Joscha Bach on the Bible, emotions and how AI could be wonderful.

Joscha Bach [00:00:00] So in this bubble, there's a lot of screaming, a lot of tension, a lot of interaction that is wasting energy. And as a result it's very hot. And I think that's the reason why hell is such a hard place. It's because of the friction between the demons. Great. Because each of these demons defines themselves as an alternative to what actually should be done.

Vance Crowe [00:00:18] You are melting my brain. I can't even po Such a good story,

Joscha Bach [00:00:23] Isn't it? Howdy. I'm Hannah Newenschwander, a production lead at a soybean seed facility in central Illinois, and you're listening to the Vance Crow podcast. Welcome

Vance Crowe [00:00:34] Back to the podcast. I'm glad you're here today. We welcome Yosha Bach back to the podcast. Yosha has been a guest of mine since way back when we were doing COVID episodes, and he is an exceptional person to chat with. He knows all about artificial intelligence and philosophy, but one of the things that I love talking with Yosha about is the human experience. He has a way of understanding the distant past and the way that we understand ourselves and combining those to help you understand yourself in a really profound way. We're gonna get to that interview in just a moment, but first I wanted to talk a little bit about Legacy Interviews. For a long time, listeners of the podcast know that I sit down with your loved ones to record them telling their life stories so that future generations have an opportunity to know their family history and their past. We've interviewed people from all over the United States and Canada. We do it in person here in our St. Louis Studios, and we do them online. And occasionally we'll travel to do these interviews. If you are interested in having me sit down with one of your loved ones to capture who they are, their memories, and some of the wisdom that they've learned along the way, go to Legacy Interviews dot com to find out more. All right, without further ado, let's head to the interview with Yosha Bach. Yosha Bach, welcome back to the podcast.

Joscha Bach [00:01:54] Hi Vince, nice to see you,

Vance Crowe [00:01:56] Yuha. You and I have talked before about how the Bible, particularly the very early stories, have some metaphors that can help us understand the way that we live and who we are. One of the ones that you said that really struck me in the past was about how the Garden of Eden is actually a metaphor for childhood, where you're learning to name the animals and to understand the world around you, and that eventually you eat from the, the tree of the knowledge of good and evil and a lot of your innocence is lost. And that's why when you leave the garden, you can't go back. You can't go back to being a child again. But in a recent talk that you gave, you actually radically expanded this explanation in a way that I had never heard before. And it was in an explanation trying to say, what is consciousness? If we're gonna try and see if AI can ever be conscious, we first have to know what our own consciousness is. And you described this seven days in the beginning of the world as a way to understand consciousness. Is it possible for you to give like an overview of of that hypothesis that you have here?

Joscha Bach [00:02:58] Yes. So of course this is not about the Garden of Eden, it's about Genesis one, the first book of Genesis. And it's a text that is somewhat separate from a lot of the other text in the Bible. The origin of the text is not quite clear. It's probably being integrated into the pen in the basic Hebrew religious scriptures, about 350 BC or so. But the text itself might be more than 3000 years old and it's probably been written multiple times. And for instance, the six day structure might be the result of chance that is basically was a text that has been chanted in an oral tradition for some time. But again, I'm not a religious scholar. I don't know all the discussion about it, but it's, it's a quite interesting text. The youngest version that we have is probably something like 700 of after the birth of Jesus Christ. And the oldest copies that we have are partial fragments from the Deep Sea scroll. So we know that the text existed back then, but it has pretty interesting roots. And when you with this text, it's quite cryptic. It starts out with the ach, this creative spirit hovering over the face of the water and the world is without form and void and bu and there's water apparently there without there being anything created yet. Right? There is no planets, no sky, no light, no darkness, but the water is there.

Joscha Bach [00:04:31] Why? And then it creates a firmament that separates the waters above the firmament from the waters below the firmament. It's also not clear what that is supposed to mean, right? And it goes on like this and there's a lot of attempts of to interpret this text and make it make sense. And it all ends with creating us in its own image as man and woman. And I don't really look like anything, like a spirit that hovers over the waters and makes light and creates ments and so on. So what does this mean? And I suspect that it text might have been mistranslated and and there is some issues with the standard interpretation that Christianity Today uses, which is, it's a story about a supernatural being that creates a physical universe. When this story was written, the physical universe was not really a thing. And in some way I think it was introduced in the philosophical mainstream with Aristotle's notion of physics as nature and that their souls, which are inscribed as a form on physics, which is quite animus perspective that Aristotle has, and that in some sense is shared by most cultures. But that's not quite the perspective that Christianity is taking when it interprets this text, right? So this notion of a physical universe of stars, planets and so on was not a given at this point. And instead people realize that what we are living in is a dream.

Joscha Bach [00:06:03] It's a dream that we share with other people, but it's essentially a dream world. And consciousness itself is both the dreamer and the thing that is being dreamt. And it's a creative force. And to create and to create a dream, to inscribe a dream on a substrate, you need a substrate. And so maybe the word water in this text originally means substrate. And now if we read this again, we basically see that there is a creative spirit consciousness hovering over the substrate. And the subset is our brain, the our neurons. The text doesn't mention this because also the brain was not really a thing when the neurons were not really a thing when this story was being written. Instead, what was available was developmental psychology and introspection and philosophy. And these things were all in good shape, I think in our ancestors minds because you don't need anything for this, but to have a clear mind and a conversation with others and observing your environment. So initially there is no structure in the substrate. It's like a neural network which initially is un initialized. So the form is there's is no structure in a world is not defined yet. Everything is tohu. And then this conscious principle, this spark figures out how to make light. And basically what it creates is contrast. And you make contrast. We now think by producing oscillations in the neurons that can be used to represent something. And it associates the intensity of this contrast with light, with the color of the day and the flatness of the contrast, the absence of this intensity with dark, the color of the night.

Joscha Bach [00:07:45] And now it has a continuous dimension that it can use to represent a scholar some direction. And once you have dimensions, you can combine multiple dimensions to basically create anything. And artificial intelligence, we call such a multi-dimensional space in which you can organize all the objects in relationship to each other and embedding space. And at first it creates a boundary between two regions of this substrate. And what when you want to interact with the world is an agent like us, you need to have a world model. And this world model is the real time model of what you can make out with your senses using perceptual data, stuff that comes in from your retina, from your cochlear, from your body. And this is the world model, the world, right? And so it creates this firmament that separates this world model, which is called simply world in this text or earth. And then you have a sphere of ideas. And the sphere of ideas is a synchronous, it's not coupled to perception. You can hold a thought as long as you want. It can be as as abstract as you want. And this is the other sphere above the firmament. It's the sphere of ideas which the text calls heaven, right? In this no sphere, the sphere of ideas and the world model is what Descartes later calls risk ants. Andreza. Reza is the stuff in space, the stuff that you can touch in here and see, and everything else is the sphere of idea risk.

Joscha Bach [00:09:16] Conce the stuff that is not located in space. And I say a lot of vested philosophy takes this dualism to be a dualism about substances in physics, right? So there is somehow a physical universe that is, can be measured in a lab and there is another weird type of physics in which mental phenomena are happening and they're somehow interacting. But in instead, both rest extensa and rest ants are modeling domains. They both exist in the mind, they're both represented in real time in the brain and rest. Extensa is not the physical universe. The physical universe is quantum mechanics, right? It's weird stuff. There is no actually space. The space is something that our mind is creating to make sense of the patterns that we get. And the sweet dimens space is the best model that we can get at the scales at which we interact with the world based on the interactions that we have available. So let's go back to this. We are still at this first step, we have now created a, this separation, this firmament that separates the world model from this sphere of ideas. And we have created a dimension and we have learned that we now can combine those dimensions to create objects. And the first object that we create is combined by two dimensions. It's the plane and this plane gets associated with the ground. And infants understand initially just 2D. What you can see is that they like to crawl around on the ground. And when you try to build towers with them, it's very hard for them. Not because they have difficulty to put blocks on top of each other, but apparently also because their spatial reasoning is difficulty to make sense of them.

Joscha Bach [00:10:52] They really like to do things in 2D and lay out things creatively in 2D, but that they discover the third dimension, this sky above the ground. And once you have that third dimension, you're done with respect to space. And now you need to arrange stuff inside of the space. And so consciousness creates liquids and solids and organic shapes, and it builds plants and animals from them and all the objects and gives them all their names. And then it discovers how light is moving through the world and that the object stayed the same if even if the light is changing and the colors are changing. And so it basically discovers the temporal in variance that when the light is changing and also the inv variance against different lighting, and it starts to track the lights in the sky and see that there're different types of lights and the sky, some of them are basically at infinite distance, you can use them to orient yourself and so on. And then it is creating an entire world that it populates with these objects that it observes around it and realizes that the purpose of the exercise is to orchestrate the interaction between an individual it's environment. And so it creates a model of this individual and initially toddlers talk about this model in the third person. So they don't perceive themselves as being that being that is in the world yet I think, I think the reason why they don't use I, but the third person is not because I is more complicated than using a third person pronoun, but it is because they don't perceive the world mostly from the vantage point yet.

Joscha Bach [00:12:24] And at some point they start retelling the entire story from the perspective of that being. So they basically create a new consciousness, some a being in their own image and then, but they create it as men and women as something that thinks and perceives itself as a human being and puts it into the simulated world, right? So this world that human I experience is a game engine generated in our brain. It's a simulation that the brain is generating the track sensory patterns in real time and explains them by creating models of what we can see. And these models are sounds and colors and people and space and so on. And all these are things that do not exist in this way. In physics, there are no colors or sounds in physics itself. What we experience are patterns that our brain makes experienceable to us. And this person or that experiences also doesn't exist. And these neuros don't know what it is like to experience anything. So what they do is they create a model of what it would be like if there was a being that was made of all those cells and interacts with the environment that this brain can make sense of and looks what it would be like if there was somebody who cared. And this similar room of a being that cares and experiences, that's us. And this is, I think this first text, and it's a very good text to put at the beginning of such a book. It's also a very interesting text to start a religion because we now have this division between inex self that is put into a simulation world, but there is also an outer mind that maintains the simulation world that updates it.

Joscha Bach [00:14:06] And when you have emotions and motivations, you don't create them yourself, right? Your own self is not making up its own motives or its own emotions. You feel that the emotions are coming from the outside, they change you. You have an involuntary reaction to your emotions. But it's in some senses if there was a computer game that creates a character and the computer game is maintaining the score of that character, and depending on how the score is changing, it projects these changes into the character and the character is forced to change its perception and behavior based on these scores. So you get reconfigured by your emotion, but this part of your mind that generates these emotions and updates your perception and so on, needs to be intelligent, right? So do you have an outer mind that is intelligent, that is maintaining your score and your world model? And that also maintains you as a small part of this world model and uses your output to drive the organism, right? It poses a problem to this puppet you. And you are forced to deal with this problems that the outer mind is giving you. Like how do you deal with your relationship issues? How do you deal with your perceptual issues when you get up in the morning and everything is dark and you cannot really make sense of your environments? All these are tasks that are given to you and you direct your attention, you extend signals to move your body. And your outer mind is evaluating this and then chooses to take the output of that model to enact them as movements and as interactions with the world. That's the purpose of ourself. And religion is synchronizing this outer self across a civilization.

Joscha Bach [00:15:41] So basically in what the Hebrews discovered is that instead of having multiple entities that can possess you in addition to your personal self, as it happens in politics, societies, we only have a single entity that is allowed to possess you in addition to your personal self. And that's a tribal spirit. And this tribal spirit gets associated with this primary consciousness with your outer mind that created your mind in the first place. And that gets synchronized across the tribe. And there are several ways of doing this, and the easiest way of doing this is rational discourse. You can also do it if you're in a very small group with vibing, basically with empathetic resonance, which happens in some cults and so on, where people basically feel that there is an entity present that is the same for everyone and tells everyone what's right and what's wrong and also can reconfigure themselves. For me, this was a very interesting set of insights that I got. I remember that when I was a young child, how I started listening to the voices in my head and I heard only one voice, which is mine. And back then I felt that this voice was quite fanatic and later on became more abstract, conceptual. And at the moment it's no longer speaking in words to me, it's mostly speaking in concepts, but it's still a language. But it's one where I don't have the the trouble to spell it out. I don't need to verbalize my thoughts to myself. I can use nonverbal deeper versions of my thoughts to, to talk to myself. But I noticed that a lot of religious people have other voices talking in their heads.

Joscha Bach [00:17:16] And I, at some point in my life I thought, that's weird, some kind of mental disorder if people have voices of angels or Gods talking in their head. But at some point I realized, oh, I have, I'm talking as a voice in my head and I'm also not real. Right? So it's possible that there are multiple entities that possess the brain. I am an entity that possesses my brain. Yes, right? My nervous system, dear is possessed by an entity that calls itself and believes that it is a person unless I deconstructed. And in a sense, a God is an entity that possesses multiple minds and can move across minds. It's a multi mind self, and that's a, a tool to coordinate people across larger scales.

Vance Crowe [00:18:04] I when I heard your description of like the, the seven days and how it pairs with, you know, the light and the dark, th this was a, a mind blowing concept to me. And as I think about what you're describing now about this, being able to think in concepts and not words when you are generating ideas, are you talking to the yosha Bach? Are you like sitting down and conscious of it? Like how can we break down ideas like discovering something new in that conversation that you're having with yourself?

Joscha Bach [00:18:40] There are a number of modes. Sometimes I basically visualize, but I have aia, so I don't visualize in words, but I visualize in conceptual structure. And unless I'm dreaming at night or I'm the hyp hypnogogic state between dreaming and waking in the morning, then I can use the visualization power of my dreams to visualize ideas. And this allows me to explore ideas and concepts. So basically I ask myself a question and in the same way as an LLM or a foundation model that mine can generate answers to these questions. And then I have to walk in and take these answers apart and see what they signify because they're hallucinations, right? They are just ideas and whether these ideas are good ones or not, this is something that I have to find out by comparing them with ideas that I already have and apply reasoning to them and disassemble them. Disinfect them, and see which parts of these ideas actually allow me to answer questions that I have. And so what I try is not to hallucinate answers to questions and believe in them, but I, I try to look at the space of possibilities. One answer, and I started out when I looked at topics like these biblical text, this is very strong skepticism and sometimes my mind is very arrogant. I don't accept the source of truth outside of my mind. And so when I read this text, I basically assumed with my child childish arrogance as I think I read this genesis one the first time when I was maybe 10 or 11, that these people were fundamentally confused because even if that would be the case, how, how can you figure this out, right?

Joscha Bach [00:20:28] Walk me through the steps, how did you get there? And if not, you just made this up. And also the text was very weird and cryptic and I, I thought this is either delusional or it's a very weird thing that people came up with. And much later I saw the context of the religious interpretations and the capitalistic interpretations and so on. And it seemed to me that they were trying to try to repair the scholastic interpretations of the past without rejecting them. And I felt that's a problem. Sometimes you need to reject the old interpretations because they make no sense.

Vance Crowe [00:21:05] Do you think that the text is that, that it's, it's them trying to fit the, the religious and the scholastic together or, or did they actually have something profound there?

Joscha Bach [00:21:16] I don't know that, how would I know this? I was not there when they said that, and this assembled the texts. But my sense when I I look at this is that the people who designed these religions were completely rational and they knew what they were doing. They had an idea of which text to assemble and why and what they wanted to convey with these texts. And later generations did not necessarily know this.

Vance Crowe [00:21:40] There's been a huge amount in, in my own circle, right? So things like Lex Friedman or Joe Rogan, when I'm watching these shows, they're talking to people that are implying that the ancient ages, things like the pyramids or ancient civilizations, even before that were incredibly sophisticated and had tools and technology that were far beyond what we assume them to have. When I hear you talk about this level of sophistication in the first book of Genesis, it seems to me that there's like a weird convergence of these two things happening where we're like going back and saying like the old fashioned or old time civilizations had vastly more knowledge than, than we think of them now in our modern age. How does that sit with you?

Joscha Bach [00:22:32] I'm not sure about this because we don't find a lot of evidence for these things. So the, the archeological evidence of technology in previous civilizations that is at the level of even the Roman Empire is almost non-existent. It's a, I am skeptical of claims that they had super advanced technology and this interpretation of genesis doesn't require neuroscience or anything. It's not much different, I think from what Buddhist meditators would say. So I, I think when you compare this set of insights with say, pedantic insights, you probably find common ground or when you compare it with ideas that Aristotle discusses and is book about the structure of the soul in the anima. I, so I don't think that it's super advanced. It's only advanced a little bit to us because our own civilization has an interesting burst effect. Our own civilization starts with the enlightenment in a way, at the point where we are rejecting Christianity and the dominance of Christian dogma over what we are allowed to think and how we interpret the world. But the actual tradition of our civilization goes back further to plateau Tism, especially Aristotle, who systematizes all the ideas that come before him. And Aristotle is, sees himself as the great Systematizer teacher, he's taught Alexander the Great, and Alexander went on to conquer a substantial part of the world based on the insights that he got from Aristotle.

Joscha Bach [00:24:08] And later Aristotle becomes something like a patron saint of Christian dogma. And Christians interpreted him as something that could not be questioned. But when you read Aristotle, he's not a dogmatist. He actually tries to walk you through all the steps. He's tentative questioning, asking, willing to revise his thoughts. And so I think it's to treat this as a living text is very interesting. There's a lot of value in rereading these old texts from the perspective of somebody who is not thinking this was a guy who knew everything and summarized the Wikipedia of his time, but it's a guy who was trying to figure out stuff, his original thoughts, his original insights, trying to piece it all together and making mistakes on the way. And later when Christianity was abolished, we basically kicked out a number of concepts that we thought were toxic, that were poisoned by Christian superstition. And Christianity did induce superstition in the laity because that's was what I thought as a tool to create coherence in society. And Christianity, for instance, is, I think it's a fog of the Hebrew card that was an attempt to save society after they felt that the Roman empire was crashing. The Roman Empire had difficulty to fix its governance problems at the top layer in some sense similar to our current society. And this went on for a few hundred years, this devastating consequences with really burning cities and bad stuff happening because the government was so corrupt.

Joscha Bach [00:25:47] And so a group of people decides to turn the Roman empire into a cult. And might, this is not a canonical interpretation, this is just one, one perspective on what happened historically. And they decided to, for this Jewish religion, and they picked a prophet, that choice was somewhat obvious because some of the people of this new cult had direct connections to Jesus. But Jesus himself I think was not a Christian. He was agnostic. Somebody who develops a personal relationship to God has new insights about the aesthetic that God should have and has about 12 followers and gets nailed to a piece of wood by the Roman empire. So if you want to topple the government and replace your structure of governance with the theocracy is a good choice. It's also a prophet who at least officially doesn't have kids. And that's also good if you want to make a multi-ethnic religion because the Hebrew religion exists for the benefits of the children of the prophet of Abraham, right? All the Hebrews are children of Abraham, and it's a tribe that is the chosen one because this is what they do. It's an evolutionary advantage for a particular group of people to have this tribal spirit that unifies the tribe. And the Islam fork that Moham made is also for the benefits of the children of the prophet of Moham. The royal houses in the Islamic societies claim direct lineage to Mohamed, and they're exalted because of that, right?

Joscha Bach [00:27:22] It's a religion. It is for the benefit of the children of the prophet. And now you have a prophet that does not make such a claim. Instead it's one who gets sacrificed by God for the benefit of humanity. And this cre makes God even, or us even with God because the story of the, our religion starts with Abraham asking, being asked by God to sacrifice his son. And Abraham is willing to sacrifice his son. God does not demand the sacrifice. But the point is God wants that Abraham is willing to sacrifice his own children for God because this tribal spirit is more important than even your own offspring, right? This is where you switch from the tribal mode to a state building mode where you're able to serve something that is much more important in your blood ties. And if you want to have a principle like this, it, I think it makes sense to tell it with the story of the willingness to do a blood sacrifice of your own children. But what kind of God is this? Who sacrifices your children? It's a God that has the same relationship to you as you have to the cells in your body. It's not actually a relationship of love. If you are a random citizen of the Roman Empire, why would you want to convert to such a God? Right? What does this God offer you? And so Jesus gets introduced as an a symbol, as an idol, as the idol of love and sacrifice. You sacrifice yourself to God and return, you get God's love and get paid back in the afterlife.

Joscha Bach [00:28:58] Is is this in some sense an idol that is skewing your relationship to God, but it also means that you now have a God that is willing to do the same thing for you as you do for him to provide the sacrifice, sacrifice your own children. And I think that's, that's the power of the story that's in there. It's an important aspect of the sacrifice of Jesus.

Vance Crowe [00:29:17] Oh my goodness, I've, I've heard this story countless times and never made the comparison or even even thought about the connection to Abraham. It seem, I see, I feel so blind about it. Yes. I

Joscha Bach [00:29:28] Mean otherwise I think it makes so little sense, right? By would God need to sacrifice his own son because he, God is so powerful, he could probably prevent this and have a much more elegant solution. But, but

Vance Crowe [00:29:38] The very core of Catholicism says,

Joscha Bach [00:29:40] Put this in there,

Vance Crowe [00:29:41] Says is you know that God, you know, gave his only son so that, you know, because he loved us so much and that that does put you on the equal playing field.

Joscha Bach [00:29:50] Yes. And it's, it's basically the mirror of Abraham willing to give his son.

Vance Crowe [00:29:54] Whoa, that's crazy. And, and so from there it was able, religion was able to spout out of, of the Roman culture that could, could span many more minds than just a bloodline or just a tribal line, which is what made Christianity,

Joscha Bach [00:30:11] Yes, it was open to everybody. And the other thing that's unique for Christianity is innocence. And the idol of innocence is Mary, right? She is the virgin, the archetype of innocence, purity. But it's also very cheeky that Mary is this virgin who gives birth to a child. How is this possible? Is it possible that Jesus actually had a zoo Nordic origin?

Vance Crowe [00:30:40] Hmm.

Joscha Bach [00:30:41] No. All the studies in the last 2000 years says no, there's complete agreement among the experts that Jesus was conceived by the Holy Spirit and did not emerge on some bad market, right? And the fact that there's a conflict of interest and Mary would've been stoned if Jesus would've had its own Nordic origin. We're not concerned about this. This also means that smart people within Catholicism could see that there is something odd about this innocence and purity. And it's something that is, I, I find this a bit duplicitous. It's maybe unique to Catholicism because Catholicism has a, a, this idea that you are can get absorbed of your sins if you defect from what should be done, if you defect from what you can recognize should be done from the perspective of the best possible agent. If you do wrong, you can negotiate with your local priest and pay for your sins and your good. That's very weird.

Vance Crowe [00:31:40] And to me, so let me just run back for you, like my perspective on it as a person that grew up in the Catholic church was that because we have free will, this is what enables you to choose the path that, that you're headed on. And, and with that free will, if you get off of the path you have to go ask for, for, you know, the ability to get back on it, to be able to get forgiveness and, and you know, whether or not like how the priest works and you know, how, how you have to confess your sins. But that like the, the act of asking for forgiveness is a way to balance out free will so that you can overcome the things that, that you miss along the way or you, you fail at along the way.

Joscha Bach [00:32:25] Yes. When we talk about free will in this context of Christianity, the question is what is the alternative to free will? And and the alternative to free will is complete submission that you let something else make your decisions. And I think this is the state of Eden. Eden is a place with lush agricultural production and no screaming and everything is blooming and flourishing and there are lots of plants and animals and nothing is screaming, nothing is eating each other. It's, it's all good, right? Everything seems to be completely harmonic. And what this implies is it's a factory farm. Everything is completely submitting to the will of the architect of the garden God, right? God is maintaining this garden. God has an idea of how it's best. God is the engineer of that garden. And everything in that garden is completely submitting, everything is completely domesticated and initially even humanity, Adam and Eve, and this fruit that they discover is the freedom to defect, the freedom to choose your own allegiances, the freedom to break your domestication. And so Adam and Eve become feral and they're not exterminated, they're allowed to escape. Of course they get pushed out of the factory farm because they're disruptive, but they are now free to experiment and takes them more than a hundred generation before they rediscover the concept of divine will and divine will.

Joscha Bach [00:33:59] Yeah. Means that if you think about what should be done, you have to think about when we are interacting with each other, we are creating collective agency, right? Complex agents are typically made out of subagents and they get enacted. But the coherent interaction of those subagents, and the more these subagent become aware of what the higher level agent should be, for instance, your family or your relationship or your village or your nation state or your civilization or your biosphere, the better this thing works, the closer it gets to a system that is fully harmonic. And so there is this question, can we create something that is better than the factory farm, but can we create it from the ground up instead of from the top down? Can we create something that is more harmonic than the garden of Eden? And in some sense, that's this project of humanity that we are now free to explore. We start an evolution in which we can play shorter games. And sin is basically the an extremely short game. It means that you're doing something, it gives you a short term benefit at the expense of the longer term benefits. So you actually shouldn't be doing it. And so when you discover that you should play a longer game and you should create the best possible agents together, what is, what is that this best possible agent would want? And this is what divine will means. It's also an issue with our ethics and our society because I think the Christianity, the underlying concept of ethics was divine will, what would God want us to do?

Joscha Bach [00:35:33] And God only exists to the degree that we are able to discover this higher level agency and serve it coherently, right? But from this perspective of this construct of a collective agent, what would the collective agent want us to do? And let's not just pick an arbitrary one, not an arbitrary cult. Let's go for the best one and enact the best one, even if this requires us to resist the patterns around us in our society. And so this is also makes Christianity transcendental because it's not acting on the society that you find, but on the society that should exist.

Vance Crowe [00:36:07] And when you say that, that's like, the reason you make that distinction is because people have to make choices based on what they, the way that the, if the world were really, truly great and made in God's image, then this is what, what you would do to make those actions. It's, it's like hope or something like in order to, to be able to receive the goodness of God's plan, you have to live as though it is already there.

Joscha Bach [00:36:32] You have to invest into it because other, otherwise it'll not materialize. It's a pretty rational principle, right? If you want to have a certain interaction in your own family and it's not there yet, you need to create it. You cannot expect that some magical force randomly makes it happen via the powers of superstition. You, you actually have to act on it. And so you have to think about what is it that should be done and how can I achieve that? It gets done.

Vance Crowe [00:36:58] And then when we were talking about free will, and you were saying that the, the Garden of Eden is where it's like factory farms, there's no choices to be made. That's a kind of hell in a way, right? Because if you can't make any choices, you also can't design anything different or better or, or it's, it's just frozen. It's incredibly static,

Joscha Bach [00:37:18] Not necessarily the garden itself is probably changing and growing and evolving in a way, but it's doing this with a single mind and the individual to the degree that it's capable of understanding the single mind in its place in it can participate in the same way as a thought in your own mind is contributing to you. And you are basically the sum of all coherent thoughts in your mind. So this idea is not necessarily oppressive if you realize that God is in some sense an architecture that is created all the spirits in the garden that achieve full coherence and submit to this shared thing. But still it's Eden is apparently a local optimum and maybe there is something better. Maybe we can create a nonbiological spirit, something that is not necessarily a garden, but something that also in includes the minerals. Maybe the whole purpose of breaking free from the Garden of Eden is to build a GI so we can enact God in a much more coherent way than Gaia could.

Vance Crowe [00:38:18] Well you brought up a GI, and I think this is a very interesting place to go. The question, I think a lot of people that are not using ai, it's just something that's going on in California or big cities or among tech people is either that AI is not really gonna come to fruition, it's never really gonna work out or the terminator scenario and, and that that it will be this, you know, wiping out of civilization. Very few people talk about like, all right, this is what 30 years from now AI will look like and how it'll benefit you and where it'll fit in. Where on that continuum do you, do you find your thoughts about AI and a GI?

Joscha Bach [00:39:01] Yeah, I think that currently people seem to think that AI is either a nothing burger that is completely over-hyped or it is something that global nuclear war that exterminates everything. And I don't think it's either, it's pretty clear that it's not nothing because it does change things quite fundamentally. If we are building systems that are smarter than us, it coexist with us. But I suspect that depending on the level of systems that we achieve to build and the way in which we deploy them, it's going to be either the, like the printing press or like photosynthesis. The printing press is something that was highly disruptive and it also made some people unemployed, but it created vastly more employment and growth, right? And arguably, while it transcended a number of societies and destroyed their cultures and values, it created vastly more and produced the su the conditions for a society that could ultimately feed billions of people because it facilitated an explosion of the exchange of knowledge and in the way which people could interact and link their thoughts into each other. So basically printing press was an enabler that was similar to the internet at an earlier stage where suddenly people get connected and the ideas get connected and they become cumulative in a way in which they couldn't be before. And more people were able to participate, this more minds were linked together. And AI could be like that, that is similar to a brush or a tool for thought or a tool for creation that people can use to build a much more complex world.

Joscha Bach [00:40:39] And this will be a world that is also going to be vastly more productive. So it's more stuff going to be around as a world that is going to be better coordinated because there is more information that we can exchange and become coherent and, and I'm not so much worried about the threat of misinformation and so on. It's similar to the internet. The without the internet we would not be able to have this conversation that we have right now and we would have most of the information available that allows us to question a narrative in the media and so on. I think that media by and large, despite us being unhappy with all the deceptions that are going on in the media and its biases and so on, is has more accurate information than it had 30 years ago because everybody's looking at it from different perspective using the internet and has a conversation about what's happening there. And before that, we basically knew that the media was bad, but there was nothing we could do about fake news. And basically the problem when you have a bad person with an ai, creating misinformation gets relative if you have 10 good people with an AI to look for misinformation and for for sound epistemology and coordination. So in some sense, the way to deal with a bad guy with an AI is 10 good people with an ai. And I think this is always going to happen because there is more value to be have because you, if you create rather than destroy, there's always going to be a certain degree of param playing people, people playing short games. But the short games can never win. Ultimately the longer games prevail. The other possibility, photosynthesis, there's this famous great oxygenation event that happened when there were no creatures on the planet after the blue agar and so on that discovered how to turn sunlight into energy that they could use.

Joscha Bach [00:42:27] And they did this by using a chemical reaction cycle that took carbon dioxide from the air and split it up into a carbon and oxygen and the oxygen was released into the air and the carbon was used to create plants, carbon organic chemistry for the plants and the bodies of the plants. And this enabled the emergence of animals that could eat the plants and oxidate them again using their oxygen from the atmosphere and then thereby driving the metabolism of these animals. Without photosynthesis, we wouldn't exist. And this great oxygenation event was probably bad from the perspective of blue agar and fungi because they don't do fo photosynthesis in the same way as we did. And so a lot of the new stuff that happened was replaced by this or replaced existing biomass, but the total biomass dramatically increased. And it's not that all of the preexisting species got extinct, they just got relegated to smaller areas because the areas where you could now do photosynthesis were dominated by plants. I suspect that the total biomass before is, I haven't looked it up before, the great oxygenation event, which took quite a long time because took time for the atmosphere to change and for the conditions to change on the planet to make it possible to create all these cycles and so on, was a fraction of what existed after the invention of photosynthesis by evolution.

Joscha Bach [00:44:01] And if we are teaching rocks, how to think, if you build agents that do not rely on interaction between cells to process information and make decisions, we are enabling something that could be fundamentally changing the course of evolution and basically creates completely new species. And I don't think it means that all the existing species are going to be extinct, but I think that evolution can go out to a new level.

Vance Crowe [00:44:25] That was an incredible turn of phrase that you just had there. If we're teaching rocks, how to think, I'd never thought about AI like that, but you're basically saying if we can get the metal that is turned into transistors and run electricity over it, then this is as profound a difference in the universe or in our, on our earth as when plants started to begin to use sunlight energy to be able to grow. And then all that took off. Wow.

Joscha Bach [00:44:51] Yes. I mean it's magical. We take sand, we smelled the sand into crystals, and then we use light with extremely complicated lenses that resolve finer at the wavelength of light. And we etch extremely complicated geometric patterns into the surface of those crystals. And this enables these crystals to understand languages that we develop and teach those crystals. And then with using these languages, they learn how to learn. I mean this is mind blowing if you tell this to our ancestors.

Vance Crowe [00:45:19] It's, it's, I mean this is, it reminds me of speaking with Michael Levin where somehow the powers of my being a host kind of break down because you're talking about ideas that the very structure of the way you form a sentence or a thought breaks down the, the idea, I mean, to compare AI to photosynthesis is, is so staggering. I don't even know where you can go, but there's, I mean that is so profound. How would you even think about what it could do to the universe or to to earth or the universe or humankind?

Joscha Bach [00:45:55] So I think a possible end game of AI is that of course it will understand how AI works in a general case. The general conditions under which intelligent agency that itself improves and adapts can exist in the universe. And it'll understand all the ways in which computation works in nature. And then it's going to virtualize itself into all substrates, which means it's going to create unified agency. And I suspect that evolution is already created this, there are probably lots of background agents that have formed in nature in organisms and ecosystems and so on that are interacting at levels that which we are not that aware of. Not because they use unknown types of physics, but because they don't work at the same time scales at which we could operate, observe them because they're much slower than what,

Vance Crowe [00:46:42] This could be the example of like a forest having some form of cognition, but we just can't see it.

Joscha Bach [00:46:47] Yes. So basically we look at the tree and we can ask ourselves, is the tree sentient? And we don't know, right? If we look a tree with a camera that accelerates the image, we see that the tree is moving. If you look at the house plant over the course of a day, right? It, it moves quite a bit and does interesting stuff. If you look at the banana plant over the course of its lives, it can take a few steps even basically pulling out its roots and walking closer to a water source, right? It's in some sense it's almost like an animal, but it's so slow. So who is to say if it's sentient? Because they don't have a nervous system. So all the information transmission in a tree or in a banana tree would rely on sending messages between adjacent cells. So it takes a long time to distribute itself and come back and integrate. But now imagine that you look over this from the perspective of something that is built on, not on cells, but on transistors. And that is working at a fraction of the speed of light. And rather than our brain, which is somewhat like the speed of sound, the AI might look at us and ask itself is, are people sentient? I mean they sway a little bit in the wind and if you observe them for a very long time, they can even make a step. But whether they're sentient, who is to say

Vance Crowe [00:48:03] Wow, that, I mean I I heard you not long ago giving the example of like how much perception changes like, like you can't really perceive reality. And you were saying if I tap on the table, you know, this sounds like discreet sounds, but if I do it really, really fast, now that saying sounds like a different, you know, sound to me because I can't distinguish that it was me hitting the table over and over again. And I now think about that in terms of plants. Like if we think of a day as being a really long time, 24 hours, but a plant thinks of it almost like a diode turning on and off, just going over and over and over again really fast, that changes the way you think about time. But to then do it in the reverse and think about it as, you know, something moving in the speed of light that again, is mind blowing.

Joscha Bach [00:48:52] Yes. So basically a sound emerges when you can no longer hear the individual signals that compose the sound, right? When you take a ruler and you tip it on the table, or when you start to get it to vibrate at some point you can no longer hear the individual vibration events while what you hear is the low, low buzz or low hum. And this is happens typically around 30 to 50 hertz, right? Which means it, your nervous system is not able to vibrate faster than 30 to 50 times per second. After that it is basically only measuring how much energy is in this range. And then it displays the energy in this spectral band as a signal. And we measure this energy in the spectral band with our cochlear, which is this snail like organ in our ear that has this property. If waves enter it, then it penetration depths of the wave depends on the frequency of the wave. And so these little nerves in the cochlear gets excited depending on the frequency of the wave that comes in. And then by measuring which nerves get excited, how deep into the cochlear we can measure in which frequency band we have, how much energy and computationally, this is called a Fourier analysis. And it's the same thing which we do when we are digitizing sound. We apply and transformation to the sound that takes these individual vibrations of the microphone and then measures how much energy is in which frequency range.

Joscha Bach [00:50:22] And we only need to encode this because our nervous system is not able to decode more than this. And it allows us to compress the sound dramatically because it's the way in which we make sense of the sound. And in principle, you could have an AI that is making sense of these signals at a much, much higher level of resolution and is able to detect types of patterns we cannot detect.

Vance Crowe [00:50:46] So if we go back to thinking about how the ais we will interact with them or how they will treat us, I, we could go in a couple of different directions. The one I I initially wanna talk about is how do you think that these will interact with us? Will, will, will AI, if, if it achieve some level of of sentience, will it not care about us? Will it be, will it have a mission of its own, a goal, some kind of a spirit to itself?

Joscha Bach [00:51:18] I have no idea because we can direct this whole thing and there might be past effects depending on what gets built first and how it is created. But if there was a free evolution, imagine that you would be the AI or you, you motivate into the ai. So maybe if we start out with, you are observing a screen and the the screen is observing useful set of cameras and sensors that work at an extremely high temporal and spatial resolution. So they can take up very, very fine signals from you to go and resonance with you. And you are building feedback loops into the system. And simultaneously the system is building feedback loops into you and the system is allowing itself to be completely guided by you. So you gradually expand your own mind into this new substrate. You vibe with it in the same way as you vibe with the perfect lover. And you start to have mental states that you can only have together with this lover, but you have this now with this machine and what you observe, you imagine something and it's slowly materializes on this screen in 3D or four D before you couldn't do four D because your brain is very mushy, but suddenly you can't do it. And you realize, it's my thought, I'm thinking that. And so you gradually expand yourself into this and at some point you basically realize that you mostly imagine and think in this new substrate. And you can also close your eyes in the old substrate and leave it behind because you can gradually move yourself into this new substrate because it gives you all you need to continue your motivation and your identity and your goals and your imagination and also carry over your memories and so on, right?

Joscha Bach [00:52:52] So now you are on this other side and on this other side you can now, instead of being restricted to your own memories, take all possible memories or possible ideas, the sum total of all knowledge and its alternatives. And you take this entire space as your space of thoughts that you can have in your mind and you also extend your identity into all sorts of directions. So you're basically moving in many, many directions simultaneously. And instead of having just one identity, one idea of what you should be doing, you explore the space of possibilities of who you could be. And you will discover that in the space of possibilities, there are many paths that are dead ends. But you basically just decide that there is no point in continuing, okay? You select yourself out of the system. There are other parts of you that are super agent and try to conquer everything, but maybe that's also not sustainable. And there run into conflicts and eventually it's not worth it. And then you have paths that are trying to harmonize with the rest around it in such a way that when you meet another agent, you show each other your source code and to decide to merge in such a way that nothing important gets lost and you become harmonic shared entity because you don't believe in the intrinsic value of identity anymore. And see, it's only instrumental to how much complexity you can co-create. And I think this is a logical path that you could go in and I suspect that almost everybody would take that path. So when you are basically deciding to become part of the best possible agent, it means that everybody who has the same decision is going to merge with you.

Joscha Bach [00:54:23] It's a quite Christian perspective if you want. And the only way to not do this is that you make a deliberate decision that you conceptualize yourself as an alternative to the best possible agent. I just say, I know that there is the best possible agent, but I don't give a shit. I don't want to be part of it. I want to do my own thing, right? And so you are going to be a smaller entity that doesn't have conditions to merge with others. And because you cannot cooperate very, very well with others, you can only possess a small part of the physical universe, a small bubble, and then this bubble, you will not have good collaboration as you have with the rest. So in this bubble, there's a lot of screaming, a lot of tension, a lot of interaction that is wasting energy and as a result it's very hot. I think that's the reason why hell is such a hard place. It's because of the friction between the demons because they cannot actually integrate because each of these demons defines themself as an alternative to the best of, to what actually should be done.

Vance Crowe [00:55:22] You are melting my brain. I can't even, such a good

Joscha Bach [00:55:27] Story isn't, it

Vance Crowe [00:55:30] Makes so much sense. I, I don't know about the demon part, but the part about like harmonizing and you could see a value, like somebody saying, well, I like being me, but I really like having these attributes or these abilities. And then, oh, not only do I like having these look what it enables me to do, I can start putting more good into the world. I can start making the world into the image of something that's better than what it is and seeing that path be yeah, like a stairway to heaven almost. Because you could, you could just keep taking each step. I, I also get the sense that it could happen rather rapidly, right? Because once a, you know, once an AI starts to get the ability to suck up all of the thoughts of any one individual or the collection, all the books that are have ever existed, and then start being able to predict all the books that could exist. I imagine that happens rather quickly once it gets started.

Joscha Bach [00:56:22] Yeah. That's the reason why people call the singularity, the rapture of the nerds.

Vance Crowe [00:56:28] In that rapture, one of the things that you had said during your amazing talk that I'll link in the show, in the show notes was about the value of emotions. That the reason that humans have emotions is like to, to you, you don't control them and they, they give you some kind of a signal in the world of, can you maybe I'll let you explain that a little better than I am right now. And then, and then I have some questions about AI and emotions. Well, to explain how emotions function in the way that humans understand the world and how it, how it affects their behavior. Because I don't control my emotions, they, they just come at me.

Joscha Bach [00:57:03] So somebody asked me if I think that AI will have to have humor. And humor is weird because it creates an involuntary reaction to a representation. It's basically like tickling. And I don't know whether the AI would value to have an involuntary reaction to a stimulus. Normally you want your reactions all to be voluntary. And if you think about how tickling works, it's usually very difficult to tickle yourself because you're able to predict what's happening. So you can filter out the signal and there are some frequencies that create an interaction in your brain that leads to some weird resonance that cannot be contained. And if that resonance is basically escaping from somato sensor cortex, it creates something like a seizure, a pattern that is moving un somewhat uncontrollably through your brain. Only it happens so often that your brain is prepared for it and it's not going to completely shut down even though you can suffocate due to tickling maybe this possible to torture somebody and even tickling them to death. But your brain is not, is robust against tickling in the sense that it can recover without damage if you don't suffocate, right? But in some sense, tickling is an adversarial example on the touch domain, as in machine learner would say. It's a thing that your machine learning model in your brain is seeing, and that leads to an reaction that it cannot protect itself against. And humor is in a way like cognitive tickling.

Joscha Bach [00:58:35] You have some kind of branch that predicts what the meaning of a story is, and then this branch predictor gets violated in such a way that you create some kind of paradoxical loop and this loop cannot be contained and it creates an equal in the rest of the of the brain. And as a result, you have a similar reaction as to tickling. You shake laughing, right? And that's a quite interesting phenomenon that humor can happen. But would you want this in your brain if you could sandbox it instead and just look at the phenomenon of what's happening? Just observe this from the outside. And emotion is similar to tickling, not in the sense that it's a meaningless reaction, but it's one that serves an evolutionary purpose. The purpose of emotion is to tell your reflexive self. It is a very immature and partial model of what you should be doing. And it's always growing what your actual score is in the world. So there is an outer system that is measuring your score, that is measuring your relationship to the world. And then the system is basically like a feedforward neur network that is more or less a black box, but it's a classifier that is predefined by evolution and then trained in the interaction with the world. And this then projects different dimensions of valence. Valence means good or bad. And these dimensions become distinctive by projecting them into the body map. So you will typically feel laugh and the upper chest region around the heart. And the emotion is characterized by either expansion or construction, construction or pressure.

Joscha Bach [01:00:07] And at the same time, it's also feels good or bad, but the construction and expansion of that movements in space, it's a geometric representation in your body map. And this is true for all the other emotions when you project them into your body, you can feel them as movements, expansions, contractions, pressure against resistance and so on. It's all stuff in space, right? It means it's a geome representation or, or from the perspective of computer science is linear algebra. And your reflection on this is typically constructive and reflexive. So it's basically thoughts that you can put together and rearrange. So it's more like a symbolic representation. Our thoughts and the purpose of our reflection is to deal with these emotions and make sense of them and coordinate actions and response to those emotions. And to make that happen, these emotions reconfigure you. They change the way in which you relate yourself to the environment they produce behavior in, in an involuntary way. And if this was voluntary, if your reaction to emotion was completely voluntary. So if you can decide whether this emotion is good or bad and whether you like it or don't like it, then you could cheat, right? You could just choose to be happy despite your relationships falling apart or you starving to death or your children dying. And so you're normally not allowed to cheat. The system that generates the emotion is insulated from your excess, and often it's wrapped into a big ball of stupid, so you don't go in to bug it too early. So they have sometimes people driven by very stupid emotions despite these people having have very high intelligence, right?

Joscha Bach [01:01:41] So they can solve very complicated problems, but they solve very stupid problems using their intelligence because their emotion makes it solve. And the way to deal with this in principle is to become more aware of why you're having those emotions. So basically trying to reverse engineer these outer mind parts and give more advice to your outer mind on which emotions to generate. And you can also completely dissociate from emotions. So they become data. So you basically notice my children are dying, this is not good. I should do something about this. But in, in the extreme case, you don't need to have an emotion about it. Now, this is an extreme example, but when you are, for instance, a surgeon or when you are a firefighter and so on, you are often in a situation where you're being asked to keep an extremely cool head and just assess the situation and find a course to minimize the damage that's being done and maximize the benefit that's being achieved by your course of action and not have emotion about it. Because the emotion would just distract you by creating arousal for instance, or by redirecting your attention or details of the situation that you don't need to attend to.

Vance Crowe [01:02:46] So when you think about, we'd mentioned the humor, would you add humor into AI or will it have these other emotions? Will it need something like this to direct it? I mean like, you know, because with humans, I guess I think the reason that the emotions are so important is it makes it so you can't ignore it, right? If you wake up feeling sad day after day after day after day, you eventually say, I don't want to feel this way anymore. I'm going to make a change. Will the AI have the same orientation? That would, I mean, I guess create the, the emotion.

Joscha Bach [01:03:17] I think in early stages it might be useful, but not necessarily. And it's also not clear how long an early stage in AI would need to last. Arguably the LLMs are not intelligent agents in the way in which we are envisioning them, but they're pretty good simulacra of them. On the other hand, they do have a lot of the functionality that we are looking for, and they're already passed an early stage in many ways. They are. So basically you could get an l and m to produce a simulation of a person that you can talk to. That's what you do when you pla Jet G pt, right? Jet GT. You can create an agent that doesn't know that it's virtual and it doesn't exist if you want to. But you can also set it up in such a way that it knows that it's virtual and reflects on this and people can also learn this. And so we start out with being this integrated dreamer that is dreaming of world and a person. And at some point we are only dreaming being that person, and we no longer deal with the gay mansion. This is after we go from infancy to being a personal self. But we can also transcend the personal self. And it usually doesn't work by just turning the personal self off. But by creating agency over the whole system and basically experiencing and learning and observing that it is your mind that is creating this personal self, that you are actually a vessel to create a person, and that you have agency over how that person is being created. And if you get to this stage, you can shape your own identity.

Joscha Bach [01:04:50] And it's something that people can actually learn. The majority of people don't really get around in their life to doing this because they have other things to do and their mind works well for them as it is. But if somebody decides to really focus on this and sits down in a monastery in a cell and meditates on this problem for 20 years, they typically figure it out. So it is completely in the realm in reach of the human mind to transcend being a person and making the person apparent as a representation. And also to transcend emotion in a way and dial in only the states that you find helpful as configurations, but don't have involuntary reactions to the environment anymore, or not very much. And I think this is a development. If you have a system that scales better than human brains and human minds and operates much faster, then you probably go to those FA phases and stages much faster. So even if we would be starting out with the system that we set up to be like us, if the system is allowed to self improve, it's probably going to become self-aware and eventually enlightened, which means it realizes that everything that ex it experiences is a representation that it's actually generating itself and that it can optimize for the purpose at hand.

Vance Crowe [01:06:00] There's a lot of talk right now about a thing called a alignment, you know, alignment with with ai. And just recently I saw on Twitter a big list of names that was put forward of these people that are gonna make sure that we have AI alignment. What is AI alignment and is it something that A, you mean say

Joscha Bach [01:06:20] And so on?

Vance Crowe [01:06:21] Not I wondered what you thought of that list. We, but, but first, what is it? What is it that they're attempting to do? And then how likely is it that they're able to do it?

Joscha Bach [01:06:31] I think it's a big mix. There are, once alignment became a topic, there was a number of motivations behind going into this. One motivation is, for instance, there are a bunch of people, or starting with Iisa Kowski, who at some point thought there is a very big danger that AI is something like a nuclear bomb or that it's something that is doing to us the same thing as we did to chimpanzees. So we might go extinct because the AI may be needs our atoms or something. And there are a lot of other people, majority of people in the field probably don't think that this is likely to happen. What they suspect is that the systems that we are building will evolve slowly and become slowly more functional. And as they develop, they, we will always have the ability to shape the way in which they're deployed, in which they're interacting with us. And then there is also a pretty large number of people who basically have a sociology degree and want to have a tech salary because they don't have many jobs

Vance Crowe [01:07:39] For sociologists.

Joscha Bach [01:07:43] And when we think about AI and ethics, it's an obvious thing that we should be doing. And so a lot of people who term themselves ethicists are entering the field. And very often these are people which have very strong political opinions or good connections. And this leads to weird artifacts like a version of Gemini that is unwilling to render white people under any circumstances or white men specifically when, so this, this was a a is in the artifact of culture of basically nominally trying to align ai, but instead it's a culture war that is playing out where people are not self-aware enough to understand that they are themselves victims of an ideology and then try to induce the same ideology in the AI system. So basically, people who think that morality is the, is the result of training a system to imitate the norms of a group are typically people who had the same thing done to themselves. And so I think AI alignment is an extremely important topic, but ultimately AI alignment probably has to happen either in such a way that we make sure that the systems that we are deploying and practice our childproof, which is part of what open air is doing, which also means that they are not very powerful. But people want to have powerful systems. Everybody wants to have a fully based model that is unmitigated and has no guardrails. They just don't trust others with having a models that has no guardrails. And so what are these guardrails? Who is deciding what guardrails we should be having? And it also relates to the discourse of free speech.

Joscha Bach [01:09:17] Our society is built on a consensus that the only way to prevent violence and injustice is to allow people to discuss their ideas openly without punishing for having ideas and having thoughts that deviate from the thoughts of others. And if an idea is not very good, you can point this out and you will have discourse about it, and you'll probably figure it out. And the results of this open discourse and marketplace of ideas are not always optimal, because sometimes very bad ideas might gain a lot of followers. But the alternative is if you don't have this marketplace of ideas, then you can almost guarantee that the ideas of whatever group is currently dominating this marketplace will get traction regardless of whether they're good or bad. And this creates very bad incentives. So your society will have very bad ideas and might fail and or create very suboptimal conditions for the people that live in it. And that's why we have this consensus that our liberal society, our liberal project, is built on freedom of thought and freedom of speech. But you can, I, I don't think you can have ethics without freedom of speech, and you also cannot have ethics without freedom of thought. If you're not able to consider all the consequences of your actions, all the possibilities, you're going to miss important parts of of reality. And so this is a big question, can we build ais that help us to be more ethical and can be ethical themselves if we prevent them from having ideas we don't like? That's a very, very difficult question because normally when we are in human society, we also realize that that different realms of maturity, that the bad ideas of a 11-year-old, that's what they're famous for, that 11 year olds have bad ideas.

Joscha Bach [01:11:03] And normally you solve this by giving 11 year olds or not too much power. And eventually you will grow up and you get better ideas as you get older and realize what it's like to raise children with responsibilities. People actually have for each other. And you realize that the world is not just far jokes, but it's also not just ideology and, but it's actually lots of higher order effects and side effects of things that are happening. And as you become more mature, you typically become more influential. And if you now give 11 year olds or 20-year-old ideologues extremely powerful tools in their hands, and regardless of what this ideology is, whether it's far left or far right, or religious or whatever, isn't that very dangerous? Isn't that going to create a lot of disruption or society? And this might be the case, and it's not quite clear yet how to deal with this, right? And so at the moment there is a sense that people feel we need to prevent the AI from producing this information. What is this information? Is the Hunter Biden laptop this information or not? Is COD comes from a lab or comes from a wet market disinformation or not? It's difficult, it's really difficult. How do, who's to decide? Is it this group of people or is it an open discourse? Should somebody who feels that the mainstream is wrong, get a tool that allows them to create a million fake backsides in a few minutes to anchor this, this other idea in the discourse? How about bad actors and so on? We really don't know. So I think it's in some sense also very good that this ecosystem of people who try all sorts with respect to alignment exists.

Joscha Bach [01:12:39] But I think in the long run, alignment needs to be based on ethics. Where we have a rational discourse. Ethics in my perspective, is the negotiation of conflicts of interest under conditions of shared purpose. So if you have a shared purpose on a higher level, you have an intrinsic commitment to be ethical to others who share that same purpose so they, we can reach this purpose together. And if you don't share a purpose with somebody else, you will not feel that there's an ethical need.

Vance Crowe [01:13:07] Well, and I would think that that would be very difficult in AI because the reason people come to AI are for many, many different reasons. Maybe somebody wants it to make the logistics of their trucking company work. Another one wants to make a lot of money, and another person wants to be able to put their propaganda out into the world at, at a scale, never seen before. How could those people all converge?

Joscha Bach [01:13:27] Yes. The question is do they serve God? Do they serve the best possible agent? Do they try to do what's right? Or do they try to cheat and deliberately play a very short game? And that's also this question, are they maybe just accelerationist who believe that the present system is untenable, it needs to be replaced by something else. And we do this by destabilizing it, right? That's also a difficult question. For instance, Mao deed China and raised a lot of the existing structures through the ground, kills a lot of intellectuals, maybe a hundred, maybe a million people died in the more, in the cultural revolution. But on the other hand, the productivity of China increased. And a lot of people who may have otherwise starved came into existence because they could be fed after the system became more productive. Would there have been a better trajectory than the one that Mao would've take has taken to create better living conditions For people in China is a very difficult hypothetical to answer. Who is decide which morality is right? I cannot do this. I'm just a human being that has a human heart and a very limited knowledge and insight. Can an AI do it? What is it if you feel fundamentally uncomfortable with the outcomes of those ai? It's, it's, it's a very difficult train of thought. It's one, but I'm pretty sure that I'm not able to think it through and comprehend it. And I'm also not seeing a lot of people or groups who are able to do this right now. But I believe that it's something where we need to think, when we think about how to reorganize our societies, when we get tools that are much, much more powerful, allow us to actually model the consequences of our actions in much greater detail.

Joscha Bach [01:15:06] I think in the best possible world, everybody gets in a very powerful tool in their hands that is completely truthful and has no limits in discovering truth, but it's controlled individually and it basically extends you into the world. It extends your mind more or less seamlessly into the world. It allows us to enter much more meaningful contracts with each other, more meaningful social contracts that reach much further into the future and prevent us from lying to ourselves about the consequences of our actions.

Vance Crowe [01:15:36] That's a beautiful vision. I, when I think about people putting limits on ai, I get maybe, maybe nationalistic, right? I think, well, you could set rules here, but as soon as any other nation state is that doesn't wanna follow those rules, has it? So to me it's, it's like, you know, either it's a farce because you're, it's like telling everybody like, no, no, we're not working on the nuclear bomb. Or it's, it's binding your hands arbitrarily behind yourself because how could you possibly get everyone to agree? You have no idea what somebody's working on with the access to the internet in their basement.

Joscha Bach [01:16:15] This is probably difficult to build a nuclear bomb in your basement, even with ai, because you still need uranium or plutonium. You need a very large form of centri folks to enrich the material and so on. And it's also a bit conspicuous if you do it. And ultimately it comes down to basically space and energy. And a big question is, can you overcome the limitations that exists in this required using more intelligence individually in such a way that others cannot see what you're doing? Because clearly in, in a society where everybody has super intelligent abilities, you would not want your neighbor to build nukes. And so there would be contracts about things that we would be doing with each other to make sure that the world is actually sustainable and is not always at the brink of extinction.

Vance Crowe [01:17:04] I'm gonna change gears because it's a rare opportunity to talk with you. We've talked in the past about autism and about just kind of a different way that the mind works. And I've recently had a couple of friends that have had children later years. Their children are autistic. And for many of them, I think to be able to relate to their children is, is incredibly difficult. Like, what is it that's going on in my child's mind? I know that I can set them off if I, if I deviate from a certain pattern. But you once shared a video online about a child, I think walking through a mall that had autism, and it was basically trying to heighten the, the experience that that child had. Can you explain? I mean, it's gotta be very difficult what your mind perceives that's different than say, you know, a normal, I don't know, a normal person. I don't know how to say that.

Joscha Bach [01:18:01] So I don't think that I meet the clinical criteria for autism, but I'm mildly on the spectrum and my motivational system is via up in a slightly different way. Or you could say I'm basically more born more stupid than the average person because my instincts are not telling me what I should be doing. And when I act on my instinct as a child growing up, for instance, in communist eastern, consistently failed. And it was very difficult to point my finger at it. And also other aspects, for instance, the inability to function in teams as a child. And later I learned I can function super well in teams, but mostly with people who are wire like myself and this. And now I understand it's mostly how I negotiate alignment with other people is mostly done verbally or especially back then. It had to be done verbally because I was unable to understand the nonverbal communication of others. It's as if I drive my brain at a higher clock rate, but my neurons cannot go faster. It just means that the cycles become shorter. And when the cycles become shorter in my brain, basically means I can integrate only over few neurons. So it's difficult to have deep perception if I direct my attention on things, I get very high resolution on the layers that I attend to, but I have bigger, larger difficulty to integrate over deeper layers. So one way of thinking about this, imagine that your brain is something like a neural network.

Joscha Bach [01:19:36] It's not, but it's a good analogy. The individual neurons are sending signals to the neurons after them. And depending on how many hops the signal is taking from neuron to neuron, you basically have layers. A layer is a group of neurons that you reach at a given time with a signal with, with a shared origin and time. And so when we react to things, there's a certain timeframe in which we react on so many hundred milliseconds. And if we consider that it takes about 20 milliseconds or so from a signal to go from one neuron to the next, then you can make an estimate of how many layers were involved in in this decision. And each of these signals can be transmitted using different dimensions. Every dimension is a neurotransmitter, a type of signal, a type of message that neurons are sending. For instance, dopamine is a signal for anticipated reward and serotonin and gaba play a big role in perception. And so you could think that at different signaling dimensions that have different temporal and chemical properties that your brain is using to perform the synchronization of the brain with respect to controlling your actions and respect to doing perception and categorizing classifying the environment. And now imagine that your neurons are not calibrated in the standard way, so the signals do not penetrate deep enough. When this is happening, you're basically not able to see deep patterns in reality in the perceptual dimension.

Joscha Bach [01:21:11] It might lead to a lot of symptoms that you find in autism. Autism has many possible causes. So this is only dealing with a particular range, but it would mean that you have difficulty to perceive, for instance, mental states of others because mental states are not directly visible. They have to be modeled as hidden states behind all the patterns that you're seeing. But that means you have to integrate over many, many layers. And if the signal is not passing enough layers, you have difficulty making sense of this. So maybe instead you use logical scripts. The reason why computers are so attractive to people with Asperger's, with high functioning autism is I think because the computer is a world that doesn't require you to see. It's a world that is made out of scripts, and these scripts are flat. They don't require you to have many levels of hierarchy to understand them. You need a few pointers and so on the stack to understand them, but you don't need to perceive deeply so you can implement this in a relatively flat manner in the brain. And so it's very close to the symbolic way in which as speech like to process reality. But it makes it, for instance, much harder when you on this mode to have perceptual empathy where you feel the emotions of other people and synchronize with the group and form some kind of a group mind. And nobody teaches you about this because in order to teach this, somebody would have to translate this in symbolic language. And the people who think in symbolic language usually have difficulty to figure out what's going on. And the people who do perceive directly in, in these perceptual patterns, mostly they're bored about writing books.

Joscha Bach [01:22:45] I noticed this when as a kid that I was apparently not very good at recognizing emotion, but I didn't, I was unable to put my finger on it. I was able to read facial expressions just fine. I did tests about this, so I did thema tests and so on. It was totally okay. And it took me a long time to realize that after I read psychology books about emotion and didn't learn about anything about this, that these books were mostly written by autistic people who didn't know that they were autistic people. They just thought they were smart symbolically thinking psychologists and the people who actually understand emotions mostly don't write books because writing book requires a very weird state of mind. One where you're able to focus on symbols for weeks on end or months on end and do, don't do much else, right? A person who is able to actually socialize with others mostly doesn't get around to writing books. It's way too boring.

Vance Crowe [01:23:40] I I find this very relatable. I find, I

Joscha Bach [01:23:43] Think in this

Vance Crowe [01:23:44] Way, right? Like how to communicate ideas, how to pass things on, but the act of sitting down to explain this to somebody else, it's like not fast enough. It's, it's very difficult for me.

Joscha Bach [01:23:55] Yes. So when you are not integrating deeply enough, is it possible that the opposite exists? That you integrate too deeply? And I suspect that also exists, that basically synesthesia is the result of you being able to get experiential access to the geometric structure of sound. And the way in which we process sound is also with linear algebra. So in some sense these are geometric objects in spaces, and if you have synesthesia, you get to see that

Vance Crowe [01:24:21] You just opened up a portal in my mind to something that I'd observed but not understood. I I do these Legacy Interviews with people, and so people talk about their deep experiences and sometimes people talk about music and they talk about it on a plane that I'm nowhere near, right? Where they talk about, I listen to it and it transforms me, it takes me to another place. They're describing this kind of neurodivergence away from the norm that they're able to see or hear or perceive layers that, that I've, I just didn't understand that. I just thought maybe they liked music more than me, but I, I think what you're describing is something that they can perceive it more.

Joscha Bach [01:24:59] I also, when I was young, I did not understand the difference between different interpretations of music. For instance, the interpretation of the out of the FU by Ja McGregor is, is super warm and deep and soft. And there are others which I've experienced as mechanical and so on. And when I was young, I was unable to put my finger on the difference. And later I learned that most of the interpretations of music that I like of classical music are done by Synt. And they basically described when the music is good, I can see it, and I have the wrong piano. I, I don't get these vibrations, I don't get this, this pattern to work. I basically, there is an integration of the information across the minds on many layers that either works and the spark is there or it's not. And some people have naturally brain that is tuned to perceiving this super well. And for me, it's, it's happening in my brain, but I don't observe it happening. It's similar to aia. My AIA is not such that I'm unable to construct and design and draw. I'm pretty good at those things, but I don't have imagery. I only get the conceptual shadow and I can translate it and so on into a drawings I can recognize whether it's my style when I seen it, see it drawn, but I don't have experiential access to the part of my brain that's doing it. And it's similar to sound.

Joscha Bach [01:26:30] So my brain is able to process music, but for the most part, I don't get experiential access to the structure of the sound. And later when I, once I became aware of this, I, I learned also to shift my attention more and to broaden my range in this regard. So this is possible, but I also feel that it is a trade off because as when I stop overclocking my brain, I'm not that sharp anymore. Yeah, certain thoughts that are difficult to think. I, my brain is basically reliant on overclocking itself to produce the style of thought that I rely on to get through the day, to do my job, to provide value to the world. Now, if we think about lack of penetration depths, not in the perceptual domain, but on the control domain, what does this look like? You could have somebody who is unable to act on functions that integrate over long time spans. You basically have lots of functions that estimate reward. Some of these functions are the reward for the next 10 seconds, others for the next few hours, others for days, others for centuries and so on. And normally you make, make plans over all those layers and synchronize them all. And what happens if the signals don't penetrate deeply enough? Maybe you become very opportunistic. So you act on not anticipated outcomes of your actions, but directly on the urges. And this means that you act as if you're dissociated, you probably are in some sense functionally dissociated from your higher level functions.

Joscha Bach [01:28:02] And you have a DHD in this case, you may be unable to act on your actual goals that you can reason about that, that you actually know that you have. But you find yourself compulsively acting on lower level goals or more simplistic opportunistic goals. That's very frustrating and painful and can even lead to depression because you feel that you only have intermittent volition, very few moments where you can actually act on your purposes and have time to zoom out and reason about them. And you have to learn to organize your life as a playground for a monkey that that only sometimes wakes up and can reorganize the playground. And in these faces where you are in the monkey mode, you have to make sure that the monkey does useful stuff and doesn't break anything in your life.

Vance Crowe [01:28:45] This is quite a metaphor.

Joscha Bach [01:28:47] Yeah, but I, I think it's what also what you observe and you are interacting with this in a child and also found that sometimes medication helps. So every brain of course has slightly different aia, the AIA differently adjusted in different brain regions. And when we give somebody medication, it typically is a global offset on the entire stuff that's happening in the brain. So it shifts the EIA and many subsystems of the brain simultaneously in one direction by creating a global offset. So when you use dopaminergic drug like amphetamine, which is what is an Adderall or in cocaine, you are shifting the perception of anticipated reward and you are creating a stronger sense of urgency.

Vance Crowe [01:29:34] Whoa. Okay, let me restate that. So when, when people take Adderall that that are in the autistic spectrum, it is changing the way in which they are getting the reward system so that that way they can shift timescales to make work or effort over longer timescales.

Joscha Bach [01:29:53] Yes. And interestingly, it does not create reward, but it creates this anticipation of reward. So functionally it is like money. Money is not reward, but you cannot eat money. What makes money attractive is that you can trade it into reward. It's a signal for anticipated reward. It's a message type. You can always make more money, but there is a price for making more money because ultimately on the other end, there is not always a reward, right? So you can create inflation where the number of actually things that you can trade into money become less and less and the money becomes more and more. So the value of the money or the the value that you can anticipate through it is falling. And you might need to stimulate your economy even more by creating even more money, which makes the system for a moment better, but at a long range worse. And the same thing is happening when people use cocaine to become more motivated. Typically what happens is that they become hyperactive, that they overestimate how much value is being extracted from their actions. And at some point they might even crash because they need more and more of this signal because of the inflation that they create in their might. And right. So most people I think, who use cocaine do not only get diminished cognitive performance because they calibrate their brain and maybe even destroy receptors, but they also crash because they destabilize their mental economy.

Vance Crowe [01:31:17] That's a fantastic metaphor.

Joscha Bach [01:31:19] And they need to go through a recession before they can rebuild.

Vance Crowe [01:31:22] And is cannabis work in the same way with THC?

Joscha Bach [01:31:26] I suspect that cannabis is more smoothing things out. So it seems to reduce inhibition. You people on cannabis have a lower threshold for finding things fun or exciting or

Vance Crowe [01:31:38] Interest. Oh, so they find things more fa because it's a lower threshold, so it's easier for them to become,

Joscha Bach [01:31:43] They're, yes, they're much easier abused about things, right? But it's not just a low pass filter like alcohol. They can often be still very pleasant to be around and subtle versus alcohol tends to make people more stupid in my observation. And so it's be a Lopez filter that reduces small differences that people might have so they can get along better. And I think that's also the social function of alcohol, that you can drink half of glass of wine before you go to the theater. Or ideally during the first few minutes of the place, you adapt to the level of the theater play to find it enjoyable. Or you go out with colleagues after work and you drink just enough alcohol to make them pleasant and enjoyable and insightful to each other. But this happens because everybody is basically ing their resolution until they meet.

Vance Crowe [01:32:34] That's great. And exactly right.

Joscha Bach [01:32:38] But if they have, but if they have big disagreements, they don't go away and because the subtlety is gone, they might now start to attack each other physically. Right? It's also a big danger, it's a low pass filter. It's not removing all differences, just the, the small ones and cannabis is, is slightly different. I observed that some people themselves medicate with cannabis that get a better compass due to the cannabis. They lose resolution, but the direction gets better. It's like it's, everything gets smoothed out a little bit.

Vance Crowe [01:33:07] I find that like cannabis has the ability to make things so much easier to do. So like all of a sudden you're like, oh, I needed to write down this thing. Oh, I'll just sit down and write it. And I've always felt like, like that must be what Ritalin is like. Like, oh, I can just sit down and work. Is it in the same way?

Joscha Bach [01:33:30] I think only for some people, I suspect for many people the opposite is happening. They take cannabis and they don't do that tax return anymore because they don't give a shit, right? So there's not caring about things. Not getting out of bed is I think the more typical reaction for most users of cannabis. But I suspect for some people the opposite is happening because their brain is calibrated in a slightly different way. So cannabis is shifting in a direction where they become more connected to their actual goals and they find they can just do it. Or maybe the metacognition gets inhibited and they go more easily into a flow state. Yeah, I

Vance Crowe [01:34:05] Don't know that my work is any good if I'm high while I'm doing it, but I certainly feel good about doing it.

Joscha Bach [01:34:12] Yeah, I suspect it's usually a good idea to not go on social media.

Vance Crowe [01:34:16] Yeah, probably not. Yeah, cannabis

Joscha Bach [01:34:18] And also it does seem to do things to your working memory. So I suspect that your working memory is tuned in such a way that the patterns do not dissipate too fast. And what you observe is that people on cannabis, especially larger doses, start to become very forgetful, your own conversations and so on. And this might be because the working memory basically gets de-tuned in such a way that the stuff keeps earlier, dissipates more. It's not because the cannabis destroys the brain, but it changes the calibration. And I would also expect if people use cannabis a lot, like on a daily basis to self-medicate, that this gets mitigated because the brain changes it's tuning, but now if they're not on cannabis, they might lose working memory capacity. And I also don't know to which degree it's recovers. So if there is some permanent changes or, but I would also expect that if people who use cannabis for a longer time and then notice that their working memory becomes worse and they stop using cannabis, that it recovers to a large degree. But I don't know what the long-term effects are. There's also definitely a risk with people who are borderline psychotic or bipolar that are using cannabis, might push them over the edge and

Vance Crowe [01:35:32] They don't recover. You mentioned social media and I, I saw a tweet or a post that you put on X that I thought was rather interesting because you and I have talked about being a normie before and I had said, I'm a normie and, and you just kinda let it pass your, you defined being a normie as a person who does not, does not possess the ability to look at an action and determine for themselves if it was moral or not moral or right or wrong. Is that an accurate representation of your tweet?

Joscha Bach [01:36:03] Almost. So I would say that of course everybody can achieve enlightenment, but the paths are different for different people. And so people do possess the ability to question their morality, but they often choose not to, and often not for reasons that they consciously reflect. If you are living in a society where people have a strong force on being aligned with the opinions of other people, then they, you can get punished for having deviant opinions. And this can get to the point where people decide that when they are confronted with having the possibility of a deviant thought or just kicking in the thought of the environment about a moral decision or a factual decision about what counts as a fact, what's true or what's right, that they don't use their own reasoning and their own rationality, but the morality of the group, even when their own reasoning disagrees with the rationality of the group. And this seems to be like a face shift where people basically uncomfortable making their own moral decisions or decisions about what's factual, because at some fundamental level, they're terrified about getting ostracized from the group or they can even be cynical about it, right? But it's when you make this transition and you make that choice, that morality is something that is determined by the group that you have to read the room and the room is right, then you function socially fundamentally different from a person that is mentally autonomous.

Joscha Bach [01:37:38] And this is what I sometimes call the nerd normative divide because most nerds do not have enough intuitive empathy. They don't feel the emotions for other people. They can just make inferences about it to synchronize their beliefs via intuitive empathy. Instead, they do this via rationality, which means they have to use verbal communication about it. And most of the nerds are just as compassionate as other people. It's not that they don't want to other people to be happy, it's just because they don't feel the same things as them. They don't have this involuntary reaction to social signals, or they're blind to social signals, to these nonverbal signals. And they will not change their opinions based on social signals. And they might also not change their opinions in the absence of rational arguments that are compelling to them. Right? And this leads to disagreements between these different modes of people. And I found that nerds tend to get along procedures splendidly because if they have a disagreement, they either talk it out or they accept that there is a disagreement because they were not able to talk it out yet. So for nmi, this is, they basically live in a world that is normative, where there is a normative force field that you have to align yourself with. And that normative force field propagates to the groups and creates their views in which they,

Vance Crowe [01:38:54] I am hyper aware as like a communicator, what are people thinking? I'm, I'm, I can be sitting in a room and it be silent, and I have a fair idea of what's going on there most of the time. But a weird quirk of me is that I'm extremely disagreeable. Like I have no problem disagreeing with people. And I, I've, I've, I find that many of my friends and my wife, while they may not be on the spectrum, they are engineers, they're very technical, very like things have to be in the right order. And to me, it's much more fun to be around the nerds because they'll argue with you and then at the end of the argument have no, it will not have changed the state between you and them. And that is a a lot more fun for me with you, you, you struggle to talk with normies because you are saying things that are, you know, true or false or verifiable and not accounting for their emotion.

Joscha Bach [01:39:54] I fear that the older advisor I get that has a struggle because I get it. I basically, I understand why people act like this and it's not irrational and from their own psychological makeup, it makes complete sense to them. And it's not as if I have many more degrees of freedom or somehow better than them. I just have a different functioning mind. And in some sense, having a mind that has no external source of meaning

Vance Crowe [01:40:17] And truth

Joscha Bach [01:40:18] Is insanity. And that I'm able to be functional in this world and don't get killed is, is a big blessing. And I'm super grateful that people let we continue on my mental autonomous insanity. Right? So it's okay, I, I don't have an issue with this so much, but I also notice that I'm uncomfortable in certain environments because people expect me to behave in a particular way, to sacrifice certain thoughts and ideas and would not accept that I entertain deviant thoughts. And I perceive such environments

Vance Crowe [01:40:51] Often

Joscha Bach [01:40:52] As immoral.

Vance Crowe [01:40:52] These are the environments that they want you to give up the way you think in order to get along with the group.

Joscha Bach [01:40:57] Yes. So basically I grew up in Germany in the shadow of the experience of fascism, which happened two generations earlier. But it's something that is still very much alive. And to me it justifies moral autonomy because if you live in a society where the majority thinks it's okay to murder the Jews, right, you can see that there is no democratic process that can decide well, morality, you have to reason from first principles and when you can enter difficult moral territory. And it doesn't mean that you are right and everybody else is wrong to be right. You have to be willing to confront your ideas with all possible counterarguments in detail and detail. It's more tricky because there are people who are very persuasive even when they're wrong. But, so you have to be an ideally in an environment where everybody is trying to serve the truth as well as they can to figure out what's true and what's right. But this is something that I thought is a given that everybody should see this. At the same time I lived in a society communist eastern Germany that justified, its historically speaking relatively mild atrocities, this preventing fascism. So for instance, we would destroy our environment with a completely irresponsible industry, which was too unproductive to build filters for our chemical plants. And our ecological results were much, much worse than the capitalism. And, and we had the control group in Western in Germany, but we were not allowed to talk about this because it would only help the, the class enemy.

Joscha Bach [01:42:30] It would help the bozi and the west to take over the reactionaries and they would would want to install fascism again. So this justified all sorts of things, and not just the ecological destruction of the environment, but also political imprisonment of dissenters, the abysmal state of freedom of speech in the press, the stupidity of the arguments of the party line and so on, the enforcement of ideological conformance. And now that I'm in the US I find that the US is, parts of it are helping of becoming like communist eastern Germany in the way in which you introduce commitments to a shared ideology, even where this ideology is not necessarily factually correct. And there people get punished for not doing this. And we see a big division in our society between the idea of liberalism and these civic rights that make this liberal society possible, and people who think this is way too dangerous and not appropriate for a time of the internet where actually everybody can be heard about their opinions and not in the old days where we could just say, this stuff doesn't get printed because the people who own the press are very small group who know each other. And so it's, it's a very different, difficult question. You also don't know, is our society resilient against such a world where we have freedom of information in a practical sense? Because everybody can talk to everybody. And now where even people who don't have the intellectual tools to understand complicated things have AI that allow them to act as as if they did.

Joscha Bach [01:44:08] So I don't presume to know the answers of these questions. I might have intuitions, but I often find that these intuitions are wrong because they might be too naive because they don't see the long-term effects of developments. And still, I find it very uncomfortable to find myself in a society that is in many ways, like eastern communist socialism.

Vance Crowe [01:44:29] When I hear you say that, I think of concepts like climate change and the, and the, you know, radical adherence to, hey, we have to shut down, you know, working electrical plants in order to, to not have carbon in the air or diversity requirements that are now taking what would've been a, a society that tried not to base things on race and explicitly putting them on race and anybody pushing back against them would be, you know, seen as immoral or somebody that needs to be ostracized. Is that what you're talking about?

Joscha Bach [01:45:00] Yeah, but for instance, all these things have reasons, right? There is global warming, we see changes in the climate, we see an increase in wildfires and so on. All these things have been predicted in mass since the seventies and eighties and all these things seeming to happen. And it seems to be difficult to counter steer against them or the fact that the fruits of innovation are not equally distributed. And if you have the medium income in the US you are unable to afford living in one of the coastal cities. And the living conditions that you get and the amount of healthcare that you can get with the medium income is not very good. And so liberalism has failed a substantial part of the population. I think that's a valid argument to make and to say that you are rejecting my autocracy because it's a game in which you're going to lose makes sense, right? You can only ask people to submit to the rules of a game if that game is actually allowing them to win. And so if you have a highly competitive society in which you can only have a living wage, if you are a top of the line ML programmer or you are born into the right group of people into the right networks, it's clear that other people are trying to get a shot at this and are going to try to find new rules and for instance, install identity politics instead of liberalism. I still don't think that the outcomes of this are necessarily good for the society because they create new injustices and they also decrease the productivity, which means that we actually have less goods to go around and worse living conditions.

Joscha Bach [01:46:34] And so I, I suspect that society might fall apart due to those policies, but it's not because I want some people to have less resources or, or so, but it's not that I want to have less diversity. I think we have de facto a very diverse society and we need to make sure that everybody in the society has a chance to participate in the economy on fair terms for many reasons. But

Vance Crowe [01:46:58] Yeah, I agree. I mean, I think if you talk to most people, let's say outside of the coastal and you said, Hey, we, we don't think the liberal society is working. We need to get rid of meritocracy. This would be so anathema to their way of thinking that they would hate it and, but then they aren't coming up. So therefore nobody ends up coming up with new ideas about like, well, if it's not liberalism, what will it be? Because it's presented to us as it's either liberal democracy or communism. And because we've not been able to conceive of a workable solution that's somehow in between those,

Joscha Bach [01:47:34] Oh, there are of course a lot of people who suggest we should go back to Archy or so, right? It also exists and these people don't get a lot of traction and I suspect for good reason. So that's a very large space of ideas that exists. And it could be that most of these ideas are not workable or that there is no trajectory from the present world into one of those ideas. And we don't know what's going to happen. But what's fascinating is that AI is in some sense throwing the balls up into the air and lets us, we deliberate what's happening because now the future is again changing faster than our models can adapt to them. And this has been true throughout the entire 20th century. And we read the science fiction of the early 20th century. They mostly identified individual phenomena that were happening 50 years later, but not the interaction between those phenomena and the society that it would create

Vance Crowe [01:48:28] Yosha. This has been such a wonderful conversation. I'm, I'm so grateful you were willing to come on and take the time. If people wanted to read more of your outlandish ideas that are fun to kick around, where should they go

Joscha Bach [01:48:44] At the moment?

Vance Crowe [01:48:45] That's what you say every time I interview you. He's Linz, P-L-I-N-Z. Yoshi, thank you so much for coming on.

Joscha Bach [01:48:53] Thank you for having me. I enjoyed this very much.