin. I worry, but not really, about being recognized late at night. But there's something about the night that brings out those deep philosophical thoughts and self-reflection that you really enjoy. But recently I started getting back into the routine. So I'm going to compete or hope to compete in September and October. - In Jujitsu? - In Jujitsu, yes. To return to the competition.

And that requires getting back into excellent cardiovascular shape. I have been running as part of my daily routine. - I understand. - Yes. - Well, I always know that I can contact you regardless of the time zone in the middle of the night, wherever that happens. - Well, part of that has to be just being single and a programmer. Those two thingsThey just don't work well in terms of a stable sleep schedule. - It's not a banker's type of job. Nine to five. I want, you mentioned single. I want to ask you a little about the other form of relationship, which is romantic love.

So are your parents still married? - Still married, still happily married. - That is amazing. - Yes. - Something rare nowadays. - Yes. - So you grew up with that example? - Yeah. I guess that's a powerful thing, right? If there is an example that I think can work. - Yes. I didn't have that in my own family, but when I see it, it's inspiring and beautiful. The fact that they have that, and that that was the norm for you, I think is really wonderful. - Well, in the case of my parents it was interesting to see because obviously there is tension.

There will be times where they will fight and all that kind of stuff. Obviously they get frustrated with each other and they like it, but they find mechanisms to communicate it, like making fun of each other a little bit, joking around, getting some of that frustration out and eventually getting together and finding their happy moments and that being the energy. I think it's clear because I met there, I think I was in my twenties, I was very, very young. I think we grow together as people. - Yes. You are still in the critical period of brain plasticity. - And also, I mean, it's like divorce is so frowned upon that you put up with it.

And I think for a lot of couples, especially from that era, the Soviet Union, that probably applies to a lot of cultures. You work hard and you work hard, you learn to work hard. And once you do, you start to get to some of those rewarding aspects of being, like over time by sharing so many moments together. That's definitely something that was an inspiration to me, but maybe that's where I've done it. So I have a similar kind of longing to have a lifelong partner, I like to have that kind of vision, where the same goes for friendship, lifelong friendship is the most meaningful kind.

That there is something about that time of sharing all that time together. Until death do us part is a powerful thing. Not by force, not because religion says so or the government says so or your culture says so, but because you want to. - You want to have children? - Definitely yes. I definitely want to have children. It's... - How many Roombas do you have? - Oh, I thought... - You should, no, no... - Human children? - No, from human to human. - Because I already have the children. - Exactly. What I was saying is that you probably need at least as many human children as Roombas.

Big family, small family. - So. - In your mind, they're a bunch of Fridmans running around. - Then I will tell you that, realistically, I can explain exactly my thinking, and this is similar to robotics work. If I'm purely logical right now, my answer would be: I don't want kids. Because I just don't have enough time. I have so many things to do. But when I use the same type of vision that I use for robots, I know that my life will be transformed with the first one. How I know I would love to be a father.

And then the question of how many are on the other side of that hill. It could be a ridiculous number. So I just know that... I have a feeling and it could be, I don't have a crystal ball, but I don't know. I see more than three or more coming to mind. - A lot of that also has to do with the partner you're with. So that's an open question, especially in this society, about what the right partnership is. Because I am deeply empathetic. I want to see, like me, what I'm looking for in your relationship is that you're really excited about someone else's passions, like whatever you're into doesn't have to be a professional success.

Any kind of success, just to be excited for them and for them to be excited for me. And share that emotion and build, build, build. But there were also practicalities like: what kind of things do you like to do together? And I think family is a really serious task. - It certainly is. I mean, I think I have a friend who said it, I think the best thing is that first you have him, he's in a very successful relationship and he has a family. And he said, first you have to define the role and then you have to choose the right person for the role. - Well, yes, there are some deep aspects of that, but there is also an aspect where you are not smart enough from this side to define the role.

I think that in part you have to take a leap. And I see having children that way. You just have to accept it and figure it out too. As long as there's love there, what the hell is life for? So, there are so many incredibly successful people that I know that I have come to know that all have children. And the presence of children for the most part has just been something that energizes them, something that gave them meaning, something that made them the best version of themselves, that is, made them more productive, not less, which It's fascinating to me. - It is fascinating.

I mean, can you imagine if what you felt about Homer, what I feel and feel about Costello is at all an idea of ​​what that must be like then. - Exactly. The disadvantage, what worries me most is the partner side. I have seen that children are almost universally a source of increased productivity, joy, and happiness. Yes, they are a pain in the ass. Yes, it's complicated. Yes, so and so, people like to complain about children. But when you look past that small surface layer of complaint, kids are great. The source of pain for many people is when the relationship is not working.

That's why I'm very worried because dating is very difficult and I'm a complicated person. That's why it's been very difficult to find the right kind of person. But that statement doesn't even make sense because I'm not on dating apps, I don't watch people. You're like the first person I've seen in a long time. It's like you, Michael Malice and Joe. So I don't think I saw myself as a woman. What is it? An element of the female species for quite some time. So, I think you have to work hard. What is it? Daniel Johnston says: True love will find you, but only if you look.

So there's some element of really taking the leap and exposing yourself to different situations. And I don't know how to do that when you're behind a computer all the time. - Well, you are a builder and a problem solver, and you find solutions, and I am sure that the solution is available, and. - I think you're implying that I'm going to build the girlfriend, which I believe. - Well, and maybe we shouldn't separate this friendship, the notion of friendship and community, and if we go back to this whole concept, maybe you know this woman through a friend, or maybe you do or something. of that type. - So one of the things, I don't know if you feel the same way, I'm definitely one of those people who just falls in love and that's it. - Yes, I can't say that's the case.

With Costello it was instantaneous. - Yes. - It really was. I mean, I know it's not romantic love, but it was instant. No, me, but that's me. And I think if you know it, you know it, because it's good to have it there. - I'm very careful with that, because you don't want to fall in love with the wrong person. So I try to be very careful, I've noticed because I fall in love with everything, like this cup, everything. I fall in love with the things of this world. So, you have to be very careful because a girl comes up to you and tells you that she loves DUSTY HUSKY, that doesn't necessarily mean that you will marry her tonight. - Yes.

And I like the way you said it out loud so you could hear it, it doesn't mean you have to marry her tonight, right? - Exactly. - Good. - But I mean, people are amazing, and people are beautiful and that's what I fully accept, but you also have to be careful with relationships. And at the same time, as I mentioned to you offline, I don't, there's something in me that appreciates jumping over the fences and not dating, like serial dating or dating. - As if you were one boy and one girl. - Yes. - And you said that. - And it's complicated because you have to be careful with that kind of thing.

Especially now that there is a growing platform that has a ridiculous amount of female interests of a certain type. But I'm looking for a deep connection, and I look for it by sending myself home and occasionally talking to Stanford professors. - Perfect solution. - In a podcast. - Perfect solution. - You are going to exercise very well. - It's good, it's part of what constitutes a kind of machine learning. - Yes, in a way. - Yes, you mentioned what has now become a pretty extensive and expansive public platform, which is incredible. I mean, the amount of people that came out, the first time I saw his podcast, I noticed the suit, I was like, he respects his audience, which was cool, but I also thought this is amazing.

People show up to receive information about science, engineering and technology and those discussions and other types of discussions. Now I want to talk for a moment about the podcast. So my two questions about the podcast are: When you started it, did you have a plan? And regardless of what that answer is, do you know where you're taking it or would you like to leave us? I think an element of surprise is always fun. But what about the podcast? Do you like the podcast? I mean, your audience certainly includes me, they really enjoy the podcast. Is incredible. - I love talking to people, and there is something about microphones that really brings out the best in people.

As if you don't have the chance to talk like that. If you and I were just hanging out, we would have a very different conversation regarding the amount of attention we allocate to each other. We would have fun talking about other things and doing other things. There will be a lot of distraction. There would be some phone use and all that kind of stuff. But here we are 100% focused on each other and focused on the idea. And sometimes we play with ideas that neither of us know the answer to, like a question we don't know the answer to.

We're both kind of groping for something, trying to figure it out, trying to get some ideas about something that we hadn't really figured out before and together we come to it. I think that's magical. I don't know why we need microphones for that, but somehow we do. - It's like doing science. - Definitely, for me it is like doing science. That's exactly it. And I'm so glad you said that because I don't really say it often, but that's exactly how I felt. I wanted to talk to friends and colleagues at MIT about doing real science together.

That's how I felt about it. I like to really talk about problems that are really interesting, rather than the incremental work we're currently working on for a particular conference. So we really ask ourselves questions like: what are we doing? Where is this going? What are the big ones? Will this really help us solve, in the case of AI, solve intelligence? Does this even work in intelligence? It makes some sense, which is why I initially called it artificial intelligence. It's like most of us aren't working on artificial intelligence. You are working on a very specific problem and set of techniques; at that point, it is machine learning that solves this particular problem.

This will not lead us to a system that comes close to generalizing the human mind. Like the kinds of things that the human mind can do in terms of memory, in terms of cognition, in terms of reasoning, common sense reasoning. This doesn't seem to get us there. So the initial impulse was: Can I talk to these people who do science together through conversation? And I also thought there weren't enough, I didn't think there were enough good conversations with world-class minds that I got to meet. And not those who had the book, or that this was the ideal.

You often take this tour when you have a book, but there are many minds that don't write books. - And books also limit the conversation, when talking about this, this book. - But I've noticed that, with brilliant people who haven't written a book, we can talk about ideas in a new way. In reality, neither of us, when we ask a question, do not know the answer once the question is asked. And we try to get there. I don't know, I remember asking world-class deep learning researchers questions about, why do neural networks work as well as they do?

That question is often asked vaguely, but like when you have microphones and you have to think about it, and you have 30 minutes to an hour to think about it together, I think that's science. I think that's really powerful. So that was the only goal. The other is that, again, I don't usually talk about this, but in a sense I wanted to have dangerous conversations. Part of the reason I wanted to wear a suit is because I want it to be brave. The reason I don't usually talk about it is because I feel like I'm not good at conversation.

So it seems like it doesn't match the current skill level. But I wanted to have really dangerous conversations that I was the only one capable of having. Not entirely exclusive, but I'm a huge Joe Rogan fan and I had to ask myself: What conversations can I have that Joe Rogan can't? For me, I know I mention this, but for me that person I thought of at that moment was Putin. That's why I mention it. He's like Costello, he's not just one person. It's also an idea for me what I strive for. Just to have those dangerous conversations.

And the reason why I am especially qualified as a Russian, but there is also judo and artsmartial arts, there are many elements that make me have a conversation that he hasn't had before. And there are a few other people that I had in mind, like Don Knuth, he's a Stanford computer scientist who I thought is one of the most beautiful minds that ever lived. And no one really talked to him, didn't really talk to him. He gave some lectures that people love, but I really only have one conversation with him. There are some people like that.

One of them passed away, John Conway, who I never got, we agreed to talk, but he died before us. There are some people like that, who I thought it was a crime not to listen to. And I have the unique ability of knowing how to buy a microphone on Amazon and connect it to a device that records audio and then post it, which seems relatively unique. As if that wasn't easy in the scientific community. People who know how to connect a microphone. - No. They can build Faraday cages, two-photon microscopes and bioengineering, all kinds of things, but the idea that you can take ideas and export them into a structure or pseudostructure that people would benefit from seems like a cosmic achievement. them. - I don't know if it's fear or they just haven't tried it, so they haven't learned the skill level. - But I think they are not trained.

I mean, we could talk about this for a while, but I think it's important. And maybe we should do it, because they are not trained to do it. They're trained to think about specific games and specific hypotheses, and a lot of them don't care, right? They became scientists because that's where they felt safe, and so why would they leave that haven of safety? - Well, they don't necessarily always see the value in it either. We are all learning together, you and I are learning the value of this. I think you're probably having an amazing and exceptionally successful podcast that you recently started. - Thanks to your encouragement. - Well, but there is a raw skill there and that is that you are definitely an inspiration to me in how you made the podcast at the level of excellence that you achieve.

But I think you've discovered that that's also an impactful way to do science. That podcast. And I think a lot of scientists haven't figured it out yet. Which is, if you apply the same kind of rigor that you apply to academic publications or even conference presentations, and you put that rigor and effort into the podcast, whatever it is, it could be a five-minute podcast, a two-minute podcast. minutes. hour-long podcasts, it could be conversational or it could be more like a lecture, if they apply that effort, you have the potential to eventually reach tens of thousands, hundreds of thousands, millions of people.

And that's really, really powerful. But yeah, for me I give a platform to some of those people, especially for me personally, so maybe you can talk about the fields you're drawn to, but I thought computer scientists were especially bad at this. There are brilliant computer scientists that I thought would be amazing to explore their mind, explore their thinking. And I took it almost as an effort. And at the same time I had in mind other guests or people who connected with my own interests. Then wrestling. Wrestling, music, soccer, both American football and soccer. I have a few people in particular that I'm very interested in.

Buvaisar Saitiev. The Saitiev brothers, even Khabib for wrestling, just to talk to them, because. - Oh, because you can, you can communicate- - In Russian and wrestling, right? As fighters and as Russians. And just like that, it's like an opportunity to explore a mind that I'm capable of bringing into the world. And I also feel like it makes me a better person, just by being so vulnerable and exploring ideas together. I don't know, like a good conversation. I don't know how often you have great conversations with friends, but podcasts are like that. And it is deeply moving. - Is the best.

And what you brought. I mean, when I saw you sit with Penrose, the Nobel Prize-winning physicist, and those other people, it's not just because he has a Nobel, but what comes out of his mouth is incredible. And what you were able to have in that conversation was so much better, light years away from what any other interviewer had, I don't even want to call you an interviewer because it's really about conversation. Light years beyond what anyone else had been able to interact with, he was a beacon of what is possible. And I know, I think that's what attracts people.

And there is a certain intimacy, that certainly for people, our friends, as we are, and they know each other, there is more to that, but there is intimacy in those types of private conversations that become public. Y. - Well, that's it, with you, you're probably starting to realize, and Costello, it's part of that, because you're authentic and you're exposing yourself completely, people almost don't just consume the words. You're saying, they also enjoy watching you, Andrew, wrestle with these ideas or try to communicate them. They like defects, they like human beings. - Oh, well, that fails. - Well, that's good because I have a lot of those. - But they like the self-critical aspects, like when you are very careful, where you are very self-critical of your flaws.

I mean, in the same way, I think it's interesting that people see me talking to Penrose, not just because Penrose is communicating ideas, but because here's this silly kid trying to explore ideas. Like they know this kid. There is a human connection that is really powerful. I think the same as with Putin, right? It's not just a good interview with Putin, but it's also a kid struggling to talk to one of the most powerful and, some say, dangerous people in the world. They love that. The authenticity that led to that. And in return, I can connect, with everyone I run with on the street and all that kind of stuff, there's a deep connection there, almost within a minute or two, that's like no other. - Yes, there is an intimacy that you have formed with them. - Yes, we have been on this journey together.

And yes, I have the same thing with Joe Rogan before I met him, right? Like I was, because I was a Joe fan for so many years, there's something, there's a kind of friendship that's as absurd as you could say in podcasting and listening to podcasts. - Yes. Maybe it will complete that a little bit or resolve a little bit of that loneliness you've been talking about before. - Until the robots arrive. - I have just a couple more questions, but one of them is from your audience, that is, I am not going to ask you the meaning of the hedgehog, I just want to know, does it have a name?

And you don't need to tell us the name, but just, does it have a name? Yes or no? - Well, there's a name you like to be called, and then there's a private name in the privacy of your own company that we call each other. No, I'm not that crazy. No, his name is Hedgy. He is a hedgehog. I don't like stuffed animals. But his story is one of minimalism. So I gave away everything I have, three times in my life now. By everything I mean almost everything, I was left with jeans, a shirt and a laptop.

And lately it's also been the guitar and things like that. But he survived because it was always in it, at least the first two times it was in the laptop bag, and he was just lucky. And so I liked the perseverance of that. And I saw it for the first time in, the reason I bought a stuffed animal, I don't have any other stuffed animals, was in a thrift store, in this like a giant pile of stuffed animals and it jumped out at me, because Unlike Everyone Else, he has that intense, mean look. That he is simply annoyed with life, with the cruelty of life.

And the thing is, especially in contrast to the other stuffed animals, they have that goofy smile on their face. If you look at most stuffed animals, they have that goofy look on their face. They are just happy. It's like "Pleasantville." - This is what we say in neuroscience, they have a smooth cortex, not many shapes. - Exactly. And this, like Hedgy, saw it all. He was like Dustyesky's underground man. I mean, there's a sense that he saw the darkness of the world and persevered. And there's also a famous Russian cartoon, "Hedgehog in the Fog," that I grew up with and connected with.

People who know that cartoon can watch it on YouTube. - "Hedgehog in the fog." - Yes. It's just, as you would expect, especially from early Soviet cartoons. He is a hedgehog, as if sad, walking through the fog, exploring how loneliness and sadness. It's like, but it's beautiful. It's like a work of art, people should do it, even if you don't speak Russian, you'll see, you'll understand. - Oh, at the time you said I was going to ask, so it's in Russian? But of course it's in... It's in Russian, but what's more, very little is spoken in it.

It's almost an interesting exploration of how you make sense of the world when you see it only vaguely through the fog. So you're trying to understand the world. - We have Mickey Mouse, we have Bugs Bunny. We have all these crazy animals and you have the "Hedgehog in the Fog." - So there is a certain period, and this is again, I don't know what it is attributed to, but it was really powerful, that there is a period in Soviet history, I think probably in the 70s and 80s, when, especially The children treated very seriously. As if they were treated as if they were capable of dealing with the weight of life.

And that was reflected in the cartoons. And you were allowed to have really artistic content, not silly cartoons trying to make you smile and run, but creating art. Like things that, you know, cartoons or short films can win Oscars, like that's what they're going for. - So, what strikes me about this is a little bit how we were talking about the suit before, it's almost as if they treat the children with respect. - Yes. - So they have an intelligence and they honor that intelligence. - Yes, they are actually adults in a small body. As if you want to protect them from the true cruelty of the world. - Sure. - But in terms of their intellectual capacity or similar philosophical capacity, they are right there with you.

And for the cartoons to reflect that, the art that they consumed, the education reflected that. So he represents that. I mean, there's a sense that, because it's survived so long and because I don't like stuffed animals, it's like we've been through all this together and it's the same thing, sharing the moments together as friendship. And there's a sense in which, if the whole world turns against you and goes to hell, at least we have each other. And it doesn't die because it is an inanimate object. - Until you encourage him. - Until you encourage it. And then I'd probably want to know what he was thinking about all this time.

He probably really likes Taylor Swift or something. And I wouldn't even want to know anyway. - Well, now I feel a connection with Hedgy, the hedgehog that I certainly didn't have before. And I think that sums up the kind of possibility of connection that is possible between humans and other objects and certainly through robotics. There is a saying that I heard when I was a graduate student and that has been echoing in my mind throughout this conversation in a way that I think is appropriate, and that is that, Lex, you are in the minority of one, you are truly extraordinary in your ability to summarize so many aspects of science, engineering, public communication, on so many topics, martial arts and the emotional depth that it brings to it.

And just the purpose, and I think if it's not clear to people, it absolutely needs to be mentioned. But I think it's very clear that the amount of time and thought you put into things is the ultimate sign of respect. So I am extraordinarily grateful for your friendship and for this conversation. - I'm proud to be a friend. And I just wished that you would show me the same kind of respect by wearing a suit and make your father proud maybe next time. - Next time, of course. Thank you very much my friend. - Thank you.

Thanks andres. - Thank you for joining me in my conversation with Dr. Lex Fridman. If you are enjoying and learning from this podcast, please consider subscribing on YouTube. Also, you can subscribe to us on Spotify or Apple. Please leave any questions, comments, and suggestions you have for future podcast episodes and guests in the YouTube comments section. At Apple you can also leave us a review of up to five stars. If you would like to support this podcast, we have Patreon. That's patreon.com/andrewhuberman. And there you can support us at the level you want. Also, check out our sponsors mentioned at the beginning of the podcast episode.

That's the best way to support this podcast. Links to our sponsors can be found in the show notes. And finally, thank you for your interest in science.