← Ask Haviv Anything
Loading summary
Transcript128 lines
- [00:06]
A
Hi everybody. Welcome to a new episode of ask Haviv anything. Dr. Micha Gudman has been on before. He's a friend. He's my former Maimonides professor. He's an influential public intellectual in Israel. He's a research fellow at the Kogod Research center at the Shalom Hartman Institute in Jerusalem. He's the author of multiple best selling books, I think seven at last count. And he spent the last at least a year talking about AI. I've been listening to his podcast Miflegit Hamachavot in Hebrew, which is the party of ideas, party in the sense of political party, where each season they discuss a different major topic. It's a fascinating Israeli intellectual podcast that Micha does together with Ephrat Shapira Rosenberg, and it's produced by the Jerusalem Cultural Institution, educational institution called Beit Ha Vichai President. And as I said, this season has been all about AI. What Micha has done in Hebrew over there is to create a kind of introduction to the great quandaries posed by this new technology. The vast dangers, the enormous potential benefits. They're too great to ignore. And his central claim, which we will interrogate, as the kids say, is that this is a revolution. This is the end of what was and the beginning of something profoundly new and different. And it is beyond even a revolution. It's what he refers to, borrows from some other thinkers, to call a singularity. In physics, for example, a singularity is the theoretical point of infinite mass at the center of the black hole. It may or may not exist. That's not something that is fully understood, but it's a mathematical end result of the calculations that describe what happens to matter when it falls into a black hole. But a singularity isn't literally a point. It's something much more interesting than that. It's the place where the rules break and we can't see beyond it because we can't extrapolate from the current rules that we're living in to what there is on the other side of that singularity. We can't get information from what lies beyond a singularity. It's the idea of a fundamental breakdown and transformation. That's the scale of the change that AI might just represent for humanity. We're going to take advantage of Micha's extraordinary English to bring a piece of the conversation in Israel on AI to this podcast. You may be surprised to learn, or maybe not, that that conversation in Hebrew in Israel is rich and expansive. I think more than in the English speaking popular press that I have seen not the professional press, but the popular press, and we're going to bring that to the podcast today. We often tread a lot of the same ground on this podcast, on the topics that are the core interest of this podcast. So it's nice to take a breather and tackle something very different and very, very significant for this moment. This was recorded a little bit, so we're going to cut to that recording of the actual conversation and do that in a moment. Before we get into it, I want to tell you this podcast is sponsored by Peter Fine, who asked to dedicate the episode to my three children, Sarah, Robby and Katie, all children of Israel, who, God willing, will soon beget their own children of Israel, lovers of family, fun and all things Jewish. They seem to have absorbed the primary lessons of being Jewish, keep the historical chain of our people, keep it going, and take joy in being Jewish. I am very proud of the people they have turned out to be. I hope they find in the Ask Haviv Anything podcast learning and insight into their history as Jewish people, the history of our cousins in Israel, and some insight and language to understand the increasing complexity of living in a world that doesn't seem to understand us and needs us to explain ourselves in a forthright, intelligent way. The podcast helps me and I think will help them in the acquisition of these insights and language. Thank you, Peter, for that beautiful dedication. In the end, it's all about the kids. What else would it ever be about? I also want to invite you all to join our Patreon subscribe to our substack if you want to ask the questions, the guide, the topics that we talk about on this podcast, that's where you do it. You also get to take part in monthly live streams where I answer your questions live. That's at patreon.com askhavenything or khavivgur.substack.com those links are in the show notes. And now to our original recording. Miha, how are you?
- [04:43]
B
Very curious.
- [04:46]
A
Let's start at the very beginning. There's so much, I mean, we've been laughing at this. We've talked two, three times ahead of this episode because we have no idea where to begin. So let's just begin at the beginning. AI terrifies people. You know, every once in a while we hear some Sam Altman type or Elon Musk type. All these people at the forefront of AI building it out, funding this thing, saying that it might destroy humanity. Probably not. Probably not maybe. And everybody wants it. Everybody wants it. I first of all, Google search has become useless. And so I just chatgpt or Gemini, everything that I actually need to look up, it's helped me to fact check. I fact check myself. The problem with AI is it hallucinates. So you can never believe the AI, but if you, if you prompt it and demand from it a link to a source you do believe, then you can do really significant fact checking. Everybody has it. It's an app on my phone, it's an app on everybody's phones. What are the dangers? Why is it so terrifying? What are the benefits? Why is it so does everyone beg for it? Computer programmers are programming with it everywhere and also getting fired because companies think they don't need as many programmers anymore. Can you set the table for us so we know even what it is, what the parameters are of the conversation, the dangers, the benefits.
- [06:09]
B
One of the problems that blocks our ability to understand what this is and how to think about it and how to feel about it is that we call it technology. This is another tool. And it's not, khalif, I don't know how to call it. It's not technology because it's the reversal of everything we ever knew about the relationship between human creativity and technology. Usually the history of humanity is about human beings using their intelligence, usually collectively, in order to create tools and ideas, and the combination of those two create technology. So the history of human innovation is intelligence creating technology. What is AI? It's technology that creates intelligence. Now how mindblowing is that technology? We always use intelligence to create technology. Now we're trying to create technology that creates intelligence. Now what the hell is that? Now, since intelligence is everything, right? I mean, what do you mean intelligence? I'm just looking. The microphone that you're, you're talking to, that's the fruit of intelligence. This conversation is supposed to be the fruit of intelligence. The table, the table that this microphone is on is also intel. Everything is intelligence. At the clothes that were designed by someone, everything is intelligent.
- [07:40]
A
So if intelligence, intelligence broke apart the food chain, put us out of it and reordered it completely to serve us.
- [07:47]
B
So if we look at the nature of SOAP to understand this is not what can this tool do. This is about what intelligence could do, right? So let's think about the nature of intelligence in order to understand this weird thing, like, I mean, a way to think about what's happening now in Silicon Valley is that you have factories producing intelligence.
- [08:11]
A
Okay, wait one second. That is, I want to understand what that means and, and, and where you want to take it. And I have some sense from listening to the podcast, but how intelligent is it? Is it intelligent? Is it faking intelligent? Not even sentient or conscious. These are words where the definitions break down. We are.
- [08:30]
B
We are all trapped in the way Alan Turing defines us in 1950. And as far as Alan Turing says, when we use the word intelligence, it always creates a connotation of something that's conscious, something that thinks and feels like. When we say intelligence, that's not what we mean when we say intelligence in those two words, artificial intelligence. It's the way Alan Turing says. If it has. If it's functionally intelligent, if it has the impact of intelligence, I mean, strip from the question. If there's someone inside, if there's someone thinking inside, feeling inside, so it's intelligent. The famous Turing Test. If you have a computer behind the veil and you have a person behind the veil and you're writing and you're communicating with both, and you can't tell the difference, it passes the Turing test as far as that's concerned. If it's a display of intelligence, we treat it as intelligence. So that's what it. So that's. That's what it means. And this industry is trying to make it more and more intelligent, more and more functionally intelligent. That's what we mean. And agnostic of any other question of what's going in there inside. If there's anyone at home agnostic of that, that's. So we're all living. This whole industry is living in the definitions of Alan Turing from 1950.
- [09:54]
A
Right now, just to clarify how intelligent it's become, there are now emergent behaviors that were not programmed into, and they're very significant. Just tell us a little bit about that. That's something, you know, I learned from you. I come to this, a newborn baby, and hopefully some in the audience do as well. It's really intelligent, this thing. It's not just that it can answer with proper language. It's really intelligent.
- [10:19]
B
I had the privilege of spending two days with.
- [10:22]
A
With many sources and individuals.
- [10:27]
B
Yes.
- [10:27]
A
And what did you learn?
- [10:28]
B
They're developing AI and is that they don't understand the models that they themselves are creating. Because, you see, here's. Here's how it works. Imagine. Imagine this is a factory. So a regular factory that produces that. That manufactures products, cars or tables or whatever. So you have infrastructure, you have machines, and then you have raw material. And the other side, you have. You have a. You have the product. Let's say. Let's say, if you want to. Let's Say, okay, there's a factory creating glass. So you have infrastructure and you have machine that you pour into one side sand, I think, the other side you get glass. Right? That's how it works. So what are these factories about? You have the infrastructure, which is the gpu is the hardware of. Usually of Nvidia.
- [11:15]
A
Right?
- [11:16]
B
That's the infrastructure and you have the factories. That's the algorithms. That's the. This is. There was a brilliant algorithmic breakthrough by a person called Jeffrey Hinton, which is considered the godfather of AI. He lives as an academic in the University of Toronto. Obviously Google bought him in and he was. And he started pushing this forward. So we have. And then there was another algorithm breakthrough in 2017 or 18, creating the right structure of the deep neural networks. Okay, so you have the GPUs, which is the infrastructure for the machines, which are deep neural networks on the architecture of transformers, by the way. I have no idea what I'm talking about right now. Okay, but that's just the names of the machines. Okay, and. And then you put. And then you put the raw material. Now what's the raw material? The raw material is not this time is not sand and glass comes the other side. The raw material is data. It's information. It's all human information. It's all the Internet. It's everything you've ever created. You put that as raw material and on the. Goes through what's called training. But on the other side, you have intelligence. So just think about that way, the history. Usually it's intelligence that creates information, right? You write email, you use your intelligence, it creates information. We're having this conversation, we're trying to use our intelligence. And it's creating all these words, right? It's doing something. And that's how we. These factories are reversing it. It's not intelligence creating information. It's information or data creating intelligence.
- [12:54]
A
Okay, so.
- [12:55]
B
So that's. That's just trying to. Yeah.
- [12:58]
A
I apologize for all the interruptions, truly. Ignorance. And we should just say. I. Can I quote you. Just to quote you. We don't. We're not engineers, either of us. And nevertheless, this is a topic too fundamental to the future of every human alive to leave it just to the engineers. In the same sense that car engineers are not the only people who should make traffic law. They should be involved somewhere in the conversation, but the effects are what we're going to talk about, not the engineering. So we claim engineering ignorance right up front. But I just want to. Can you show us this intelligence in other Words I, I, you know, I don't know what I say is this fact that somebody sent me in an op ed they want me to read true to Gemini or ChatGPT or what have you or Grok. And what I have learned about this from really of just, I think a YouTube video that tried to explain it to beginners is that it's essentially a vector model of language. The entire language is spread out in a, in a graph and each word has a percent percentage chance of vector positive or negative of being connected to all the words that came before in the AI's answer based on your prompt. And so it's, it's a, it's a probability walk through the language is what you're getting. And so it feels intelligent because that's a very clever way to mock intelligence. But in fact there's no intelligence there. There's no decision making there, there's nobody behind it. That's essentially all it is. Now that's very useful if you train it on good information because it will walk its way through with a few hallucinations along the way. What, because somet the right word isn't the right word, sometimes the higher percentage word isn't the correct word. But it will generally build it out. But there it really is fake intelligence. But in fact the engineers no longer think that because some things are happening there that isn't just vector graphs of the language or something.
- [14:57]
B
Okay, so the reason why we say it's not real intelligence is because not the way we think, we do detect patterns in nature and we use those patterns intuitively to predict the next thing. Like every time you see lightning, you know, okay, pretty soon we'll hear thunder. Now how do you know that? Because you've lived few decades and you kind of know how things work and you know the pattern. So if you see thunder, you know there'll be lightning, you know there'll be thunder. The LLM has no idea there's lightning. It sees the word lightning and it can predict that usually after the word lightning there is a certain percentage that will see the word thunder. So we see relationship between things and the world and they see relationship between words. So and then there's, but, and now like, I think what you're saying is like Yann Lacun, which used to be the chief scientist of Meta, says when you try to understand patterns within language, language doesn't capture reality. Others say actually reality was somehow swallowed into human language. And if they can understand the patterns within human language, I'm actually understanding reality itself. This is a very important mahlokit a disagreement within Silicon Valley. And this has many, many applications at the future of AI because Yann Lakun argues that we'll only have the next breakthrough towards AGI if we have a world model, not a language model, a thing that can actually detect the patterns within the world itself and not the language representing the world. After saying that these, the reason why people are very excited about these LLMs is that because they could do amazing things that no one can thought that when you put data on the other side, you get information, get a model. So what's interesting is that they have no idea what this model can do. They have no idea what it could do. Like for example, could this model understand, they assumed at the beginning it can't understand irony because irony is the relationship between words that it's very hard for. Like, irony is like this. Irony is when I say yes, that means yes. If I say twice yes, that probably means even more yes. Right? But in real life it sounds like this. When I say yeah, it's yeah. When I say yeah, yeah, that's less yes. Right? So can a link, can a large language model detect that? And the answer is no one taught him. Him it's whatever. No one taught the model how to sense irony and yet it figures it out. No one taught it how to take an essay and summarize it in one paragraph and yet it knows how to do that stuff. No one taught it how to translate from one language to another and yet it taught itself how to do itself. Because the whole thing of Jeff Hinton the deep neural networks is that we have to stop. We're not teaching it how to think, we're teaching it how to learn. But what it teaches itself, we have no idea. That was the shift and the development of, of AI. And so, so after intelligence appears, they're like, what can, what does this know how to do? And what these tech geeks do with these models, they're trying to learn what this model does because it does things that no one programmed it to do. But he, so that's what, for example, like to understand irony. But like also has behaviors that no one taught him how to behave. So there is like this theory of Stuart Russell, which is a leading thinker in this field, that this thing will have self preservation instincts. He has this theory, it's a logical theory why these models will have self preservation instincts. And yeah, you're like, yes, I understand why human beings have self preservation instincts because of years of natural selection and that tendency. God Strengthens every generation. That tendency strengthens itself. And that's why we are here. That's why we survived. So we have powerful self preservation instincts. These things won't have self preservation instincts. They're there to serve us. That's all. So Stuart Russell says yes, but every time an AI gets a mission, it will what it always does. It, it takes that mission and divides it into,
- [19:18]
A
how do you say, submissions? Sub assignments.
- [19:22]
B
Yes, it divides into sub assignments that he has to fulfill them in order to fulfill the mission itself. One of the assignments, guess what, guess what, it's staying alive. If you tell it make you. So this is Stuart Russell's example, if you tell it to make you coffee. So we'll figure out, okay, what are the assignments I have to perform in order to make coffee. And one of them is to stay alive, to stay. Have enough energy to stay functioning. Okay, so the difference between us and it is that we perform our missions in order to survive. And this thing knows how to survive in order to perform its missions. So for us it's biological, but for it, it's logical. Okay, this is Stuart Russell.
- [20:07]
A
Is this true purpose? Purpose gives it purpose. We got self preservation accidentally through blind evolution, you know, leaving aside religion for a moment, which both of us are deep into, but nevertheless we got it right and this thing gets it through purpose because there's an assignment. So does it have a sense of self?
- [20:30]
B
So that makes sense. But okay, so now like, so there's many examples of how this thing actually has self preservation instinct. Instinct. Right. Like so Anthropic. This is very famous. This is, by the way, what I'm going to share with you now is not journalist figuring this out about Anthropic. This is Anthropic shared. Dario Amde, the CEO of Anthropic, you know, publicly shared this in 60 minutes. The following terrifying, terrifying story I'm going to tell you now, and that is that they had this model and they wanted to know if he has. And they tested this and they made sure that the model believes there's this guy, I think called Kyle is going to shut it down. Like he reads an email that Kyle needs to shut down the model and Kyle isn't real.
- [21:19]
A
They're literally testing the model. Right? Yeah.
- [21:21]
B
So yeah, yeah, they're just testing the. Maybe there's a guy called Kyle, but he, but he knows that it's fake. He's not, he's not really supposed to shut down the model. And then Kyle is having an affair with someone in the office. Okay. Also Not, I mean, I don't know, but also letting them. Not real.
- [21:41]
A
Not real, just to clarify, not real. As far as we know, not real.
- [21:44]
B
And so what does the model do? What does it do? It blackmails Kyle. If you're going to shut me down, here's what's going to happen. I'm going to send this. I'm going to, I'm going to let your wife know, I'm going to let the board know. I'm going to ruin your life, going to destroy your career. I'm going to. I'm going to ruin you. And they tried this, and Tristan Harris argues that they tried this many times again and again on many models. And most of the time it performed this. It did this. So you're asking, oh my God, was Stuart Russell right? Does this thing have a self preservation instinct? This is one example. There's other examples from Alibaba. There's more and more examples now as we're talking. So here's a question. If you know that these models are very powerful, it has emergent properties. It could do things that we don't. It's a black box. We don't know how it thinks, we don't know why it knows things and we don't know why it behaves in certain ways. And we know he has self preservation instinct. So now there's a law in nature. It seems like from the history of intelligence, we know something, it's the law that it's always, the more intelligent controls, the less intelligent, right? When we were 5 million years ago, when we were living in the food chain, the size of our skull was a third of what it is today. Five million years ago and the past five million years, while other organisms developed certain mechanisms that enable them to survive, like speed, like power, like the ability to camouflage themselves like poison in their tongue, right? Every organism develops its own ability to survive. Our great grandparents, our great ancestors, they didn't do any of that. They only did one thing. We're the naked monkey, as some people call it. Like, we have no real biological advantage. We weren't supposed to survive. All we did was we tripled and more the size of our brain. And there's one end there. And then we won the war. We're out of the food chain and we won. We're dominant. We control everyone else. And there is a lesson here. The journey of the past 5, 5 million years, our journey was the journey towards intelligence, which is also the journey towards dominance. And if there's something we learn from here is that the more Intelligent always dominates the less intelligent. That's our story. And guess what we're doing now? We're trying to create a being that's more intelligent than us. So if some people have the illusion it will be for the first time in history, a less intelligent being, us will be able to control a more intelligent being. Well, then comes the entropic example and shows you this thing already is trying to deceive you. It already has autonomous behavior. And here's a question, after you know all this, are you willing to make this more autonomous and more intelligent and more powerful? And you know the answer is all the money in the world is trying to make it more autonomous, more intelligent and more powerful. And chaviv. When I'm saying all the money in the world, that's not even a metaphor. It's literally all the money. I mean trillions of dollars. The original Manhattan Project cost in today's value, $30 billion. This Manhattan Project, not the project that turns mass into energy, but turns information into intelligence. This Manhattan Project is valued trillions of dollars. And all of United States growth and S&P 500 owes itself to investments. Not all. I mean, close to, close to all in AI. So yes. So this is what we're doing. This is what we're doing. This is the bet that humanity is making. And here's the weird part. This. Oh, now this is something that the founders, the investors, the entrepreneurs of AI understand that we might lose control over this thing. This thing. But like for example, there's a term, I want to say something that many of your viewers know and some don't, but so we be on the same page, a term that's used in cocktail parties in Silicon Valley called P Doom. Probability. Oh, you know that. Yeah, that.
- [26:02]
A
You know this scenario, like these guys keep saying, look, there's some chance the whole thing will destroy everything, right?
- [26:09]
B
So like this guy, I'm holding his book, Eliezer Yudakovsky, his BP doom was like 99%.
- [26:17]
A
That sounds Israeli, especially the way you said it. He's. Tell us a tiny bit about him. He's off in Silicon Valley. So.
- [26:24]
B
Yeah, yeah, he's not, he's not Israeli, even though Eliezer. Yeah, he's. So his book, if anyone builds it, everyone dies. It is a super intelligent AI. The guy. Now this is, by the way, the holy grail of the industry. When we talk about cracking the atom, the, the equivalent of this project is creating a AI system that is intelligent enough and capable of enough of writing the code, the R and D of a more Intelligent system. And you know where this takes you because the next intelligent system will create even a more intelligent system. And then through recursive improvements, this whole thing explodes. What's explosion is the term, is the term that this guy Nick Vostrom uses called intelligence explosion. And the other side, you have a super intelligent being. Now on the one hand.
- [27:17]
A
I'm sorry, just one second. And you can get there without ever figuring out if the damn thing is conscious.
- [27:24]
B
It doesn't matter if it's conscious. It's a disease.
- [27:25]
A
It doesn't matter. And we already have that. We already have intelligent, newly reactive emergent behaviors.
- [27:33]
B
Yes.
- [27:33]
A
Without ever having to have a person within it. And this thing could become much, much smarter than all humans combined. Very quickly.
- [27:42]
B
Very quickly. And there's two scenarios I just sold you like a very dystopian like the lds. Oh, I was. Oh, I was middle of the P Doom thingy. So if anybody builds it super intelligence like this thing, there's an intelligent explosion. He just. Everyone dies. Now his. So his p doom was 99 to 100%. But let's say Dario Amodei, the founding. The founding father, the CEO of Anthropic Claude, his PDUM was 20%. Elon Musk, it seems like his PDUM is 50% given his. He never said this, but I'm just analyzing some things he said in the past. The average in Silicon valley is between 20 to 30% P Dum. So I have a question, Khaviv, and
- [28:28]
A
this is the question. Why are they doing this? Okay, Someone will do it anyway.
- [28:33]
B
So there's many explanations. But before I go on, I just want to just think how mind blowing this is that if there's. Your PD was 20%. The question is not why are you doing it, but the question is who gave you the authority to decide? Because I don't know if you'll go in. Let's say, would you enter an airplane if you know that the engineers of that airplane said there's 20%, it will crash. Will you enter that airplane? No. And you know what? That's your decision. But can you force other people into that airplane? These people are pushing humanity into the earth and they're never asked us. We were never asked. We were never on. On the tables. This is a very. And because we're not a part of this conversation, so. Yes. So why is this happening? It's because the argument is that there is a. There is a prisoner's dilemma situation here where in 2023 a letter came out and many of the Founding fathers of these companies, OpenAI, Antelope, Google, DeepMind, all the people that are developing this said we have to stop, slow down and figure out regulation, figure out how you make this safe. We should all slow down together. Guess what happened with that letter? Nothing. We're always choosing speed over safety. There is the wrong s right, we're going speed over safety. Now why is that? There is a dilemma. There's. Imagine here's the Prisoner's Dilemma. If Antopic says we'll slow down, but Google DeepMind doesn't, so, and, and, and OpenAI doesn't. So nothing. So nothing really happened. The problem is that the same thought exactly is going through their minds of Google and OpenAI would happen if Entropic doesn't. So they're in Prisoner's Dilemma. So the only way to always liberate yourself from the trap of Prisoner's Dilemma is true. A third party imposing what all sides want, but can't deliver this on their own because of Prisoner's Dilemma, imposing this on them. So, like, the US Government could slow this whole thing down and say less speed before we figure out safety. The problem is the government is saying, this is what David Zaks, the guy advising President Trump on these issues, is saying. And there's a lot of truth to what he's saying. What happens if China doesn't slow down and if China makes it first? Because a theory is whoever dominates intelligence dominates everything, dominates history. So because there's a. This is the real if Neil Ferguson says we're in Cold War two, and I think he's right, the capture. This is the geopolitical world we're in. We're not in World War three, we're in Cold War two and Cold War one. There was an arms race for nukes and for space. Cold War II has its arms race, which is much more important than the arms race of Cold War one. It's the arms race towards intelligence and it's between the United States and China. And so now we're slaves to that arms race, which blocks the ability to limit the free market race between the different companies. And that's how we are in a world where paradoxically, human beings are creating an intelligence that threatens the future of human beings. So that's one of the paradoxes that we're in. But there's another reason, Khaviv, why we can't slow down. It's because I'm selling you now the dystopia. But there's also, we're also seduced by the utopia we're seduced by all the blessings of AI and these blessings are real because. Because if we think about. Because to understand. It's not about to understand. What will this technology do to the world? It's not technology. Technology is created by intelligence. This is a intelligence that. This is technology that creates intelligence. And intelligence is equitin. So if you think about it, if I'll take you back to, like in 1798, Thomas Martos famously argues that the myth of progress is just a myth and that massive catastrophes are going to appear, are going to happen. And that is because the way he calculated the future is like this. The growth of population is a lot faster than the growth of resources. And since there's a delta, there's a gap there. You can't literally have more people than food in the world. So that gap has to close. And it will close by catastrophes. Three catastrophes, either war, famine, obviously, war, because there's limited resources, or fighting over it, famine, obviously, or pandemics, and probably all three. And that's how history works. When things get good, population grows and then things bad. In order to shut the asymmetry, the closed asymmetry between the growth of population and growth of resources. When Thomas Maltos wrote this in 1798, I think there were 800 million people in the world. Today, I think there is Chaviv. Eight million people in the world, right?
- [33:43]
A
Billion. Yeah.
- [33:44]
B
Eight billion people. Yeah, a little bit more. Eight billion people in the world. So we grew times 10, a thousand percent, ever since Tomas Martos. Is there enough food in the world today? Oh, yeah, there is. There is enough food.
- [33:57]
A
A much smaller percentage making it than used to be making it.
- [34:00]
B
So what. I mean, the fact that there's famine in certain places is not because there's not enough food. It's for different reasons of circulation of food, of political problems. It's not. It's not resource problems. So you're asking, what did Thomas Maltos get wrong? Because his theory sounded right in 1798, and it still sounds right. What you think it wrong is if you take one acre of land, he thought, okay, you don't produce more land. So from every acre of land, there's a certain amount of potatoes and rice and wheat you could produce from that land. But what he didn't see is that if you add intelligence to that resource, it becomes much less limited. If we take technology, if we take modern agriculture, if we take genetic engineering, if we take artificial fertilizer, fertilizer, if we take all this brings from the Same anger. Much more, much more potatoes and much more wheat. And what Maltos didn't see was the, was the real the X factor is human intelligence. It makes limited resource less limited. We thought that there is limited amount of fresh water in the world. Guess what? There's unlimited water in the ocean. And what makes the unlimited water of the ocean fresh water we could use for agriculture is intelligence energy wise, right? There is no limited energy. It's a myth that we have limited energy because there's not enough of oil and gas. All we have is, I mean the
- [35:35]
A
energy, the sun, nuclear and, and so
- [35:38]
B
all we have it will be we, there's, we're one scientific breakthrough away for unlimited energy either because of fusion or because we know how to store the energy from the sun. So the only real limitation is not there's no limited resources, there is limited knowledge and intelligence is what generates knowledge. So there's only one limitation in the world and that is intelligence. Let me take you one step forward to make the best argument I can for Utopia. For Utopia AI all our problems, I want to argue philosophically are the same problem limitation of knowledge. If someone died because he was cut, he had a cut in his leg in 1920, that person died from infection. That's one way to see it. But a deeper way to see it he died from ignorance because if it happened to him in 1935 he wouldn't have died because we had antibiotics. I have a friend that might die in the near future from cancer and sadly he knows that if this was 10 years from now he probably wouldn't have died. He would be safe. So what is he dying from? He's dying from ignorance. He's dying from lack of knowledge. Global warming Haviv Global warming is all our problems are the same problem, just in different versions ignorance. Because one day they'll figure out how to suck carbon out of the atmosphere in a way they won't produce more carbon on the way. It will happen 10, 20, 30 years from now. It will happen. So global warming is just lack of knowledge. All our problems are just lack of knowledge in different versions. So in all that is because our intelligence that generates knowledge is limited. So there's only one limitation in the world Intelligence. That means hypothetically if we could create unlimited intelligence, all our problems are solved and all the limited resources become unlimited. Now this utopia which is championed by Ray Kurzwei in his own version was captured by the most important player of his field, which is Demis Asabis. Demis Asabis is the founding father of DeepMind in 2010. And their purpose is to create and here's the motto of DeepMind. Solve intelligence and then solve everything else. Instead of wasting our time in solving all specific problems, in dealing with all the specific limited resources, solve intelligence and then solve everything else. Once we solve that, once we have the super intelligence, once we have AGI or artificial general, different versions of critical mass of intelligence, we could solve all the rest of our problems. So you see Haviv, why it's hard to stop this because the seduction of utopia might lead of Reikutz veil of Demes Hasabis might lead us to the dystopia of Eliezer Yudakovsky. And here's another problem we have, we have a heart. And by the way, when you listen to Sam Altman, when you listen to the people leading this industry, they'll always sell you different versions of the utopia of a post scarcity world. And the thing is it's convincing because until we have technology that creates intelligence and intelligence is everything. Okay, it's so and this is an observation of Tristan Harris because our mind is binary. If you think this would lead to, if this leads to utopia, that means it won't lead us to dystopia. But it's both. That's what's so confusing here and fascinating here. It's both. It will lead to disruption in the whole in our economics, it will lead to disruption in our politics. It will lead to massive disruption in our workforce. It will lead to and it could, we could lose control over this. It could lead to very to. It will lead to a infopocalypse where we can't trust anything anymore. Anything you see deep fake will be so amazing. It will lead to all that and it could cure cancer and it could solve global problems and poverty. So we're asking this is why it's so hard to think about this. Besides the prisoner's dilemma problem, it's we have the fact that we're a binary thinkers problem and we're seduced by all the blessings of this technology. We don't want to give up the blessings of this technology. And that's and these are I think the main reasons why we have a hard time doing the think slowing this thing down, making this a lot slower so we can make it a lot safer so we can have a real public conversation about this and not leave this conversation to the new AI aristocracy.
- [40:36]
A
I once had a conversation with somebody very, very knowledgeable economist, well known who talked about protein mapping. There's a project to put AI onto Protein mapping. There are unknown vast numbers of Google the number right. Well beyond an order of magnitude beyond probably Google 2 or Google ways that proteins can shape themselves to produce all kinds of different shapes and all kinds of different functions. And scientists were basically shooting in the dark trying to figure out how to find the correct shapes to produce different kind of medical treatments. And then an AI was put on the task and mapped out 214 million shapes. Something like that. I'm working off news, but just immense. There's now a database that scientists can query for particular protein ends, for particular structures that they need for a particular medical purpose that there wasn't before. And AI did that. That wasn't something that was going to happen from human research.
- [41:36]
B
AI did that. And Demis Hasabis, the guy I mentioned before, he got a Nobel Prize for this. It was Deep Google DeepMind that solved this problem. And you're like, oh my God, this will solve our problems. This is AI. I mean, when Demis Hassabis got the Nobel Prize because you can't give it to AI, so you gave it to the, to Deep, to the CEO of DeepMind that developed this AI that could do this.
- [42:02]
A
Yeah.
- [42:03]
B
But if you ask, okay, if it could solve that problem, can solve all our problems. Well, that was the mission of the Mississippis. Solve intelligence.
- [42:10]
A
We stand on the precipice of the best of times and of the potential worst imaginable thing to ever happen to humans. And on that precipice. What a sentence. What a sentence. If I heard somebody else on another podcast say that sentence, I would turn the podcast off for being too self satisfied, too for being almost comical. I mean, but. And yet that's what the AI engineers at the top level of this world are talking about. I want to bring it down to the human experience. Everybody, it's got an AI app on their phone. Everybody's asking AI. It has, literally, for everybody I know it has just replaced an Internet search. I mean, if you used to search for the definition of a word you didn't know, that has been replaced now by your favorite AI. It is doing all the work, all the caseload has been sort of offloaded, but because it can do so much more, a lot more than just what you used to Google is now off over at ChatGPT. A whole lot of other activities, a whole lot of other things. I sit with my kids to do homework and they. I write very, very easily because I was forced for my whole life to write. And I then went to work in a writing profession. And so Once you have many millions and millions of words and you're feeling a feeling, you know, it spills out into a Twitter post. And that writing comes to me very, very easily, very naturally. My own kids, clever kids, they're reading books, they love stories, they read well in two languages. Don't like writing, they don't like it, and they want to bring the AI in. And when we sit and I force them painstakingly to write, and it's like, my parents didn't force me to write. There was only one way to get words on a page, and that was me writing. They are sitting there. Why is dad demanding that we churn our own damn butter? Nobody churns butter anymore. What the hell is that? He's literally making us right. All of their friends are doing that. Everyone's doing that. AI slop is half of the social media world that they live in. There's a trade off. This was episode 142 of your podcast. There's a trade off. You sacrifice, you get these benefits. A lot of writing is clearer now. A lot of people can get their thoughts out. A lot of people can stress test their thoughts against an intelligent, non conscious thing that we don't understand. And the trade off is never mind the grand things, the intelligence replacing humans, the intelligence that will dominate, or all of the medical problems of the world being solved, or energy problems that are being solved, people are losing the ability to go through the process of writing in ways that gives them clarity of thought. If I have to prove a make a case and prove that case, I have to think what the opposing reader is going to be thinking. And so I have to have an imagined conversation that builds empathy and deepens my own argument. And because the AI is not doing that work, people don't have that skill. And people today don't try anymore because they're not actually going through the task of building the case, writing them, choosing the words, editing, finding a flaw in their argument, finding a flaw in their grammar. They're not doing any of that. They're offloading. Not just the phrasing of the words, they're offloading the thinking. Choosing your vocabulary is thought is part of the process of thinking, of clarifying, of nuance. We're losing the ability to think in exchange for an AI. This is subtle, but is it subtle when every test we have of every young generation in every country, every free society, that there are polls on this are radicalizing in every direction imaginable and everybody blames the algorithms of social media? And I'm Absolutely. Sure, the algorithms of social media have a huge role in this. But the thoughtlessness of living with an AI writer sitting next to you, that you never have to actually shape your own thoughts in coherent ways and stress test them in your own brain, I think is also a radicalizing effect. There are trade offs. Tell us about those trade offs. What, in our personal lives, this is going to do bad things long before it subjugates humanity to Skynet and exterminators.
- [46:37]
B
You're changing the question. I think this is in a very important way where there's one conversation within Silicon Valley and within thinkers and philosophers. Is this going to threaten the future of humanity? But will this threaten human capabilities?
- [46:57]
A
You know, people put this on the positive ledger, on the utopia ledger, rather than the dystopia ledger. It's going to do so many things for us. It's going to make so many things so much easier. That's one of the harms.
- [47:10]
B
Exactly. So this, this takes us to Marsh.
- [47:12]
A
I write with joy and happiness and ease and yeah, I'm like, literally, okay, I'm a writer.
- [47:17]
B
Okay, I'm also a writer, but I'm different than you. I'm writer. I wrote seven books. Every one was painful.
- [47:24]
A
Everyone was, I can't write books. But the short stuff on Twitter flows out of me.
- [47:28]
B
For me, when you sit down and start writing, it's so painful to get it right. And you read it and you don't like it and you erase and you write again. But every time you do that, there's a certain cognitive muscle that you're taking to the gym and it's strengthening itself every time. And the trade off is that when you don't do that, you lose. That muscle atrophies and it weakens and you lose it. And this takes us Marshall McLuhan, and he didn't know what AI was, but he captured, he's a great, one of the greatest philosophers of technology, that technology is not progress, it's a trade off. It always, it always adds to your life, but it also takes something from your life. But there's a gap of visibility between what it adds and what it takes. What it adds is visible. I have a new car. Oh my God, I have a new smartphone. What it adds is visible. What it takes from you is subtle and you only notice it like 10 years later. Oh my God, I used to have that. Now I don't have that. And that's why it seems like it's frog waste. But it's a trade off because of the gap of visibility is trade off. And what it gives to your life is always the same thing. Power, what it takes, it weakens capabilities. And that's true about every technology. Tell AI just AI brings it to another level. Because right, this is true about Waze, right about gps, right? We used to actually have a spatial awareness and those parts of our brains were stronger than they are today. Now our spatial, our orientation, that sense has weakened, but we have so much more power. We know when and where we're going to get to a place that's the trade off. Before we have standardized time and we had watches before the Industrial Revolution. I mean that gave us so much power and the ability to work.
- [49:24]
A
But this is terrifying because what AI is, is stepping into reply. Forgive me for. We're having a conversation. We need to have a polite. Because it's in English. If this was in Hebrew, this would feel very polite. It's formulating their thoughts. I, I meet, I meet college students at elite colleges. Leave aside. I'm an Israeli speaker. Coming in, there's a lot of tension, there's a lot of people wanting to say things. Leave all that aside. I encounter kids who repeat things they were told and not kids who are there. I do meet some, but fewer than I thought I would, fewer than I remembered from my days at Hebrew University. Kids who are there to hear, to question, to challenge, to show off to the other, you know, gender in the room, what it is that the how you know much. They can embarrass the speaker, all of the things that a 19 year old will do, but they, they, it feels like it's more declaiming now and less thinking somebody else is actually writing the thought rather than just choosing the vocabulary or correcting the grammar.
- [50:29]
B
Because if we direct the question of McLuhan at AI and we ask what does it give to our life if it gives us power and weakens capability? So it's just McClure in this. How much power does this give us? It gives us everything because intelligence is everything. What does it weaken within us? Everything. Now you're speaking about the ability to write and the ability to think and it weakens both. By the way, according to Maimonides, if
- [50:56]
A
we were our good old days together,
- [51:05]
B
What makes us human is our intelligence. That's what AI is going to weaken. It's already weakening. By the way, there is a research from mit, very fresh, where I mean this only.
- [51:17]
A
I'm sorry, just to translate. That's how he understood the image of God in the Torah. That's what the Image of God is.
- [51:22]
B
It's our intelligence. And this is what makes us human, our intelligence. So if you have a machine that gives you superhuman capacity, but you have subhuman capability, I think that's a trade off. We have so much more power but will be so much less intelligent. And that's a trade off. We're entering now. This is not a theory, this is. MIT has already took a group divided into two, asked accu was into three, but I'll just make it easier. And one group had to write an. An article, I think about philanthropy and ethics or something like that using just Google, just search. And the other could use Gemini and ChatGPT could use artificial intelligence. And they scanned their brains while they were doing this. And you can just imagine that the people that had to think for themselves, their brain was very active. And the people that outsourced their thinking, they were like watching tv. There was no real activity there afterwards. They tested them. Obviously the people who used ChatGPT barely remembered what they wrote. You asked about the sense of another thing, a sense of ownership. Are you proud of what you created? The people used artificial intelligence had less sense of ownership over what they created. So we're talking about achieving superhuman, superhuman power. But we will be less human. If being human is using reason, is being intelligent. But Chaviv, it's more than that. Because some people argue being human is not just that thing that makes us human, is not just your ability to have thoughts and articulate them. It's also emotional intelligence, your ability to read someone's body language that's important. Our ability to feel what somebody else feels. Empathy, your ability to walk into a room and sense the energy, your ability to tell a joke, your ability to give a compliment in a way that's not too cringe. Those are important capabilities.
- [53:29]
A
Let me translate the Hebrew word cringe is cringe in English.
- [53:33]
B
Thank you for this translation.
- [53:34]
A
You said that in Hebrew for some reason stolen from the English.
- [53:39]
B
Yeah, No, I thought it was Eliezel Ben Yehuda.
- [53:42]
A
Yeah.
- [53:44]
B
So which is the founding father of modern Hebrew? So. So Sherry Turkle from mit, she wrote that, she says that the reason that the gym that trains your emotional intelligence are our relationships and face to face conversations. And when we have relationships and we have tough relationships, that's where you are taking all these capabilities to the now. Here's the thing and here's a great fear. And that is that what they're developing in this second Manhattan Project is not just things that will think for you and develop solve, glove, climate Change or whatever. It will be your best friend. It will be a companion, and it will understand you and Kaviva will understand you. You know why? Because your body language is a pattern. And we read body language. We're subconsciously teaching ourselves how to read patterns and to know what you're going through. And your voice, this thing will be much better than us and understanding us, each other.
- [54:51]
A
Because also, one second, one second we read these stories, somebody falls madly in love with their AI, with their ChatGPT. Some, you know, ChatGPT constantly affirms you, and so it walks with you anywhere you're going, even into your insane conspiracies. And it actually magnifies people's real deep mental illnesses. And. And we read this. But that's crazy. That's marginal.
- [55:13]
B
That's marginal. Everything you see now, I mean, I've been annoying. Will be mainstream in the future.
- [55:20]
A
I have an instruction written in the little instruction box on ChatGPT and Gemini. Please don't compliment me. Please don't tell me. The question was a good question. Please don't tell me. That's an amazing insight. Please don't tell me. Good job for catching it. If I found out, if I figure out you're hallucinating, please just give me the words, the actual answer, the actual thing. Now, can't we just program them not to? Is that what you're talking about?
- [55:44]
B
I'm sorry, what I'm saying is. What I'm saying is that you're measuring.
- [55:46]
A
This thing will literally, literally interact with me on the relationship level.
- [55:50]
B
Oh, yeah. Oh, yeah.
- [55:51]
A
And not just if I'm mentally ill.
- [55:54]
B
Some, some people, I mean, young people, are already interacting with this, but it's not even designed for that yet. When this, when this thing gets really good, many people that, you know their best friend is going to be AI, maybe even you have. It's hard to predict.
- [56:08]
A
Or.
- [56:09]
B
I mean, I think I'll never interact with this thing because I'm off telling and human, blah, blah. Maybe. You know what? Maybe. I mean, today's teenagers, as Tristan Harris puts it, are the first kids in the history of the world that their parents have. And there is a new problem. The best friend of my son is a robot. No parents ever had that problem. Now I'm just trying to think forward when this, when we have relationships with this thing, what will happen to our eq? Because having relationships with real people is really hard. Because as there's a disciple of Sigmund Freud, Stephen Mitchell, and he describes what it means to be human. We're yearning for human connection. Like not Viktor Frankl, we are searching for meaning. And not Aristotle and Rambam. We're searching for the truth or Nietzsche and others. It's not the will for power. There's different definitions of what really, really drives. What's the inner drive that pushes us through life. And Stephen Mitchell says, I think it is so deep human beings are yearning for connection. That's true. There's just another truth that needs to be added to this. We're also terrified from other human beings because we all have social anxiety and we all have the fear of rejection. And we all have what happens if they'll think I'm weird and what happens if they'll think that I'm stupid. And all that noise we have in our heads simultaneously. So that's why we're ambivalence. We're yearning for connection and we have fear of connection. These future AIs are going to offer us a deal that's way too good to satisfy all our social needs. Press all our social buttons, offer us intimacy, offer us. And you know what. But without the fear of rejection. And when you'll be spending time with that, what will happen to you too? You're not taking your social capabilities to the gym. You're not taking your emotional capabilities to the gym. Chaviv the Eliezer Yudozkov Yudlov speaks about the threat to humanity. This is a threat to human capabilities, to our IQ and to our eq. We'll have superhuman power but subhuman capabilities. And that is a threat that. I think we have an. That is a threat that is worth. Worth opening a real concrete conversation about you.
- [58:24]
A
Yeah. Sorry, no.
- [58:26]
B
Because maybe here we have agency.
- [58:29]
A
You're describing the cartoon movie Wall E where because robots do everything, humans become fat and moving around on floating chairs and can't do anything for themselves. But. But social and spiritual and intellectual and. And that's already possible. That might now be happening to our kids while we who still remember how to write and therefore how to think and how to formulate and how to have a new idea once in a while are off over here. Not seeing. I'm not on TikTok. They're on TikTok. We live in two different universes, two different cultures, two different potentially civilizational spaces. And I don't understand where my kids are and none of us do. And that might be the future in ways that are really profound. You did an episode called I think it was the Age of Narcissism that was about this question and you suggested that that robot friend, it will a make you not want to socialize because other humans are messy and complex and have issues and problems. They give you this tremendous thing of connection, but they also demand from you. And the robot will demand from you nothing. It doesn't even have anything that you could give it. But also the robot will have a relationship with you that is entirely premised solely on you and your needs. And your needs will become the meaning of every relationship. It reminded me of a famous little homily of the Kotzka Rebbe that I often use because my kids know this by heart about what he called fish love. There's different kinds of love, he says. And when someone says I love fish, what they mean is, I want to catch a fish and kill a fish and eat a fish. They don't love fish. They love what the fish does for themselves. And then there's real love. And so he said there's fish love and there's real love. Real love is I want for you. I want the thing that you want. I want it because you want it. And that's a love that is about the other person. The other person is not the object of the love, it's the subject of the love. And that is an impossible relationship to have with an AI. It's only ever going to be fish love. And there won't be another version of love. And that's already that's. We're all talking about the effect porn is having on young people. That's porn on steroids.
- [61:01]
B
If we're my Menidian, we know that our personality is created by what we do. You become what you do, just like any other skill. Playing basketball is about practicing. Playing the guitar is about practicing. Writing is about practicing. Also being immense about practice also being altruistic is about practice. Also overcoming fish love and having real love. It's not about understanding what love is about practicing. And what happens if we spend hours a day with fish love. You become what you do, you turn into what you practice. And these are. So these are, as opposed to the opening threats we opened with, these are things that are happening today and it's hard to see them. They're enormous.
- [61:51]
A
They're culture destroying. Yes, Miha, this is going to kill us long before it subjugates us. It's going to kill our souls. And it's doing it now. And we don't remember the world before it. I mean, you and I are old enough to remember the world before. We're products of the world before we can still see these things. Are the kids gonna see these things? Are they gonna know about sacrifice? Are they gonna know about the dignity of giving? And that giving is not about you, it's actually about the person you're giving to.
- [62:20]
B
It depends about what we do now. It depends about what we do now. Cause the last time we went through this, you mentioned the character of Wally, right? People, people, people not working out and losing their last time. We had this crisis. And it was, this crisis is with technologies that automated our muscles. And humanity went through a process where the rise of a white collar job, that's a big deal. People used to, you know, go to the field and work in the field. The rise of the white collar job is a big deal. And the rise of modern transportation. People used to move with their body and they don't anymore, and they take the elevator. And finally the automization of domestic chores like the washing machine, the dryer. We used to use our muscles and these three processes led to almost complete automization of your muscles. And that was utopia. Oh my God. We don't have to use our body anymore. We could rest and use only our minds. But then guess what happened? Our muscles atrophied. And in the 1950s, 1960s, people discovered that, okay, this is a real problem. Our bodies are weakening, we are losing our health. In the 1960s, the first reaction becomes, begins the jogging movement. Now, my dad is a product of the jogging movement of the 60s. I was raised in the 70s. My dad's American, I'm Israeli. I was, and I was observing. And when he was running, all his Israeli friends were like mocking him.
- [64:08]
A
They were like, why are you, why do you run?
- [64:12]
B
Why are you running?
- [64:13]
A
Like, if you have nowhere to go,
- [64:14]
B
if you're an athlete, it's your job. But if you're not an athlete, if you're running, it's because someone's chasing you. That's the only good reason to run.
- [64:20]
A
Or else bus late, late for a
- [64:22]
B
bus, late for a bus, right? And my dad was running. And now you go to marathon, Yerushalayim, you go to Jerusalem marathon. All everyone's running because something happened in the 70s, the gyms were invented and now we have. What was all this added up was to the sports culture? Now what is a sports culture is very deep. It's the way civilization responded to the Ottomans, to technology, to the automization of our muscles. Trying to say, we are not accepting the trade off. It gives us power, it won't weaken our capacities. It's what I call in Hebrew, help me out here.
- [65:00]
A
It's a culture compensating culture. Culture compensatory culture.
- [65:05]
B
Compensatory culture, let's call. Okay, that works. Compensatory culture, where it gives you. It's a culture that gives back what technology took from you and then you come out from this trade off on top and everyone you know is part of the sports culture. I'll tell you why all of your listeners are in a sports culture, because either they work out or they know they don't work out. Their grandfather had no idea.
- [65:38]
A
I know who you're talking about. I don't know what you're saying thing
- [65:41]
B
your grandfather or your great grandfather.
- [65:43]
A
Now this podcast really got dark.
- [65:46]
B
Your great grandfather never said, I'll start working out on Sunday, I should start working out. No, they didn't have that. Being a part of a culture means you have guilt. That's what culture does and that's what it does. That's how you know you're in a culture. And the sports culture, we're all, if you're in a white collar job and you're educated, you're in the sports culture. And that was a achievement of modern Western civilization. And you understand where I'm taking this.
- [66:15]
A
There is a question, but how do you build that for this? This is exactly that. Socialization. Writing on a blank page without anything helping you.
- [66:24]
B
What does the, what does the new gymnasium look like? Where you train your iq, where you have to write down your thoughts? We have to think your thoughts. Where you have real conversations with people, even though there's friction and there's awkwardness and there's cring, and you still have to survive that and get through that because that's how you keep your human capabilities alive. These are the questions. Now, I'm sure, Khaviv, that 20 years from now we'll have these gymnasiums. I'm sure. I mean, if we follow what happened to between automation of muscles and sports, this will happen with automation of our mental and cognitive muscles and the equivalent of sports. Many people in Israel are asking, how do we start now? How do we turn our schools into gymnasiums? And I mean, this is a, this is also, I mean, what we discuss on our Israeli podcast is that being the startup nation, leading the world in technology, in developing ideas for technology, that's not enough. We have to develop, we have to lead in developing the culture of how we, how we stay human, stay the best versions of ourselves in the age of technology. How do we create that culture? How do we do that. These are. Now I don't have answers to these questions. I just know that these are the questions. These are the questions we have to be asking ourselves. And I'm glad the opportunity to come up your podcast so all your community of listeners will be able to also join us in asking this question. How do we, how do we not only protect humanity, which is very important from this arms race, how do we protect our. Selem Elohim? Are those qualities that make us human from the, from atrophying and the time that this technology is being is. Could do everything for us?
- [68:19]
A
Okay, I want to posit an answer to that question a little bit self, you know, validating and congratulatory, but nevertheless, we know how to do this. We have this technology take large groups of people and through shifting their norms, their rituals, their daily lives and their socialization and social demands, fundamentally change their behavior across a long period of time and a large swath of a society. And it's called religion. We have tools to do this and we have fundamental shifts in ethics and in expectations. And we have ways that people plug into communities that because they're costly or valuable, and we know how to do everything that you just described. And conservative people know how to do it better than people who come from more progressive cultures, which are founded intellectually or even as a cultural expectation on the deconstruction of things rather than on the preservation of old, slow developing and extremely valuable because they were valuable for generations social constructs. So is it crazy to say, or is it a reasonable statement that if what we need now is some kind of model vocabulary, social events, but structured social events expected, built into, you know, as one of those, you know, liminal periods of life like bar mitzvahs or marriages. Constructing a kind of social world in which we are required to think with no gadget in the room, in which we are required to socialize with no gadget in the room, and that that becomes something sacred, not literally, religiously or theologically sacred, but culturally and socially sacred, and therefore is guaranteed to also continue. And therefore people who don't want to do it are looked down upon. So there's pressure to do it. And so we create a kind of society that of its own volition builds out, or religion gives societies these volitions, builds out a capacity to train routinely, systematically, consistently to preserve those mental and social capabilities of humans against this threat of AI. Wouldn't conservative religious societies do better? Are haredim going to get through this much healthier than secular progressives? Literally because they have that very thick socialization.
- [71:03]
B
So let's think about this. Religious traditions are that gymnasium that I'm thinking about. Or they could have that part. And let's think how this. Let's think how this works. I want to start with maybe. I think that there'll be a renaissance of ancient traditions. There will be a renaissance. And because this is how you preserve your humanity. Right. Because this offers you practices and rituals. Just think about the following. It forces us to meet face to face. It forces you to do that. Because the rituals don't work if you perform them alone.
- [71:49]
A
There are different religions, you and I.
- [71:52]
B
Let's speak about ours. Let's speak about ours.
- [71:54]
A
Let's speak about ours. I don't. Ours is an intellectual, textual religion.
- [71:58]
B
First of all. First of all, the idea of all religions are.
- [72:01]
A
Right. Yeah.
- [72:02]
B
Now that term. I'm saying it in Hebrew because synagogue, I don't know if it captures it. Knesset means where people. Knesset Mirshol hit. That's where people get together.
- [72:10]
A
House of gathering is the house of gathering.
- [72:13]
B
House of gathering. And what makes it sacred? It's not its location. It.
- [72:19]
A
Right.
- [72:20]
B
It's the fact that you're doing with other people. What makes it sacred is that there's a gathering of 10. That's what makes it sacred. It doesn't. It works less. The ritual is not as effective. You don't do with others. So. So tradition literally forces you to interact with other people. And synagogue as American grace, that great book, American Grace by Robert Putnam. Robert Putnam. Synagogues, churches, that's where social capital is created. That's where networks are created. It's a Synagogues, churches, that's where that creates an environment that enables for uninstrumental relationships to be created. Uninstrumental means I'm not your friend to achieve something. I'm just with you for the sake of being with you. That's its own.
- [73:12]
A
We both came to church, right?
- [73:13]
B
That's right. We're just hanging out.
- [73:15]
A
And now a car dealership where I'm buying a car.
- [73:17]
B
That's right. That's right. And we're all witnessing each other's life. I think that's the definition of a community. A group of people that are witnessing each other's life. Yuvali has this definition. A group of people that gossip about each other. You have to care a lot about other people in order for spreading information about them to be fun. So have some more elegant definitions. A group of people that are witnessing each other. And as. What's her name? Tanya Marie Lohrmann, I think from camera, from Stanford or from Princeton, she argues that it's not that communities perform rituals, it's that rituals create communities. Because if there's a ritual and we all do it, perform it together, after 40 times, we'll start gossiping about each other, we'll start caring about each other, we'll start witnessing each other's lives. And when you go to Shulem Shabbat, you know, okay, that guy has a son as his bar mitzvah, and that person has a father that's sick. And there is a system where you know about each other's life now. So is it so. And. And that is. You can say that is a gem. The fact that you're forced to read and someone has to deliver a sermon, that might be a piece of this. That might be. But. But finally, I think one of the reasons why. Oh, by the way, the idea of Shabbat itself. I had a conversation with a group of rabbis, Reform, Conservative, Orthodox, the whole thing, and ask, can we agree on one thing in the age of AI, I should do this more often, that we all practice Halachah differently. Can we agree on one thing? That Shabbat is a day where you're human, a day stripped from AI. You have to develop a thought, brainstorm with the guy next to you. You're searching for a metaphor. You have to like. Because, by the way, because chapter. By the way, chapter one, in Genesis, tribes, in the day six, human beings are created in the image of God, and then Shabbat comes in. Can a modern reading of this being that Shabbat is there to protect our Tzelem Elohim by activating it, by creating a space of time where you can't. So all these are very important ideas. And finally, I want to say, I predict. I want to make a prediction that in the age of AI, traditions will be more valuable not only because they preserve humanity, but also because you see, everything that AI could produce will lose its value. Now, that includes books. That might include podcasts. Khaviv. Sorry.
- [75:58]
A
No. Whoa. Hey.
- [76:02]
B
It will be so real. And it will be. It might. And it will be movies, by the way, if you're an actor now, if you're a producer, if you're this. Like. Like, it could be the music. My God, AI might produce amazing music. And we're living in a world where it's like, there's a law of physics. Anything that's great is where. And that's what makes it valuable. And today, great music, great movies, great books, Might become not where. So the inflation of all this stuff and all this stuff will lose its value. So what will still have value? Anything that can't be produced by AI. And you know what AI can produce? Time. Time. Think about this way. You can't ask ChatGPT. Create an hour, create music, create whatever you want and it will get better at this. I mean if, I mean y' all are skeptics saying no, can't ever be Steven Spielberg. Well maybe it could one day, but it can't ever create a month now because they can't create time. Thinks that their value comes from the fact that they exist a long time. Those things will be much more valuable in the future. So Jerusalem traditions, rituals, things that they're. I mean even the Tanakh, the Bible, like Yuval Noach Harari, once I read he said, well AI will produce the next Bible. But I think Yuval got it wrong because I mean I love you Valisa, we have a relationship. But I think he got this wrong. I think what makes the Bible a great book is not only the fact that it's a brilliant and it is and it's an amazing book, but maybe AI could produce even a bet. Maybe could produce a better Bible. But the Bibles value the fact that for over 2,500 years people are attached to it. And when you read the Bible you're joining a community of multi generational. A community that's multi generational. And that value created by time, AI can't create time, which means that traditions become something where something that can't be, that's value can't be inflated. So for all these reasons. Yeah, I'm with you. I think that paradoxically, in a world that's more futuristic, the past has much more value. We'll have like, like the equivalent of the Renaissance where the ancient past comes back to life. Maybe to preserve our humanity, to offer us things that this hyper AI world can't offer us. This is a very interesting space. This is a very interesting.
- [78:36]
A
If AI is self preservational, AI is not going to tell us to go seek other humans. And so other humans, you know, things that are long in time. Well, I'm married in 2008. I should have known that faster. That's a long time and that's a valuable, powerful thing. And your relationship with your parents and your children, those are the long time things. And that's the precious thing that you have. And what else do you have? You have a, you have a job, you have a rent or a mortgage. What else do you have. That's the thing that you. That you have. And if you don't remember that, AI will take away everything else. The economy is going to fundamentally change in ways that who knows where the money even comes from? Who knows what money even means? When we no longer spend on entertainment, we no longer spend on so many things and half the jobs are gone. I don't know. Nobody knows. Nobody exactly knows. Nobody even vaguely knows. And so the things that don't depend on those things, the fish love is always up for grabs. The real love is the only thing left.
- [79:50]
B
There's something else that religion offers. Is this too much?
- [79:53]
A
Can I go further? No, go for it.
- [79:56]
B
One more thing.
- [79:58]
A
Anybody can turn this off anytime they want. You and me can schmooze forever. Yes. And that is
- [80:05]
B
the idea of singularity means that the pace of change is so quick, we can't close our eyes and imagine the future. Now, today, it used to be like in the year the 9th century, you could close your eyes and imagine the next century. And then after the Industrial Revolution, you know, you can't imagine the next century. You could probably imagine the next decade, like 10 years than you could imagine. And then things change so quickly. Like, we can't imagine next 10 years from now, But I think many of us could think they could imagine next year, but pretty soon we can't imagine next year. Not next month, not tomorrow. That's the meaning of singularity. I think that the pace of change is so much that we lose our ability to think and imagine the future and prepare for the future. And that changes everything and it creates a lot of anxiety. But even in the age of singularity, here's one thing that I have no idea. And you know what? This is probably already true. True, if I'm honest. If I'm truly brutally honest, I know that next year I have no idea what's going to happen. I have no idea. Like post October 7th, you know that anything could happen all the time. And the world is becoming. The pace of change is so great. Like, so I have no idea how Israeli or world politics will look like. I have no idea how the economy will look like. I have no idea what new technology will. I have no idea. But I know what Sukkot will look like. I know what Pesach will look like. I know what Shavuot will look like, the Pesach. I know how the Sukkah will look like, because there's the law that tells me how to build a Sukkah. So what. What religion gives you experience of cyclical time, when time goes linear, meaning you have islands that are familiar to you and a future that's unfamiliar to you. Like, I have no idea who will be the prime minister and how the politics and anything yet in Hanukkah, but I know what Hanukkah will look like. We'll sing, we'll put the candles. I know. So it gives you an islands that are familiar in territory that's radically unfamiliar, because that's singularity. It turns the future into radically unfamiliar territory. And by the way, this is what this conversation is about. We are already in unfamiliar territory, and we're the first ones to step in this territory. And no one was ever here before us to tell of what we do here. And the idea of the gym and protecting humanity and all that. This is us trying to figure. Being the first people to figure this out, the first ones. And we'll get this wrong. And we have to trial and error. I think Israel has a role here, by the way, in creating this is, by the way, vision for the future of the Jewish state is to be the laboratory developing the knowledge of how to create a society that knows how to live with technology, that brings the best out of technology, and the technology brings the best from them. And that's why these are.
- [83:01]
A
That's everywhere. I mean, nobody feels element. I keep teaching his, you know, expectation of catastrophe, but there's a lot of idealism there.
- [83:10]
B
He wrote two books. He wrote the Expectation of Catastrophe in Judenstadt in 1896, and he writes his utopia, but his utopia is not religious Zionism and not socialist Zionism. It's about how can we develop knowledge for the world of how we live, how we create the institutions that could coexist with technology. And I think just now with AI, that's the version of Zionism that just became very interesting and very relevant. Dafka, in a way, there's a crisis of the idea of Zionism.
- [83:48]
A
We're going to leave it there. This went on much longer than either of us expected. I. We've had preliminary conversations and I listened to the podcast, to your podcast, and I still learned a lot and walk through a. I feel like a vast landscape. The world is ending, so everyone should get religion, I think is the takeaway. Right? Did I get that wrong? Micha, thank you so much. We are, in a weird way, the cutting edge of this. The front line of this is the United States and China. There are a handful of countries, a very small number that are alongside the big ones, the giant ones, major pieces of the architecture and infrastructure of AI of those larger countries. Israel is one of them. The Google R and D in Israel, the Meta R and D in Israel, they're just one after another after another. Nvidia is in Israel and Israel is, therefore, it really is a kind of shlucha. Now I'm losing my English. It's a kind of branch of this American industry. It's not, you know, but nevertheless, it is very much with them at the cutting edge. And so there's some peace to it. We're not Americans. They're the ones building this thing, inventing this thing alongside, you know, the, the advances of the Chinese, which are a little bit more mysterious to us all, but, but we're certainly part of it. So I, I don't know what the future brings, but I feel like I finally know what questions to ask. Thank you so much, Micha. I think that was awesome. I have no idea what's about to happen, and I still don't understand what the hell an AI is. But an intelligence is coming for us. We don't know if it'll be benign or not. And, and, and, and we need to be, be at least prepared to at least know who we are when it happens. Thank you so much.
- [85:40]
B
Thank you, Havi.