I will probably get it again. Anyway. So just to describe the format quickly first. So what will happen is that Ben and Reza will talk for the first hour and a half or so, or maybe two hours. Then we'll have a break. Then we'll reconvene to have half an hour or so of questions from the audience. This will be recorded, as you just heard, and transcribed and form the basis for a published version later this year. First of all, some housekeeping notices. Please mute your microphones during the first portion of the conversation. Like I said, there'll be time for questions at the end, but please keep them brief. So no more than 60 seconds and one question at a time to give our speakers a chance to answer questions from several people in some degree of depth.
If you have a question, you can either unmute yourself and ask away, or write it in the general chat in the Discord, and I can call on you or ask it for you. Also, please, if you can, could you mute any notification sounds from Discord, which you might have opened another tab, by right-clicking the server icon and choosing the relevant options, which I unfortunately failed to do myself before beginning today. In the paper, The Tyranny of Scales, Robert W. Batterman asks, how can one model the behavior of materials that display radically different dominant behaviors at different length scales? Although we have good models for material behaviors at small and large scales, it's often hard to relate these scale-based models to one another.
Macro scale models represent the integrated effects of very subtle factors that are practically invisible at the smallest atomic scales. For this reason, it has been notoriously difficult to model realistic materials with a simple bottom up from the atom strategy. The widespread failure of that strategy forced physicists interested in overall macro behavior of materials toward completely top down modeling strategies familiar from traditional continuum mechanics. The problem of the tyranny of scales asks whether we can exploit our rather rich knowledge of intermediate micro or meso scale behaviors in a way that would allow us to bridge between these two dominant methodologies. These are words we would do well to keep in mind when considering the scale of artificial intelligence in the world today, or perhaps the differing scales proper to artificial intelligence.
Before inviting Reza and Ben to give us their reflections on this notion and others, I'll read out one more quotation, this time from the abstract. of the paper, Intelligence as a Planetary Scale Process. Conventionally, intelligence is seen as a property of individuals. However, it's also known to be a property of collectives. Here, we broaden the idea of intelligence as a collective property and extend it to the planetary scale. We consider the ways in which the appearance of technological intelligence may represent a kind of planetary scale transition and thus might be seen not as something which happens on a planet, but to a planet, much as some models propose the origin of life itself was a planetary phenomenon. So Ben, do you want to start us off? Yes, thanks, Ross, and thanks Reza for the
invitation to speak with you today about the topic. I'm really, I'm honoured and hope to, that our conversation will take us in some interesting and fruitful directions. I'll start off with this prompt then around the planetary uh and the question of scale um here is to take the bait so to speak i suppose for me the question of intelligence and planetarity um uh relates to scale but not but but in in ways that are not necessarily reducible to the planetary scale as such that is there's a way of thinking about the planetarity of intelligence that is that over millions of years and then, you know, obviously, you know, even much longer, evolutionary processes have converged to fold matter in some ways that are generic,
some ways are idiosyncratic, to perform in such a way that are able to perform a panoply of feats of activities that we might recognize as intelligence. Again, some of them are convergent, Some of them are divergent. And so just as a kind of fundamental observation, that whatever we might recognize as intelligence on Earth is, in the broad scheme of things, something that the Earth has in fact done. that is a feature of planetary processes in and of themselves, as opposed to a kind of immaterial, numinous force that exists separate from that underlining geologic and biochemical churn. It is, in fact, one of the things that this churn does.
At the same time, we are interested in the planetary scale. Hold on. I'm going to start sharing some slides here just in the interest of hopefully clarifying some of my points. One moment. All right. um so in in the book that i assume everyone here is more or less familiar with of the stack um was a meditation on the question of what planetary scale computation is uh which is not the same thing as a planetary scale artificial intelligence though obviously they are uh they're they're
related. The question that animated the book was understanding computation, not to argue that it was not these other things, but to argue that computation was also a, that can be understood in terms of a philosophy. So there's a philosophy of computation in terms of algorithmic theory, the limits of computability, and so forth. But there's also a way of and necessity to understand computation, not only in these terms, and not only as a kind of force that animates certain human-scale appliances, but also as something that has become infrastructure. Not only has computation become an infrastructural form,
but the infrastructures that animate social systems have become themselves increasingly computational. And in this regard, that the infrastructures have become increasingly cognitive or capable of increasingly granular and also increasingly large-scale forms of cybernetic recursions of sensing, modeling, and affecting that not only have technological effects, but also have indeed geopolitical ones as well, and are reorganizing our political geographies in their own image. Another way is it might argue that in a certain sense, this is not an instantaneous break with the past, but has some genealogical links with it.
the Antikythera mechanism was sometimes considered one of the first important analog computers was, of course, a cosmology machine. It was as a calculative device was used to orient location, to perform general algorithmic processes in relationship to navigation, in relationship to location. And so this correspondence between the artificialization of intelligence in relationship to astronomic cosmology, but then also the location of the calculating agent, it's us, in relationship and locating him or her in that cosmology, one could say, has been primordial to
the genealogy of computational technologies in the first place. And indeed, more recently, with the development of the Event Horizon Telescope, a matrix of sensory nodes distributed across an entire side of the surface of the planet linked together like the modes of the parts of the retina of a larger sensory apparatus allowed the planet, not just, was a machine not just for the orientation of the user within the planet, but the mechanism for the orientation of the planet itself in relationship to its astronomic condition. And the introduction, therefore, not just of, but the introduction
through the measurements and modeling of a scale that may have been inconceivable to the to the designers and engineers of the Antikythera mechanism. All of this might be understood as part of the history of what we might call planetary sapience, which could refer to at least two things. One is, as I said, the recognition that the emergence of sapience, again, and it's perhaps in some distinction from sentience, is a planetary phenomenon. It is geological. It is biochemical. It is something that is not distant, is not separate from this world. And so some of the ways in which newer critics might argue for a, in response to the recognitions of the Anthropocene,
in which anthropogenic agency is understood to have been a kind of tyrannical alien force upon the planet, we might do well to more closely recall that this sapience, this anthropogenic sapience, the agency of this anthropogenic sapience is itself something the planet is doing through us in a way. But also the emergence of this larger apparatuses of cognitive infrastructure, such as those that have emerged in relationship to climate science. One of the models of planetary scale computation that I think we want to pay very close attention to are those provided by climate science.
That is, it is, and as I will, as I've argued that the very reason we know climate change is happening in the first place is because of the, this planetary scale sensing, modeling, simulation infrastructure, producing simulated models of the past, present, and future that spit out hockey state curves. that allow us to not only to understand that climate change is happening, but through this to comprehend a temporality of planetarity in which we are involved, to recognize the effects of anthropogenic agency in relationship to those outcomes. And indeed, anything that may have come, that may be abused from the theorization of the Anthropocene
is then in this regard a secondary effect of computation. a fact that should be taken perhaps more seriously in asking the question, not just what planetary scale computation is, but ultimately what is it for? What purposes and futures and kilos may we decide for it? Now, thinking about scale, one of the other projects that we're currently interested in is around what we might think of as simulations as an epistemological technology. And in this regard, we think of simulations as shown in this image, not only in terms of simulated experiences like AR and VR, but also the role of simulations in cellular simulations, epidemiological simulations, and any climate simulations.
epistemological technology as we have it is a technology whose main main effect main social or cultural even philosophical effect may be less than what it allows us to do than what it that then how it reveals to us how the world works that in the deployment of this technology something fundamental is revealed about the world itself the copernican traumas from the use of this from this technology, and that this is its more fundamental, as I say, rather than its immediate instrumental use. And so the argument that I would hold is that AI should be a similar kind of epistemological technology. We don't really know what AI is and what AI is for at this point.
I will argue that we have produced many, some impressive, some less impressive forms of machine intelligence, but really the ultimate purpose of these and the ultimate lessons for them are probably in front of us. There's also cultural issues to be considered here. That is that many ways, different cultures would define the question of artificial, what constitutes the artificial? Obviously, more ways to define intelligence. project that Bray's is involved with is a book that we're co-editing down at Greenspan of Bogdan Konya called Machine Decision is Final, which is on the history of AI in China and some of its implications for the planetary future of AI. It's a topic of extraordinarily interesting. But again, the question of what AI is, is not necessarily, it is not clear what it
is or what it is for. And we observe, obviously, that it's currently used for both very smart things and very stupid things, but it's also largely undiscovered. What we don't want to do is to fall into trap of what I call premature ontology, that is to confuse the initial implementations of a technology, the significant is AI, for what AI actually is. And I think some regards, some of the forms of the sort of clash AI ethics discourse probably do this too much, where there's an identification of many of the noxious things that AI is used for that leads many of its adherents to the conclusion that, in fact, AI is nothing but those
things. I think this would be a mistake. I think we need to take, as the man said, a long-term view of things. Natural intelligence, or biological intelligence in the form of primates, billions of years have developed, tens of thousands of years since the cognitive revolution for homo sapiens. So an evolutionary perspective of machine intelligence should take a similar kind of long-term view. A few then things to table in terms of our discussion around these models of AI that we should be considering. And I mean this in a plurality of models of AI to try to get a sense of not only what it is, Turing's Gaudensk experiments of the commonly Turing test in which
a human, a computer is, for him was a sufficient condition of intelligence. That is, if he can fool a human into thinking that perhaps it is a human, he asked, must we not see that there is some kind of intelligence at work there hasn't unfortunately become a necessary condition of intelligence. That is, unless the computer can perform thinking the way humans think that humans think, then we would just from being intelligence. And so this is established, particularly in the West, less so in China, but particularly in the West, models of a deeply anthropomorphic, but also deeply individuated, as Ross had mentioned at the beginning, a model of intelligence as something that is contained within the boundaries of a single organism rather than in the interactions
and mind modeling and counter mind modeling, social intelligence and conversation and cybernetic intelligence that organisms or other entities may have in relationship with one other. So roughly speaking, this question of the models of AI and this, if anything, it's a call for a a multiplication of the such models in order to understand a little bit of where that dynamic of convergence versus divergence, what is generic about internal intelligence and what is polyphenous about the diversities of intelligence can actually be properly understood and therefore properly mapped and further artificialized. So I'll draw then very quickly for conversation a kind of
bimodal schematic of model A and model B. Model A, which we're probably a little bit more familiar in the discourse at least, and from science fiction movies, but is the one that I'll try to argue against, would hold roughly that AI is almost human and should be more human, that it appears in animate guises as a agent or a buddy or a secret supervisor. Ideally, AI would replicate human representational thought in silicon such that it can seamlessly automate normal human tasks. The more then that AI complements human intuition, the more successful its integration with human culture, which is its purpose, according to this model. According to
the model AI is or should be, therefore, an uncreepy reflection of our psychology and our political economy. And so the spectrum of AI interaction would run from, on the one end, docile subordination, my buddy is working, my A servant is behaving properly, to active malice. The AI wants to kill me. But it's in-conification remains. But because of these correspondences, the AI, and because this model would hold, AI might one day pass human intelligence on some kind of shared track, a kind of simple singularity model. That there's a single track of intelligence, that there are things in front of one another, and that this is going to be passed. The solution
to AI bias for this model is usually some form of politicized sentimentalism to return to what has been lost, to recapture what has been lost to the Copernican traumas of AI in its early forms, to make it more naturally human and to recover a sense of humanness that it has destabilized. Model B, which we might take as the more positive model, is one that is driven more by, I might say, a kind of landscape theory of AI, one that focuses on how all forms of intelligence are dependent upon how they sense the world and not just not just as information but produce
information about the world um and how the process of immersive processes that that that produce cognition are in fact dependent on those sensory mechanisms which are can be widely distributed so for this i would be defined as a heterogeneous collection of different sensing and signaling signal processing technologies that augmented diverse complex systems, including distributed emergent condition. It may be more like a synthetic rainforest than a robot teddy bear. On the one hand, deep learning obviously has a lot of functional isomorphs with anomalian neurologic processes. It is a model of bottom-up intelligence for sure.
human and human AI interfaces can obviously mimic you as well I'll show you in a moment particularly in terms of NLP this model would hold that ultimately AI or intelligence or broadly more broadly would only partially reflect and overlap human it is ultimately it is messier and ultimately somewhat different to those systems intrinsically. And this is, they would argue, is not the problem, but the point. This model would hold that risk should be addressed more by contestation and explication with multiple contrasting patterns. It would hold that intelligence exists in the world in disparate forms, many of which could, in principle, be augmented by AI, not just
augmentation of human intelligence. And therefore, most importantly, a durable and sustainable, let's say machine intelligence culture, one that is predicated upon its secular de-anthropomorphization. So I then would make some of these arguments. We can go deeper on this a little bit more half of model B at the expense of model A, despite the presence of this. Let me just say a couple more things and then we'll pass it to to Reza um one of the questions that we're sort of we've been considerable interest um of late uh both in the both in in Strelka and afterwards and and in some
of the more the work for the new book has to do with status of the artificial of what constitutes the artificial what constitutes artificiality um what is the philosophical not just philosophical status of artificiality, but what is the philosophical importance of artificiality. Now, the term means lots of different things. One of the ways in which artificiality is identified is in terms of anomalous regularity, anomalous regularity. There is too much an unexpected regular systems. And so you see in the image here, it's very clear which of the trees were a person, which of the trees grew on their own. Artificiology is then in this way in which
agency can recognize itself within the system, that it recognizes itself because it recognizes anomalous regularity. And I think this is the way both it's, let's say it's recognition and this importance, that it's part of the ways in which what we call anthropogenic climate change, which is essentially a kind of artificialization of the climate, was something that is not necessarily intuitive. You can go out into the forest and contemplate as long as you like, but the planetary scale effects of climate change are going to be found intuitively in that phenomenological experience.
It requires the kind of dissociative abstraction that the infrastructures of earth sciences and climate sciences provided to us, such that we will determine and perceive forms of anomalous regularity in the climatic systems, that we were therefore able to recognize our agency, our anthropogenic agency. And so in this sense, the agency actually preceded the recognition of the subjectivity that may have brought it about. Agency came first in this regard. Now, another point of distinction before I close with some of the questions on language. Another way of considering the question of artificial comes from Herb Simon, the old school symbolic AI.
a fellow from the 60s who made, I still think, a very important distinction between what he called the artificial and the synthetic, or one that may have practical importance in a certain way. So Simon argued that artificial diamond, that the artificial is something that is fake, but it has a substantive correspondence to what it is pretending to be. It superficially resembles that thing, but it intrinsically is not that thing. So a piece of glass carved to be a diamond. Molecularly, atomically, it has nothing to do with a diamond. It sort of superficially looks like one. That's artificial for Simon. The static, on the other hand, is something that is human-made,
but fundamentally is a kind of the thing that it resembles. In other words, a lab-grown diamond, which is human-made, but is atomic but is clearly identical to the diamond that you might find in the ground it would be a synthetic diamond it is it is genuine it is real it is just human made and so then the implication then is that there's an important distinction to be considered between artificial intelligence in the simonian model which would be intelligence that looks and behaves intelligence but actually isn't that is invites attribution errors in both directions that we see intelligence It's not really there and we don't see it when it is. Versus synthetic intelligence, kind of human-made machine,
probably machine, but could be machine in biology, form of intelligence that is genuinely intelligent in some way, even if its attribution errors work in a different direction, where it's more likely that we have difficult time recognizing it because it doesn't accede to the anthropomorphic presumptions that we've become. So to complicate this and maybe give us a sort of like example for us to work with, we'll talk about the question of artificial languages and natural language processing. And maybe to the question, to what extent are artificial languages artificial in the Simonian sense and to what extent are they synthetic? That they are a kind of synthetic linguistics. So an example of artificial language in that Simonian sense, a language that sort of seems like it would be, as a cause example, Joseph Eisenbahn's chatbot from Eliza from 1966, which you can sit down in front of, and it's the Rogerian psychoanalyst.
You will basically take everything you say and turn it back to you in the form of a coin without really understanding what you're talking about, which may be that this is certainly something humans do. but in this case it's a simple script it just it's just a script if the person says this then take these nouns and verbs and adjectives and flip them in this order and then derogative in this way it's a simple formula the whole program is about two pages but what as you probably know some of Wiesenbaum's lab uh sort of lab employees uh became quite enamored with ELISA and poured their hard outs to it. But this was an action in the positive, that they were identifying a form of intelligence that really wasn't there. This is artificial intelligence in the
Simonian sense. Now, one of the criteria, other than the Turing test, that those who have long been working in relationship between natural language processing and artificial intelligence, which clearly after the transformer models such as GPT-3 have sort of changed it. They're called the Winograd Schema, named after Terry Winograd, the famous human computer action designer, engineer, Sergey and Larry's mentor at Stanford and so forth. What Winograd Schema are, as you see here, if I were to say to you the trophy doesn't fit into the bronze suitcase, you know immediately that the it is refers to the trophy because you know what the size of brown suitcases and trophies from your experiences in the world but a program like
ELISA would have no idea um it doesn't have any semantic understanding of the words manipulating one of the things that I think poses the question as to what point has the artificial languages flipped in something like a synthetic language has to do with the fact that many of the larger models now the next versions of these large nlp models will surpass trillions of parameters uh gpt3 is in measured in the hundreds in in in order of magnitude smaller not only their ability to parse schema but to uh respond to relatively complex uh in what it responds to prompts that are clearly respond to the prompts in ways that are more comp that are more semantically structurally complex that prompts itself. Now, I'm not suggesting that the large language model
is conscious in the sense to which animals are conscious. I am suggesting and posing the question is perhaps we have the gradient has flipped to some degree that we are now in a world of synthetic intelligence more than artificial intelligence. Another example of I would visit to the Sony's AI lab in Tokyo years ago, there was a point where a lot of roboticists, and again, we're back to Brooks and some of the physical models of problem solving, taught this robot, that both these robots are language such that this robot could say these terms and to tell this that it should pick up one of these yellow pylons and put it on top of this thing here. Now, what happened, it was a little
bit similar to the more famous Facebook example, is that after a few days, the researchers had no idea what these things were talking to each other, that the evolution of language accelerated to such a degree that it worked, that these guys could talk to each other fine, way in which the language had evolved to produce markers and signifiers and something like a grammar to explore this space of something like an information exchange, it took life of its own. The evolution sort of acceded. So one of the things that we're kind of interested in is natural language gets built into more things in the world, just appliances, but objects, you know, what more colloquially think of Internet of Things,
things of scales, not just gizmos. and to the extent to which NLP becomes not only appropriate interface by which humans would refer to and command these things like Mickey Mouse and Fantasia, but indeed how these things would talk to themselves in relationship to us. And the question that is posed is that it's perhaps in the shift from an artificial to a genuine synthetic language that it's when the majority of speakers of English are not only are non-human Chinese or any of the other or language, whatever language, whatever strange version of English and Japanese mixed together. there will be a similar kind of acceleration in the evolution of language in such a way that the
forms of language and that is the very peculiar and beautiful form of biosemiotics, this human language, will become one among other forms in a larger cacophony of signification. There's other things to say about very large and small models and why I would make the argument that large models are actually are better as foundational systems for this. A little bit of an argument against the stochastic parrots conclusion, but maybe these two will be better discussion. So I'm going to stop the sharing from there and hopefully put enough on the table for us to kick the wall around for a while. Thank you so much, Benjamin.
Many, many thanks. I mean, there's a lot of research to explore. I try to kind of like break down the topics that you brought up. So we kind of go and talk about them in more details. So the first one, the question of scale. So you made two sorts of, you know, kind of scale sensitive questions with regard to, you know, planetary sapiens.
One was deep time scale in terms of Darwinian evolution. time-sensitive scales, and the other one was kind of like a Copernican scale that you brought it up in terms of like looking at the picture of the black hole by way of different sense, so on and so forth. for many of us obviously we can kind of grasp the question of the scales right from the time of Darwin
to Thomas Harvey in the pair of blue eyes when the protagonist was hanging from a cliff sees himself eye to eye, face to face, to a fossil. Its eyes turn into a stone and he starts to contemplate he's experiencing the mean times of the past, meaning the fields of deep time. But the thing with the Copernican, post-Copernican time scales, we are actually not talking about Darwinian deep time scale.
We are talking about the discovery and invention of apparatuses, theoretical or otherwise, for us to be able to take what you might call to be conceptual pictures of something. Right? Yes. It would be a planet. it would be a microbe, it would be a molecule of gas in an isolated system, a lot of balls. So, essentially, Copernican scale sensitivity has everything to do with at least two trajectories of how we can detect scales.
One, evolution of techniques in the broadest possible technique in the platonic sense, and then move it to, I don't know, Hume, Descartes, Kant, and so on and so forth. Another one would be historical evolution of scientific theories and what scientific theory, well, scientific theories at their core, they are constituted by a model, a model that is comprised a structure, and this structure is being informed by a certain sort of theoretical framework of the time.
For example, relativity, Keplerian celestial body movements, Newtonian gravity, and so forth. You can go on on this. And of course, this instruction needs this theoretically inspired or influenced structure ought to be constrained by certain sort of fidelity criteria, such that when we are talking about a model, this model does in fact correspond to a certain sort of a slice or sector of reality we are trying to picture or model.
And such constraints in modeling can be all sorts of stuff. resolution in terms of at what scale we are trying to take a picture of a phenomenon, such as movements of molecules of gas, or the dynamicity, you know, for dynamic systems, how can we actually take a system that is already in the process of motion, right? perturbation, so to speak, and all sorts of other stuff. So I think that I can see that when we are talking about planetary sapiens,
we should take into account the Darwinian story of the deep time scales, right? Even David Icke or all the conspiracy theories take that to account, right? From lizards to the sapiens, from sea monkeys to human beings, right? That's part of St. Nalus' conspiracy theory in terms of thinking about time skills, which is basically a Darwinian take on this sort of stuff, even though with a twist. in the conspiracy theorists. Well, with Copernican scale, we are on a different sort of modeling of scales.
As I said, we are somehow beholden to our current scientific theories, which inform the sort of models that we can make out of them, and two, the sort of fidelity criterias through which we can attribute or allocate thus and so model to such and such target system in reality, a phenomenon, so to speak. In that sense, my question for the first time of this scale sensitivity is that, to you at the very least, I know that you would say, well, both of the scale sensitivity problems are important.
and time, post-Copernican scaling systems are always beholden to theoretically constrained models. You would say that both important. But I really am interested, honestly, that for planetary sapien, At the very least for humans, by humans I mean homo sapiens, humans simply as Kantian agents, you know, agents of certain sort of cognitions, critical or practical.
which one of these two scale sensitivity time scales or theoretical Pernican theoretical models gauging the scales which of them are more important for this project for us to picture planetary sapiens good question well you're right They are to the extent to which we will hold them at arm's length from one another in this way. We want to make a model of the models such that they are, that we would differentiate them in this way. You're right. I think the answer that I will, you know, will want to come back with is obviously that they're in a way nested within each other for at least in terms of, at least in terms of this project.
and I'll try to honestly explain why I think that's the case and what but first let me sort of highlight I think some of the important points of concurrence with what you author here in terms of the conditions of recognition of deep time, let's say, or of Copernican scale. Obviously, the reality these terms refer to, the billions of years of the geotrauma of the Earth and the billions of years
of primate evolution, were processes that were ongoing long before this very, very recent phenomenon by which some form of those Kantian agents, as you described them, became partially aware of the conditions by which that agency is even possible in the first. And so part of the question, I think, is to the question of the project of planetary sapience going forward and which of these models are more important. I think that has to be asked in the question of which were these models, which of these conditions was more important for the emergence of the sapience that is possible to even ask this question. And clearly
it's both. I mean, in a way that you can think of it as a kind of nested condition of an astronomical condition of physics and the physical and chemical processes that make Darwinian possible, unless you want to argue, as some do, that even physics and chemistry are, in fact, Darwinian processes, though not necessarily need to go there. But that the condition by which it becomes possible to produce these kinds of models, produce the kinds of abstractions through which the possibility of self-recognition becomes possible, that it becomes able to recognize its own preconditions.
Clearly, we need to locate them in both. But they are always, and I think you've indicated to this, are always partial, always growing, that the mechanisms of technical abstraction that allow for Galilean, allow for heliocentrism as an astronomic model to be possible, require not only the technical abstraction and the technical extension of thought to allow for those geometric and mathematical abstractions to take place, but they require a kind of quieting of intuition in the phenomenological sense,
an overcoming of phenomenology in a way overcoming of the post-surly um uh intuition in order for them to order for them to arrive at this but that but that really the capacity of us to comprehend something like deep time is limited by just the anatomical uh you know the anatomical load that it's possible for us to consider, which is probably in many ways why the question of AI becomes important here, to the extent to which it represents a different kind of substrate for abstraction that is able to do pattern recognition and model making and a different kind of
mechanisms that would allow us, like the telescope, like the microscope, would allow for a kind of an understanding a perception of time perceptions of space that otherwise would have been otherwise would have been impossible in that in this regard so let me sort of try to from this as well I think one of the other questions around this question of apparatus and scale has to do with the the ways in which metrical systems are generative here and you know I think something like Eames powers of 10 as a way of the ways in which we teach Copernican scale to children um that that there's an anthropo that the center is intrinsically anthropocentric and that is a
quant and that there is a step gradient of of orders of magnitude below above us and that these step gradients represent sort of layers in an onion to be that there that have a kind of an intrinsic but interrelated quality to them way or another is part of the paradox then by which on the one it's only through the application and deployment of the technical systems of alienation abstraction is the possibility of of of Darwinian scale deep time scale or Capernaum become possible to think at the same time those mechanisms become generative of the units through which that thinking takes place and it becomes just as limited through that, through that construction. We're sort of a well-known, well-known phenomenon.
Yes. Superb. Superb. Okay. So now there's other things to sort of sit around the, the modeling here that I think is, I mean, now there's another point, I know the question, I mean, one of the forms of models that I think is interesting. Maybe I should actually ask you another question before, before, because I want to actually, the first question was supposed to bait you, but you weren't baiting the trap, right? Well, I mean, I was going to say a little bit more why I think the Darwinian scale increasingly in terms of the ways in which life itself is understood in terms of what you're referring to as Copernican scales and the ways in which the sub-shatter that the Copernican scales might have been traditionally held
as their own is increasingly understood through Darwinian models. is a way in which we may understand these models beginning to bounce against each other in interesting and productive ways. Yes, yes. So if you can entertain me, the question I wanted to ask, and this is really the genuine question that I really wanted to ask. So obviously, Kant and Descartes and Hume came before Darwin, right? a fully, full-fledged, you know, idea of agency, of rational agency. They didn't need to have Darwin to basically say what an agency is.
So this brings us to two sorts of stuff, sorts of trajectories, historical trajectories that, well, yes, there is such a thing as a rational agency. It doesn't need to wait for Darwin or artificial intelligence, even though there are some implicated in the condition of possibility of the agency with regard to the works of Descartes, Hume, Locke, even, and Kant, and Hegel, a condition of possibility that might implicate something like a theory of Darwin. but it doesn't need to wait for theory of Darwin or AI to emerge to talk about agency, right? We can talk about rational agency.
However, there is a different sort of scenario which we can actually talk about, and that would be the materialist take on the question of the agency, right? Which extends from Darwin to Helmholtz, Robert Meyer, and Freud, right? where basically we are forced to talk about the agency and cite objects and contents and structures and functions in terms of certain sort of energetic principles
that are basically being held across the board as universal principles. You know, the first law of thermodynamics, energy cannot be created ex nihilo, nor can it be lost, accident. And we see it in Darwin, we see it in Robert Meyer, we see it in Helmholtz, and we see it in Freud. And I think today's discourse about AI, it seems that it's a certain sort of unnecessary sentimental clash within the materialist wing of the condition of possibilities of agencies, rationalist conditions of possibility of agency.
to the point that it becomes almost a certain sort of what you might call to be a soup opera of the agency, right? Between this sort of extreme bipolarization between the material tendencies in terms of conditions of possibility of the agency and the rationalist, a periodist conditions of possibility of agency, a lot of Kant and Hegel and tradition of philosophy of mind. And I want to know, ultimately, are you Switzerland in terms of this clash
within the materialist camp and the rationalist camp of conditions of possibilities? or you somehow take side with one of these more than the other? Well, okay. So, I mean, once more, my answer would be a little bit is to challenge the opposition here as well. I don't disagree with the soap opera condition, but I don't know that that's necessarily the map of what the intellectual stakes really are here. I mean, I don't disagree with what you're suggesting here, that there is a way in which, you know, there's not only Kant and Hume, there's, you know, there's Nets and Hegel in terms of trying to understand in a way
that is through different kinds of way to prioritize the conditions for rational agency, the conditions as a condition of mind. And I mean, to your point, I think they're also, each of them, including Freud, entertained the problem of irrational agency as perhaps even more fundamental to their theories of mind, to the question of how is it that rational agency is possible given the agency of irrationality. Yes, and he also talks about unconscious inferences. Inferences are supposed to be conscious, because otherwise it wouldn't be, right?
But Freud has this idea that there are certain sort of things as unconscious inferences. yeah and and and i think that and i i take freud to be an extraordinarily beautiful gothic literature of the mind um at the same at the same at the same and valuable in this regard but at this you know there we do have a certain sense of like fortunately over in the same way that certain romantic ideas of nature had given way to earth sciences and so forth, that certain, I think, durably instructive models
of how to think about thinking that to be derived already in psychoanalysis can now be located within, as Ray and others do, can be located within new shit that's come to light through neuroscience. And I don't think that this needs to be in kind of opposition. I would say, like, I mean, I think that you picked up on in terms of the ways in which I'm trying to locate what you're claiming the Darwinian model and the Copernican model in relationship to each other. that the thing that I think that you think is that the is that uh what Darwin demonstrated and you know I think you may be picking up on this line I had said earlier that I think Darwin was the most important philosopher of the 19th century and I still think that's true um is that uh is that
he is that we now have that rational itself now understands the in essence the the conditions by which rational agency itself is possible for it to emerge in the world and in the planetary context. In the substrates configuration. Yes, the fish discovers water. What would be sufficient substrates for an agency to actually even begin to arise? That's right. And no, I don't, no, I don't, I will not, I completely, I totally follow you. And I'm with the project of understanding that, that it is this minimal conditions of substrate for trying to understand what would be, as you, as you put it, the general conditions within a general intelligence, artificial or not.
such that we might hypothesize, even in astrobiological, some kind of convergent geometries of intelligence, of induction and deduction and so forth, and the capacity for recursion and so forth, that may exist in a proliferity of different substrates. And that this takes us back in a way to some of those questions about the a priori conditions of rationality that are equally dependent upon that, but in a way, because they have this geometric similarities or nevertheless have other kinds of intrinsic properties to themselves. Again, I don't think he's necessarily autonomous to one another. I think it's part of how this process works. I mean, I'm on the energetic side here as well,
and maybe this goes a little bit to the AI, a little bit to the where... That is absolutely... I mean, the energetic story is absolutely right on. It is. It is. But also there's the cybernetic one as well. This is all, you know, I mean, you know, if you follow, and I'm sure that you do, some of the more interesting work in the more recent, you know, by recent, I mean, last 10 years in philosophy of biology, information theories of life has sort of come back again. Yes. that are quite interesting and exciting, particularly around phenomenon in natural biological systems of the dynamics between convergence and divergence, of convergent evolution and divergent evolution.
And so for me, that actually is kind of like... The revival of biophysics. The revival of biophysics, but also, yes, the structures of genetic information, of sort of information complexity at molecular level, protein, all of this, that there is an energetic model to be pursued here. There is also, I think, an information model to be pursued here that is material in its way. But anyway, let me kick it back to you. So I kind of was rather surprised that you have such a high esteem for Freud. I had no idea about this. I grew up on this
and so you either the literature that you grew up on you either you know you find a way to escape it or come back to it and repair it I guess for me I either find yeah there's three options one you find a way to repair it and it's like fixing a bad relationship one that you come to terms with it and allow it to you know be a kind of fond uh fond eccentric uh friend which is probably the way i feel at freud growing in the dark or for one or the other you you absolutely want to kill it uh that you you you you hate it because you hate the person it made you it turned you into before and you want
to kill that person and therefore you want to kill that of course for me freud for me freud is like you know uh yeah it's like it's it's it's as i say it's a beautiful gothic literature it's a much more beautiful gothic literature than say agamben um who i who who i you know as you as you know i i find to be deep pernicious i i i think i think freud is you know it's it's it's it's beautiful but i don't but but you know it is what it is yes i mean uh i i actually want something now that Freud came up that actually Freud is far more closer to your vision of AI
than people like you know hard d'actionists or hard naturalists like Dan Danette or AI denialists like Herbert Dreyfus and you know those sorts of people. So why is that Freud is actually closer to our vision of AI? And I think that we do share the same sort of vision of AI and the conditions of possibility. Well, precisely because Freud is obviously coming from the legacy of Kant and natural philosophy. but also his coming from the legacy of Meyer and Hengholtz, energetics.
Back then, it wasn't called energetic. It was called economics. Economics energy, right? Economics. So arguments about energy are economic ones. Yes, and at that time is also the importance of this, as you said, natural philosophy. But I mean, without romanticism, with a capital R, romanticism, Freud doesn't occur. Freud is an extension of that. It's a way in which I think... His early works are kind of romanticists, but he actually tries to overcome that sort of romanticism, earlier works. And what happens here is that Freud's energetics,
no Kantianism, creates a perfect, what you might call to be, positive thesis of AI. What is this positive thesis? like intelligence can be implemented within a continuum of inorganic and organic inanimate and animate but only localized for as every sort of implementation of intelligence requires also explanations for the local conditions of possibility. But insofar as intelligence is a very slippery sort of concept,
it can be implemented within a silicon, within any sort of substrate, you know, kind of like back to Darwin, or in organic materials. But for this intelligence to have thus and so characteristics, we ought to explain the local conditions of possibility of the emergence of this specific of intelligence as well. So the question of intelligence for Freud is that it is not about that intelligence can only happen in life, in vitality, right?
It can happen in an organic realm, inanimate. but then you are forced to also explain what are the local condition possibility for something having the characteristics of an intelligence is sort of expanse of continuum between inorganic and organic and that's actually to me is a very healthy understanding of AI yes i i i do too i it's i mean you realize that it's a an idiosyncratic somewhat idiosyncratic way of uh reading a freud but i that if i i one that i
you know would i don't i i and it's probably similar to what the aspects that i we want to bring forward it's a kind of what i'm hearing you is is it's the kind of it's a and obviously lots of people have written about this, but it's kind of Freud as a proto-cybernetics, a proto-cybernetic theorist in terms of his model of mind, the early and later models of mind. Yes, absolutely so. And that's because of Helmholtz, Info Helmholtz and Meyer. The dynamics of totem of this is a sort of an early version of extended cognition. And protodynamic systems, protodynamic system and complexity. That's right. That's right. That's right. And in that sense, you could see that there's a way in which Freud is a way in which that romanticism's horror with the denaturalization of nature came to find peace with what it had discovered in a way.
I mean, not what you mentioned, though, in terms of this agency of the anatomic, that's essential, obviously, to Freud's theory of drives was Thanatos, which people tend, you know, and also religion. instead of just being merely a kind of self-destructive drive, it's a kind of drive towards a drive towards cyberneticization, a drive towards dissolving the boundary between the biological and non-violent. Denaturalization. It's a drive to our denaturalization, ultimately. And boundary dissolution. I mean, my point is also the boundary, this sort of foldback into the matter of the world.
Yes, absolutely. between the living and the non-living, which he identifies as a primordial source, which has been sort of derogatorily referred to as just death. But that's not the point. No, that sort of interpretation is very Schopenhauerian. But I mean, usually in all translation of Freud, usually drive is being translated as instinct, like death instinct, right? But drive is not actually instinct on the side of life, generation, gestation, and all sorts of vital forces. And on the other hand, drive rather than instinct.
And drive is a strive to denaturalize instinct. It doesn't have a conflict with it. It is like a complementary force that ultimately, what Freud and O'Brow or Frenzy have said, it reveals the unconscious instinct of the vital instinct. And with regard to the AI, we can say somehow this, if we want to talk as intuitively, that AI ultimately reveals the precariousness of vital intelligence.
it does yes uh the but also yes the precarity of vital intelligence to the precarity of vitalism as preconditioned for uh preconditioned for this um but it also in a way it also redirects it in other kinds of ways um which is maybe where you know there's also a bit of the interesting side effects i think over the last 50 years of ai research has been the growth in studies of animal cognition uh which are directly connected where you know we went from animals can't feel pain uh with the early you know with
cluster and heidegger to much more sophisticated understanding of what octopuses do and what ants do and what cuttlefish do and what other kinds of forms of, you know, what other kinds of living substrates of intelligence that are probably not Kantian. They don't rise to this. They are sentient. They're kind of like, they are sentient. They are sentient, but they are not probably cases sapient in the Solarzian sense of sapient, but there is intelligence there. You can, you know, one of my former PhD students. I completely agree. They are intelligent. They are intelligent. Yes. There's no hard distinction for some animals who we call them sentience
as opposed to sapience. I think that there are certain sorts of organisms, whether we like it or not, are nebulous realm. Yeah. I mean, we're in the nebulous realm in lots of cases, which is probably the lesson of Freud, as I take it, is that even in the moment's highest degree of self-reflexive mastery, there's still all kinds of deep, dark shit going on underneath the hood. uh so is um and that this is not you can't separate this from that process that this that the a notion of intelligence drawn from the Kantian tradition that would see it as a kind of
arid uh uh almost almost an arid and entirely metaphysical process the freudian the freudian legacy would want us to sort of move this back into those other kinds of other kinds of mechanisms but i think we're on the right track here and that and around this question of substrate and that is i mean this is where it links back a little bit with dreyfus's original complaint was that against people like simon who were arguing that purely that purely like heuristics was all you needed to do and that you can solve problem solving mathematic in mathematical equations and then it the the location of this in the world is irrelevant. I think one of the ways in which people like Rodney Brooks again, but also the
neural network, the bottom of the learning models, became more successful. But then also this attention to, say, astrobiology on the one hand, our discussion of animals and the the mineral of machine intelligence as a kind of converging vector, what this allows us to see is that the link between the range of possible conditions of substrate is quite diverse. Yes, it's a continuum. It's diverse, but it's regularized in the spectrum and the spectrum is to be understood Because in the understanding of that substrate spectrum, you also understand the phase-based possibility of the forms of natural and official sentience and sapience that can emerge from it.
And this is, once again, I think where the Turing test kind of flips out, which always was a very Freudian little vignette of looking at yourself in the mirror and you're not sure whether the thing you're talking to is you and blah, blah, blah. where that flips out into something that is much more generous and open about challenging us to what is the possibility of our recognition, of how would it be possible for us to even recognize intelligences that are difficult to recognize? How do we overcome those attributes like we did that we fall into with Eliza and Siri and things like this? and in this case
this is another way this becomes back to these questions of planetarity that we begin with so before asking you a question I just want to make one single intervention I don't think Kant was arrogant I didn't say I didn't say that. No, no, no. I know that. No, I'm saying that he wasn't in terms of understanding the whole idea of intelligence in the sense that Kant, one of his first essay is about and it's actually
there are two essays in that sort in 1600 to 1800 One is Christian Huygens, the inventor of first perfect lens and telescopes. And another one is Kant, who actually entertained the idea of possibility of alien intelligence. Yes, Saturn and Jupiter. Kant loves them. Yes. He loves his Saturn and Jupiter people. I would say I want to say that Kant's idea of intelligence isn't really not one to entertain the idea of sentience
he does not want to even though it is somehow implied in his transcendental philosophy it's not really his job to do that and there is a reason for it precisely because he's coming to crush Aristotelian philosophy of physics, life, ultimately, where basically sentience and sapience are being eliminated by virtue of final causes in Aristotelian system. And Kant completely wants to crush this by telling us an account that if we are talking about sapient intelligence,
we ought to have a certain sort of parameters. And otherwise, we won't be able to say that, for example, if we see an alien today, like Arthur Stark, run the movie drama, a piece of rock. the solaris problem yes we can't say that is it actually sapiens so I think the problem of recognition is really important here even though it's understated in the sense that strongly agree and misrecognition and misrecognition or abuse of recognition
in the stories that Nick Bostrom and his co are talking about AI. You know, at that point, what are we talking about? Are we really talking about intelligence or are we talking about speculative theological intelligence? Well, I'm not going to defend Bostrom on this as well. I don't know. I think this is just lots. How about David Rodin then? no i don't think that's i don't think that's some recognition misrecognition question here uh i mean the way in which i would want is is i think the question of recognition and self-recognition of here as well is deeply is is i agree with you kind of a core here and
but the problem is that it or the problem or the opportunity to consider is that it errors in both directions that is there's a tendency to not recognize forms of intelligence that are too divergent from one's own self-image. And again, you notice I'm following your point about different forms of the different kinds of anthropomorphic problems with AI, one of your pieces, that there's an inability to recognize intelligence that is genuinely different from the way in which humans are. But then there's also the tendency to not recognize intelligence that are different than the way that we think that we think. That is our own misrecognition.
Our own misrecognition of self is the primary problem here, which then we express and project onto everything else. Project is projected. Yes. I mean, the real problem of intelligence, sapience intelligence, recognizing its own status in the universe or world, of which it is a part, is more like the picture of Dorian Gray. It is not as if. You don't see any outside. But rather, you... And it's not a narcissistic image, where basically the narcissist looks into the river
and sees a kind of like a rippling image of his own self. The other is more like the picture of Dorian Gray, where you see a fundamentally distorted image of your own self. And you go on and see and reconstruct that distorted image. because essentially all constructions of intelligence, philosophically, if not technically, are based on this idea of recognition on the other side of the mirror. And the thing is that- Right, this is the problem. We usually see the picture of Durian Gray on the other side of the mirror.
Right, yes, I follow this. And this is, I mean, this is where this, this, this, this, we're converging here in some, in some really good ways. The other misrecognition then is, and this is why, you know, to Freud in the neuroscience and this is, you know, Ray's work, right. You know, is Ray's work is that it is in fact, the moments within the philosophical importance of the neuroscientific resolution was the fact that we don't recognize ourselves in this. that this is a model of thinking that we are already forming and always have performed that is deeply uncanny, that is divergent from the way in which we think that we think. And yet, and so the stereoscopic, as it were, has to be the bending of our own conception of thinking towards what we actually understand it to be.
I think where we would be as well, to following further as well, is that that's a challenge for AI, for the artificialization of intelligence as well. That we have to be able to, one, recognize that there are modes of natural intelligence that we don't recognize ourselves in that may just as well be artificialized as the one we do. and the range of artificial, this full spectrum intelligence that we identified, the spectrum of the substrate, the possibility for artificialization is across that spectrum, not just in this anthropomorphic center, but also that this condition then of recognition and misrecognition is, again, to just underscore the point, is then how the, this, the, what you call,
where the model, not only the model of thought, but the model agency becomes retrained, which is like part of in a way, I mean, we can talk about this if you want, where I, you know, like, like Bertrand Russell, like a lot of others, I think Kant got the Copernic implication of the Copernican trauma backwards in a way that it has more to do in a way of, of, of this bending of the sort of centering of perspective within the system rather than the recent. But this is perhaps the tangential point. One of the questions where I think the why NLP is such an interesting area of experimentation here is it's one, it's an area in which there is this incredible expression of capacity.
it's unclear to what extent this is actually correspondent to comprehension. And so this relationship between comprehension and capacity within these large systems is unavoidable. And it forces us to re-ask those same questions about not just AI, but about this. Scales. Go ahead. Scales. Go ahead. I mean, yes. For example, I mean, the last screenshot that you showed us, that it seems as if the AI is understanding and some person who is an insider might say, well, you know, here is an evidence that the AI understands this conversation.
But according to what criteria? You see, that is a whole point. Understanding comes with a scale And parameters And constraints It is not as if That's why I hate philosophy of mind Because people Use the word Consciousness in vain Like the word Christ What is consciousness? Really, would you be able to tell me At what level At what scale Under which parameters you are actually referring to something like consciousness. The same thing about understanding. No, I've read Philip K. Dick. I'm too far gone to even talk about it.
So another question I wanted to ask you, Benjamin. So you talked about simulation. Yeah. simulation kind of, you know, but I mean, obviously simulation in the ethical sense means that we are merely replicating the behavior of a system. then it's something called emulation and there is something called inaction where basically we are simulating not only the behavior of the system but also the behavior of the environment which
is in action and reaction with regard to the target system in that sense how much do you think that digital revolution can capture the idea of generalized simulation in the sense that a system is nothing unless and until we also take into account the environment it interacts. confrontation of actions and reaction between the system and the environment. How much do you think that a digital revolution can capture that sort of generalized simulation?
Because when we are talking about sapience, we are not talking about isolated systems. We are talking about individuals, right? And individuals are social beings. Yeah, that's right. they are. And so any, I mean, the only theories of mind are ones that are predicated on that reciprocity that anyone's own extensive mind is constructed in relationship to how you imagine others are imagining you. That this is part of whatever the sort of functions of mind have always emerged in those kinds of dynamic processes in which they're relevant. And so it's those recursive that could have been recursive dynamics that are the system in a way that needs to be captured.
So the answer to the question is to what extent can you have, I mean, this goes to the correspondence question that we asked at the beginning about models and the duty of models to be corresponded and the validity of models as a dependent on the degree of correspondence that they're actually able to make to the world. Of course, this correspondence, I mean, whether those are enacted or representation, I mean, this correspondence, I mean, I think you'd be the first to say there's also sort of dependence on level of abstraction that are functional. The function of the simulation is dependent on the level of abstraction, which you could think of as a kind of scale that's necessary to do so. So I can make a model simulation of heliocentric solar system with styrofoam balls and explain
it to a child. And this model, this simulation is functionally accurate in terms of its structure. But, you know, can you, you know, at what level of abstraction does this begin to break down? So there is no way. I mean, it's obvious that there is no way in which it's an infinite recursion problem. And my friends over on the other side of campus here who work on cellular simulation, which is one of our interesting areas now of simulating something going on inside a cell. well uh you know the problem that has been intractable with this is the one that you identify is that the cell is defined is entirely in relationship to its membrane uh and the system defined in which is interior to it uh and so all of the kind of majestic
metabolic and even thermodynamic processes that constitute the system city of a cell are usually dealt with as second order questions but increasingly all i mean look one of the things that you have that it's there's no doubt that in terms of simulations of material simulations of atomic simulations molecular simulations protein folding simulations with the new deep mind all the way up to climatic and earth science simulations that the capacity that the capacity to produce simulations as a level of granular and pivot and actability uh is has radically has radically accelerated. And to a certain extent, if we wanted to think of planetary sapiens in this regard,
you can think of it in a certain way in which it's become possible for the planet to make models of its own processes, to comprehend its own processes in some form of toy abstraction in ways in which were simply impossible to do before or were done at a level of ineptitude or sluggishness that would have been characterized in the early scientific revolution. This says nothing of the question of what such simulations are for, however, of what if we are able to create simulations of climate systems that we're able to replicate in real time what the entire Earth's climate was down to a square foot level.
uh you know it's computationally possible to do this the the question the question is essentially is in a way should something like this evolve and i would you would have to think about something like that in evolutionary terms that if you were that if there were to be a planet that was capable of modeling itself with that degree of precision not just to know how or something like that but rather to model itself in this kind of way, the question of to what condition is this adaptation can be understood as fit in some sense? Yes, of course. Such ones can be explained by at least the set of tools that we have, information theory
and complexity theory that we can actually solve. for example, the evolution of a cell, RNA, and so on and so forth. Right? And I think it's easy to take for granted. I think that we can do so in 2022. You can model cellular processes in ways just 20 years ago would have been science fiction. Absolutely. But my question here for you was that So, in regard to such modeling, how much, so you actually started to talk about digital versus analog, and you showed us a picture.
I want to ask, do you really think that digital revolution or digital modeling kind of constrain in a sense such simulations that if they are digital in some sectors of reality or so on and forth? or you think that... It's a level of granularity and it's a level of manipulating a nuclear explosion. But digital and analog are different sorts of worlds. No, precisely because in one sector we are talking about continuity in analog and in another one
we are talking about another sort of world where basically certain sorts of information are being deleted. Sure. and they can represent each other in different ways. There's ways in which digital simulations can represent analog processes and analog simulations can represent digital processes. And there's a degree, there's apps always lost. These are lossy processes. So you are not against, you are essentially what I want to ask. No, but let me, let me. So you are not some sort of Alex Galloway against the digital revolution on this sort of modeling. No, I mean, the way in which I explain it to Alex In terms of the relationship between the analog and the digital, I don't see it as a digital relationship. It's one or the other. In fact, the relationship between the analog and the digital is analog, not the digital.
I think one of the weird things that Alex, this sort of thing, is that they establish this relationship between the analog and digital as a digital relationship. That it's either one or the other. That it sort of flips. And I don't think we went this way. I think these are, yeah. Superb, superb. that's that's that's magnificent another question i have for you but at the same time i'm not at all i mean the argument the other version of this not to argue that somehow that that the forms that the that the formulation of continuous processes into discrete processes and the and the representation and emulation of those continuous processes as discrete simulations is somehow is is is lossless and in ways that are insignificant always insignificant it could be extraordinarily really significant, but I'm saying such things that models are always about abstraction,
and abstraction is always about, and that models, the function. Compression, compression, yes, compression. Well, abstraction, yes, I mean, I'll hold on to abstraction, but yes, and so it's not, I don't think you need to be some kind of absolutist about the capacity for absolute digitalification. I mean this really in a much more practical sense, to the extent to which you're able to produce models of increasingly mathematically complex systems behavior. It seems to be, you know, over the last 50 years has exploded our capacities in terms of model making, but it's been responded, it would seem, at the level of planetary governance with an equal diminution in our ability to actually use these models for the purposes of self-composition.
Yes. No, I'm very much on your side on this talk. In a sense, I would say that, look, you know, digital revolution by its nature, I mean, if we understand what digitalization is and digital computation means, and actually replicate all sorts of decimal systems in terms of, you know, one and zero, it gives us aggressive or progressivist modeling opportunity of the world.
And it would be very premature to say that, you know, by the very nature of digitalization or digital revolution, there are certain sectors of reality that we are, we can never talk about. Well, I would say that about this, we can translate this sort of statement to an epistemological statement. That there are sectors of reality that we can never know about. Well, Adolf Gurunbaum had a very great idea, trolling here about this sort of stuff. That, well, those sectors of reality you don't know about and you don't have a case of, maybe you don't even fucking talk about them.
You know? Maybe. I mean, I, you know, I've also, I'm also enough of a, you know, I've been, I've stood in Stanislaw's work enough to be comfortable with the sense that there are, that there's things that it's anatomically impossible for us to think, which are nevertheless part of the universe. Not even anatomically, we cannot even theorize them, we cannot even model them. We can't and perhaps we never will in the same way in which dogs are... Are they representing those cases? No, no, I mean, they're there, and I think we need to recognize just, you know, the ways in which... Yes, but... We'll never learn calculus. It's okay, but it's okay. They might be there.
They might be there. We might entertain them speculatively, and philosophers and physicists do that sort of stuff. But look, if there's no input from those cases in our current theories that create a crisis in our current theories or no empirical evidence, no theorization can be corroborated with evidences, then at best they remain in the realm of pure theoretical which is very respectable. so yeah it is it is but at worst they are in a speculative philosophy which is really bad at worst well it depends on what we mean by that um but yeah i mean look there's you know i don't know
there's people i don't mean by what i mean by this to say that you know we need to take things like mind reading and uh you know past things like this seriously just because we don't have evidence for something we have an idea for something i knew you i knew you were going to say that um is that you know i don't mean parapsychology or something like this i mean simply to say that there is that we have to kind of presume there are that part of part of what built what part of what makes up the physical universe are are processes that just as it would have been on just sort of unthinkable to 500 years ago or 5,000 years ago that will remain almost impossible for us to conceptualize,
perhaps forever possible for us weird little primates to conceptualize for a long time, but that are not remote and distant and unimportant, but in fact constitute the fabric of things in ways in which it's impossible. So that's all I mean. No, no, no, absolutely you are right. I mean, absolutely you are right. I'm also another kind of epistemological Promethean to say that, you know, part of the purpose of part of the key, if there is a telos to the emergence of sapience in this, on this, you know, that we are managed to live through somehow, it would be to build cognitive infrastructures that will allow us to recognize and form conceptual abstractions about things that would be otherwise impossible for us.
It's absolutely. No, no, no. So we're on the same, we're all sort of. No, I completely agree. I would say. Cyber-heg aliens enough for that, so. Sure, sure. No, no, I completely agree with what you said. The thing is that, however, when, for example, we say that, you know, we abduce in a Persean sense, Charles Sanders Peirce, about the hypothetical entity in the universe, we use it, we abduce, we entertain presence or the possibility of such entity according to the conceptual means that we have at the present.
Right? It is not as if we are merely positing something like supra nature like God or angels that will never be incorporated within the canon of scientific theorization. Right. So tell me what you don't like about the term speculative philosophy. No. Is it the Whitehead Association? that you mean sort of do or like how do you mean it? No, speaking of the philosophy, what I have in mind is kind of an age of philosophy
between Kant and no Kantians. Basically, people who were disgruntled with Kant began to basically forward certain sort of what you might call to be what Kant's ultimately called vagaries of thought. Meaning that you can entertain certain sort of entities, metaphysical or otherwise, without actually explaining why they should be posited as such. Yeah, okay. That's not what I would mean by it, but I hear you.
I think this sort of work is not to my taste either. Suppose the way the connotation that it would hold for me is more, I suppose maybe in some ways closer to the sort of the philosophy that is sort of generated to the ways in which epistemological technologies have transformed some fundamental understanding of worldly processes that requires conceptualization and abstraction and architecture. Contrafactuals. You see, contrafactuals.
We can always use counterfactuals and hypotheticals. Yes, I don't mean counterfactuals. I mean more like the ways in which Jennifer and all the other people came up with CRISPR a few years ago. Okay, there's a philosophical project. What are the concepts that are necessary to be generated in order to understand what it means to be alive, what it means to be a self-composing creature, what the implications of the manipulations of the deliberate artificial manipulation of the substrate of life and thought that is implied by this technology. As opposed to the generation then, the engineering of new forms of philosophical techniques in relationship to this, rather than the often equally appropriate projection of a philosophical heritage onto this phenomenon.
What would Heidegger say about CRISPR? What would Aristotle say about CRISPR? This is the projection model that I find that the speculative mode is sort of an inverse of. It's the generation of engineering of medical technologies in relationship. Actually, there are two trajectories of engineering, of artificialization in philosophy. one that has given us by arch idealists by way of to and hegel so what is and i want to get to the our last uh question we open it to uh question answers artificialization as a matter of becoming the object of a concept namely a form like that form or a form for plato concept for hegel right
And that's absolutely, it's a trajectory of philosophy. There's always philosophy in history of philosophy has evolved by virtue of that sort of engineering. And another way, which is a little bit more maverick, so to speak, is when we posit certain sort of counterfactual qualities. And essentially, we engage in world-making, a la Nelson Goodman. but world making not as this sort of play of positing world just because they need the world
need the united with ever more worlds no we can only power world according to some ingredients that we have essentially reconfigured these ingredients that we have inherited from previous world, delete some of them, deform them, translate them, supplement them with some other sort of ingredients, and then we pause this world. And then from that position of a new world, we look back on the sort of claims that we have been making in our old, meaning current world. And that is actually far more radical.
I think that AI, in that sense, is more on the side of world-making rather than artificialization in the sense that I just made, being the object of the concept. understood yeah i yeah i mean the ways in which you talk about when here is it also i mean it's quite close to i don't know if you've read lem's theory of science fiction but this the role of counterfactual comparative and normative in the way in which you describe it here is quite similar to to what he sees as the philosophical project of the counterfactual as well um i distinction of
this with the what you put as you put it the hegelian idea of the of the artist the object of the concept it's a it's a nice way of look it's a nice way of summarizing that particular understanding of it i would want to differentiate i think one of the ways in which i'm using from this is that there's a way in which artificialization is preempts the concept in important ways. In other words, you have agents that are remaking the world in ways that maybe... How can it be? Because the concept is already a artifact. So you wouldn't be able to actually create techniques for a certain sort of
target system if you have the concept of either the technique or of the system. No. Let me take the example of climate change, again, is a sort of case to say, well, anthropogenic climate change. The anthropogenicness is the artificiality of it. This is why we sort of mean this. That you had for 200, 250 years, 5,000 years, however you want to model it. But doesn't it simply mean man-made at that point, rather than being the object of the concept? It simply means man-made. It would be well, it would be something that has this no i would take it a step further than this yeah yeah in the sense that we have a certain sort of theory of self-conception and self-transformation we have done dozens of
things to the planet right what i'm saying is that the concept emerges from the artifact as much as they run if you think it this way that we were doing climate change was going on for however long period of time, but the agency necessary to transform a climate was ongoing long before the concept that this agency was doing ever emerged. The concept of climate change and the concept that anthropogenic age on this scale was even possible does not emerge until Vernansky and others in the early 20th century, and then really doesn't become a concept until the late 20th century, but the artificialization that was taking place long preceded it. And the important,
I think, function of the artificial here is that at the moment, particularly late 20th century, where we came to recognize, through the technical abstractions of science, to recognize the anomalously regular effects of this 80, and came to recognize its artificiality, that the concept of this, of climate change was possible to emerge. And more importantly, the concept of planetarity that was capable of this form of transformation emerged after the agency that caused this transformation already had a long-term career. I see, I see. A long-term career. And so it's the way- So would you be able to say- So in summary, it is the artificial is also,
or can be, I'm not denying the Galien model here, can be the way in which subjectivity recognizes its own agency. No, I think you're right. I think you are completely right. And can make concepts about an agency and its own subjectivity that would otherwise be impossible. Yes. No, no, I actually think you're right. Then at this point, perhaps we should actually make a distinction like Freud between the unconscious philosophy or concept of technique that it has been going on before the emergence of something like the conscious of agency, the unconscious philosophy of technique
before the emergence of any sort of anything remotely looking like Kantian agency. Or concept in Haysl's sense, right. Yes. And the conscious philosophy of technique that we are so, you and I are talking about, that we now know that, you know, techniques have consequences because our multidirectional sort of art in the medieval sense of crafts and so on and so forth, Matisse, that creates consequences. So there is the unconscious use of technique and the conscious use of technique.
And it seems that, yes, that history of climate change comes from the unconscious. Right. And then it becomes, that's right, that's right. And then it becomes, at a particular point, once again, through the forms of abstraction of processes that, again, these computational apparatuses have made possible. But it makes, I mean, I think this is quite different, actually, than, you know, some people say, oh, well, this you're saying is just Heidegger's revealing.
No, that it makes possible only this recognition, as I say, the recognition of anthropogenic agency and therefore of anthropogenic subjectivity in composition also produces, back to Hegel point once more just to say, produces the concept of planetary through which this subjectivity agency could cohere for the future and could, in the Goodman sense, build worlds through and around. I see, I see. So just last, I promise, this is the last question. So the unconscious technique versus the conscious technique, how much do you think that philosophers of technology actually, for one reason or another,
align the distinction between conscious technique and the conscious technique such that they become anti-technology or techno-optimist. Right. Because of that bipolarization. Right. So where there's, I mean, in terms of philosophy of technology, the capital P, capital T, I mean, I'm happy to say it as many times as necessary. I think that we've wasted a century with Heidegger on this topic. The whole field is monopolized by a Heideggerian and post-Heideggerian discourse that needs to be filled from the question.
that's having been said um as i sort of intimated like one of the uh perhaps one of the the um the virtues of that tradition was that it was willing to talk about something like the unconscious agency here but in a way that was so negative that there was a way in which the fact that tools use you the fact that your thought comes to be formed through equipment and and so on yeah for them for them this is for obviously for them this is a kind of you know um a kind of fallen state uh in in a way in a way or well as a kind of um and so the idea that there you know that the being has you know has sort of relations to the world that are um that have a
certain authenticity that then becomes manipulated by technology so obviously not what i'm not what i'm what i'm suggesting at all but i i think i think there is i think what we're talking about here much more than just unintended consequences in lang and winner's sense and speak philosophy of technology of unintended consequences that any complex system produces greater complexity in its exhaust or its way that there is no ultimately conscious fully conscious technique where you somehow can count the secondary and tertiary effects of a technical of a technical agency. I mean, as I said, that's well understood even in the Langdon Winner sort of discourse here as well. What I'm suggesting is like there's actually an important lesson to be drawn from this, is that the subjectivity that produces concepts is doing so in relationship to its
recognition of, perhaps misrecognition, but recognition of the agential effects of its industry over time. And that in the recognition of anomalous regularity, it's not only able to produce, to recognize agency, but again, to recognize subjectivity and to produce the around it in a way as well. And I'm not suggesting this as a fallen state, not suggesting this as a sort of inauthentic state. I'm suggesting this not only even at a visual conscious agent wielding a tool in the, and, you know, again, in the, in the, in our kind of Heideggerian hangover, but rather technology part of the technogenesis as part of the ongoing co-evolving process of anthropogenesis as well,
that the anthropogenesis and it's our species level capacity for conceptualization is emerging in relationship to how we see the effect, how it is that the technogenesis is producing this artificiality. So a lot to unpack here, but... Yes, absolutely a magnificent response. On the techno-optimist side, yes. I think part of the techno-optimist side, we could define as the naive presumption that agency is conscious, that technical agency is conscious and consciously aware of effects, that effects can be modeled, captured, predicted, and somehow contained. At whim of every society or civilization, red law.
I mean, this is where the kinds of, I think increasingly tendentious and expansive evolutionism are totally correct. The idea that we'll define a problem in a very simple engineering sort of way and we'll build a widget to address this particular problem and we're done. And I think this is another form of logical misrecognition. Superb. Thank you so much, Benjamin. Thank you so much. Sorry to interrupt. I was just going to point out that we are in 45 through. We were discussing having this first be maybe about two hours long.
So I was going to ask two things. if either of you had any remarks in particular that you wanted to bring this section to a close with. But then also, I guess what you want to do about the question and answer session, do you want to have a quick break after you have your closing remarks? No, I think we are fine. I'm fine. I'm fine. Yeah, I'm not sure I have closing. I'm not sure I have. I don't have any closing remarks. I think that Benjamin closed my own remarks. No, I mean, I think I'm sure more will come to mind when we're going through the transcript. I'll come up with something kind of brilliant. Absolutely majestic.
Absolutely superb. Majestic and glorious ending after the fact. Absolutely. I mean, you can add these things in as and when. But it's been a fascinating conversation so far. I wanted to thank you so much for, you know, it's taken some unexpected twists. I didn't expect that Freud would be a significant part of any section of this book, but I'm quite glad that you went there. So that's very good. So, I mean, yeah, well, let's roll on to the Q&A in that case. I mean, I've got one or two questions from people who couldn't keep live for one reason or another.
But yeah, anyone else who's already on the table, if you have a think about what you'd like to ask if you have something. And just either put your hand up or just ask it. But let me begin with one that was dropped into the Discord earlier. and that is, if I could summarise it, actually a question that was asked before this particular talk began and it is calling back to the previous conversation that you did with Nick Cernacek, Reza, and the question is something like this. So it's, yeah.
So he says, last conversation Reza talked about some fallacies of contemporary AI research. And my question is, if you're in a place where there is good AI research, which escapes these fallacies. I don't necessarily remember which fallacies that's in reference to, but maybe you do, Reza. Yeah, I remember that Nick asked me fallacies. I said that, well, you know, the first fallacy is kind of Turingian orthodoxy, you know, kind of syntax semantics, the whole model of computation, a lot of church and Turing. If I remember correctly, the second, they're not fallacies, they're dogmas, ideological dogmas, orthodoxies.
fallacies by any means. They're cultural norms. I would even say that part of the work that we've been doing on this history of AI in China project, it's very... Yes, the orthodoxies are being entrenched. That's right. And have been in many ways. So the second one, yes, exported, absolutely. So the first one was church theory in orthodoxy. The second one, if I remember it, was the hard naturalism. You know, the idea that why AI is inevitable proposed
by people like Dan Dene, so on and so forth. And the third one, I can't remember the third one. Yeah. No, I think that I are still working with this kind of orthodoxies, as Benjamin just said it. The thing is that these orthodoxies have been so entrenched in practical researches, that they can easily be exported across different sorts of cultures. So, you know, Benjamin introduced China on AI.
I had no idea about Chinese research on AI. Benjamin put me to read some texts about the history of Chinese AI and where it's coming from. and I actually think that some of these orthodoxies have been exported to the Chinese. And that's what I want to call conditions of possibilities of AI, contemporary AI. I don't feel comfortable to call them conditions of possibilities
of capitalists they are. Precisely because they don't have anything with capitalism. At least these three sort of things that I said, these dogmas or orthodoxies, because they are so theoretical and so realm of orthodoxy that they can be exported across the board. I mean, I agree with everything I think it makes more specific that there's a long history of discourse around AI in China that sees it, sees AI as something that is more about the augmentation and coordination of collective
intelligence, not so surprising, rather than the mirroring and reflection of individual experience of mind. That the tradition to understand is something to where a group of people, an initiative can be facilitated, can be coordinated in a way that is greater than the sum of its parts. AI is understood in logistic terms, on the one hand, is part of the tradition. the other is a bit different this has to do with a kind of moral which is again has to do with more sort of a moral AI relationship to moral philosophy
and how it train and be trained by different kinds of traditions of proper behavior and to sort of structure things according to these terms it's a much longer discussion that i don't want to you know to actually sort of treat probably you would have to go sort of much deeper into but suffice to say like i quite agree with with one of the sort of cul-de-sacs that i think the religious image at a particular time has to do with the fact that it that the question of intelligence itself and what intelligence is and how it needs to be artificialized begins with a rather parochial sense of what intelligence
is that comes from a particular philosophical tradition or traditions, a convergence of these, sort of a way an understanding of AI is simply a kind of acceleration of implications of these traditions, whereas if one were to sort of understand the geomateriality of intelligence, and the humanitarian emergence of intelligence and all its cacophonous forms, and to suggest that any of these forms could be artificialized in different sorts, and that any of these could be artificialized and located within any of the other traditions of coordination of enactment of cognition and incursion that exist
and that form the basis of human cultures, the range of possibilities for what AI is and what it could be and how it could work is much broader than what we've arrived at in the form of the dumb subservient teddy bear. Dumb subservient teddy bear. yeah that we in all of i mean that with all the animism intel you know we sort of acting we name it we project consciousness on it we project intentionality and emotions onto the stuffed animal that all of the ways in which we you know that we would try to recognize the
intelligence that is there we do so we force it through this mechanism of a kind of anamorphism It's like in Disney movies where they talk like American teenagers, that the extent to which their intelligence is recognizable, it's to the extent to which we ventriloquize them. Yes, yes. I agree. This is where we're at. yeah or mortal equalism you know voice to the dead right this is actually really I remember that this is a chapter in my book intelligent spirit that
if we go on and merely projecting our our anticipations in a dynamic way and our, you know, sort of kind of wishful thinking about what intelligence is onto the world without giving ourselves a critique of intelligence. And that requires a philosophy of intelligence, of which we are lacking. then all we create it is mirror mirror on the wall queen merely sees uh her picture in the mirror every time and she's aging and she's not a snow
white yeah that's right and so that's unfortunately that's what happens when the turing turing test becomes a necessary condition rather than a sufficient condition yes i suppose yes I suppose the optimism that I would have to regard is a little bit like this discussion we had at the end about the unintentional agency at work and then subsequent historical moments where some pattern deduced and this anomalous regularity is recognized. And, you know, the fact that something is recognized long after it's actually been created may also be taking place. That we may actually be generating real synthetic intelligence in ways that we don't really understand yet.
Beyond our minds and machinations. and that it may be that in a generation or so, we might be looking back at some of the foundations that were laid with foundational models and perhaps other kinds of things. Suggested like actually what we built was genuine synthetic intelligence, even though we did it by accident. That's what Chinese people tell themselves. That's Luke. We have been making this according to our projection but what if all along we have been creating something else I think that's bullshit you do that's Chinese Republic of
China people in the Republic of China making you know slogans of their own AI culture I think that AI culture is very I think you might have missed what I said, but go ahead. We have talked about some of the contribution in the book. And I genuinely think that some of the theses that they bring up are not, you know, better than the run-of-the-mill Western techno-optimist of AI, unfortunately.
Yes, in many cases it is. And I think this is probably a lot of the things that we're including in the book are included in a way that's meant to be more symptomatic rather than a little bit of a kind of stated list. You probably don't want to read my text. I'm almost done with that. No, no, I do. It's very bad, very critical. That's fine. No, no, no. The fact that so much of the discourse is actually this sort of Western frameworks is part of what we've identified throughout the whole thing. So no, the fact that there's a piece that's sort of critical of the other ones is not a problem at all. In fact, what we're writing is a lot of, as I said, we're sort of writing a little intercede to the piece and try to stitch these together in a little bit of this fabric.
Superb. No, no, there's, there's, I can think of three or four pieces that we are including more as symptom than anything else. For example, the way in which the, in the inscrutable history of how AI thinks gets so easily mapped onto sort of originalist ideas of the inscrutable Asia and the weird, mysterious stuff. and AI's spiritual core can be understood if we locate it in this spiritual Eastern context. I can't tell you statements I've read through that basically do this maneuver. This is Neo-Confucianism. Neo-Confucianism.
And so we're including one of these as an example of the point. I remember Nick I don't know whether he still that blog Outside In wrote about a philosopher that I actually quite admire Mo Zungs and and he's going on and trying to of introduce him as the father of Chinese AI. Well, my dear friend, it just doesn't pan out. Well, who are you saying, who are you saying, dear friend, to me?
No, dear friend, I mean Nick Lant. Oh, I see. It doesn't pan out. No, I mean, he's not the only one who's made this connection. Of course. It's a sort of expat sort of thing. No, no, I mean, he figures his work as well as part of the sort of, you know, his reconstruction of a Chinese theory of technology. But I mean, as you know, in the book, we're starting a little bit later on. We're starting after World War II and around the Sino-Soviet split. We're not really into the, you know, it's all, it's sort of cybernetics on forward. Anyway, maybe we go to another question. Yes, yes, yes. Do we have any questions from our live audience?
I've got something. To Anna. I think one of them should actually say something. I have something semi-formed and queued up, But I'm happy to deliver to anyone else who's ready with something. OK, I'll go then. So, yeah, I suppose kind of sparked by your mention of techno-optimism, just think of a couple of tweets that I'd seen from Cathy Wood and someone else from ARK Invest. over the last week about how, let me just bring them up.
Apparently, over the last two months, the estimated time to artificial general intelligence has fallen by between 60% and 70%. And the confidence is that within six to 12 years, breakthroughs in AGI would accelerate growth in GDP from 3% to 5% per year to 30% to 50% per year. apparently new DNA will win. I wondered if you had any comments on this. I only have a trolling comment here and I give it to Benjamin. I would say that I have the same exact sentiment about cryptocurrency. Well, I think there's more to AI than cryptocurrency
in a certain way. No, I mean, I'm just trolling. I know. So Kathy, he's like, I mean, I mean, I think about Kathy, I mean, Kathy's also, was also a born Christian, right? Not only she wasn't this fun, but was also someone who is. Eschatological view of economics is, is, is something she holds dear. And the ways in which, I mean, the thing is, this thing is sort of interesting about arc and i think there's got there's there's a whole book to be written just about her funds i think about the ideology of this fund um and this worked worked well is that is is that it's always constantly couching every move in messianic
messianic terms that you can't just invest in something because they got a good quarter no this is the future of mankind. This is, the fate of the universe depends on this in some kind of book of Revelation. Not really a psionic, it's also mysterionism in the accurate system that something is being promised to that is fundamentally what you might call to be nominal or speculative at this point. The thing, I actually should apologize for my friends because I do think that cryptocurrency will do, will at some point expand because it's a certain sort of technology that is very modest. And
if it carries the mercy, it will expand, becomes the prominent technology. But AI at this point, precisely because of its own bipolarization, it's hard to predict what will happen to AI at this point. I guess maybe the other side of it, I'll take the short on that, that I think AI is probably a more, there's probably more a clear path to where this could, in the sense to which some of these here are, in a way, more purely technological than with crypto, where they are more social, cultural, symbolic,
which is much more complicated and unpredictable in terms of how it would all sort of play out. You think so? Yes. You think that AI is actually more straightforward than crypto? No. No, let me qualify as a point. I'm saying is that the path towards a more wide-scale adoption of AI and a path towards something that from a 2022 perspective we might recognize as AGI. Let me put this in. I mean, the term AGI is also... We are not talking about stuff in my tongue. We are talking about artificial intelligence, not AGI. No, but the thing is, is that one of the clear paths for artificial intelligence is that it will become more general.
I see. The path of the future that I would sort of draw out is one in which you'll see the trillion parameter models very soon, which will have an increasingly little capacity, that they are designed for text modeling, but they also are particularly math. that you've got people using gpt3 as a text model but they're you because you can do you can predict dna sequences with it of like what sequence the amino acids would go in and it works this is this this generalization of of of functions to which ai would you're going to have a generalization of yes yes i completely agree hold it you have a
generalization of transfer learning where you've got language models that are able to learn winning schema type things better because they're also simultaneously analyzing doing video analysis and they didn't see things going on in the world so you're going to see a i do think the shift from trillion parameter models and a multiplication of trillion parameter models will also be a convergence of these models such that AI and function will become increasingly generalized and will become increasingly a utility based on large foundational models. Whether that is what Kurzweil or Bostrom or someone calls AI is a whole different story that's not even my point.
But the thing, so I think that Wood is, you know, even though I think she's, you know, a kind of a very interesting, but kind of a kook. I do think that we're seeing, we'll see kind of an accelerated, sort of more general uses of AGI. What the impact of this on GDP, that's tea leaf reading. Nobody has any idea of where that was sorted to go. But that's also my point about crypto, is that the technological problems of crypto are comparatively much more straightforward than AI, obviously. the complexity is in the much more difficult questions of adoption. And the question is the symbolization. The adoption. Yes, I would say I'm a very AI skeptic, even though I have written a book on AGI.
I would say that AI progress is simply the prosthesis of what we already have, the certain sort of techniques and certain concepts that we already have obtained throughout the history. With regard to crypto, I wouldn't say that. I say that crypto is a revolutionary instrument. And yes, it comes back to the question of adoption. It is easy to adopt step-by-step growing of theses namely AI it's very easy for any sort of
civilization to adopt that sort of stuff but it's not easy for a revolutionary instrument such as cryptocurrency to be adopted across the board but I would say that I think that in medium time, midterm I think or probably long term, I think cryptocurrency has far more consequences, massive consequences, its adoption than simple AI. I'm not talking about AGI at this point. Well, that's a different question. The question for one, and I think one of the things you're seeing increasingly
is the kind of convergence of crypto and fiat in general. the central bank central no I'm merely thinking about cryptocurrency as a fundamentally new technology yes but I yes the digitalization of money has been going on for quite some time and I think that with the increasingly state-backed cryptocurrency is not mere digitalization of money really of course of course I understand this but I think that I again the central bank currencies, I think in many cases, may push out the major cryptocurrencies in the long run. Yes, they do. They will do. And they may perform functions that this experimental phase of crypto has been pointing to,
as forced people to think about. I'm not at all pushing back on the idea that a fundamental transformation in the infrastructures of value, how we nominate index value, identify value, the granularity of this. All of this is fine. At the same time, I think that I want to talk to you about. But doing this on a deflationary currency with the Gini index, we're probably not going to work. There's my last point. You see, there is a friend of mine who is a mathematician, a magnificent mathematician. I highly regard her as one of the greatest mathematicians of all time. and
she sent me a slideshow about her recent talk it's about the problem of anarcho-communism so anarchism essentially has very small scale local community sort of decision making and so on so forth liberation and communism. It's a nightmare. Anarchism is my nightmare. Yes, on the paper. On the paper. On the paper it is a nightmare. And communism on the paper is also a nightmare because it's kind of the sort of model of communism that we have seen.
So she tries to create a mixed level cybernetic cybernetic anarcho-communism. And she asked this very specific question. Orbit, you know, I know this. Cortes Yarvin, yeah. So Orbit was actually programmed in the vein, explicitly, of anarcho-theodalism. Anarcho feudal. And libertarian monarchism. Yes. But what is the prospects of us adopting a platform such as Orbit
or any sort of platform of similar functions for cybernetic anarcho-communism? mixed level estate governmentality. Yeah, there's a lot of people, a lot of really smart people asking very similar questions to this. The term that I have, I had this for in the stack was what I called synthetic catalog of how you could construct an artificial economics that would allow for proper price signaling, planning, other kinds of structures, all the kinds of stuff that high-end do, but obviously you can in some ways.
So no, I'm not suggesting that there's anything there in this question. I think the questions that are, let me put it this way, I think the questions that people are asking through crypto and through the block and through the prism of the blockchain about how it is that you could reorganize society by reorganizing the way systems identify, index, circulate, model, distribute, and plan organization of value is a fundamental question. And the fact that it's captured so many people's attention is, I think, is quite healthy. I don't have I don't think I think the technologies
that we are currently working with in this regard are Yes are kind of in his infancy are primitive and I also yeah and I also don't and also you know I just don't see I don't you know there may be ways in which you know experiments in anarcho-communism in this kind of regard end up bending you know providing sort of gap spaces and bend things in certain kinds of directions in one way or another. But at the end of the day, I think that the future of governance is operating, is planetary scale. It's not common. And so to the extent to which this future of economics operates, one is a future of economics that operates in accordance,
in relation to the ecological systems upon which it's dependent. I just, for some people this is, I understand for the urban people, this is a path towards secession. And, you know, on the right and for some people on the left, it's a path towards secession. A libertarian for freedom. Yeah, well, of course, it's always where the anarchists and the libertarians kind of, you know, end up at the same parties eventually. But that's it. This is a different story. I see this, all of this as a sort of space for investigation, not the end game itself. I see. The end game itself. I see. It's a very good answer. Superb. You mentioned Nick Lansford on this as well.
I think for him, this is very much a means to an end of, this is how you can build a kind of camera-less succession from this. At the end of the day, I think some people will live on pirate ships, but that's not the main story. usually people actually live on the raft of Medusa rather than on the ship. Yeah, or a crew. Anyway, no, but I guess I wanted to get back to the question about technical optimism that he asked about Kathy Wood as well. Look, I mean, this question about GDP, like what the fuck is GDP, and why would we have GDP in 20 years, let alone be questioning about how this would work in relationship to it. I think if AI ends up having
the impact that what says AI will have, one of the things that it most likely and hopefully would have gotten rid of is GD as a way of modeling this kind of structure in and of itself. And so anyway, there's much more to be said about any of this, but if you want to sort of specific question go further maybe that would be helpful ross uh do we have any more questions let me just quick check in the discords if anything else has been dropped in and also on the on the zoom i mean people are very silent i mean someone should uh precisely because we have uh touched on a very sensitive subjects namely cryptocurrency i know
Lucas, some other people want to say something about this sort of stuff. And incidentally, if I could do some off-promotion, the themes that you and Benjamin were just talking about for the last five or ten minutes are not a million miles away from the course that I'll be running at the Center. Yes, yes. At the end of the year. So, yeah. I don't have any more questions from the Discord server. So if anyone wants to jump in from Zoom for free. Or otherwise, maybe we could draw it to a close. If you like.
Final thoughts? This is really great. I mean, I very much appreciate that. We've been going for quite some time, and I really appreciate the opportunity to kind of stretch it out and go for a long run with you both. It's been really great. Yeah, yeah. No, it's been really, really excellent. I've really, really enjoyed this great deal. And I look forward to, I mean, we have a good basis from the transcript to go through here as well, but there were, one thing I sort of, is there probably four or five points within the conversation where it seemed like we got to a point that there's something, there's a very interesting idea here that would be worth developing further that would require not doing so um on the fly um but will be uh will be new are generally new ideas for me in terms of
where to pull the the string of all this material that i've been working on for so long so that for me that's an incredibly good sign hmm yeah i mean it's a very rich conversation i thought i i thought the conversation went really superb. I don't know. I mean, I don't have anything to say on this until I catch Benjamin red-handed during the exchange, during the transcript. That's all I can do at this point. What are you going to catch? I mean, probably some stuff about naturalism, AI, this sort of stuff, you know,
run-of-the-mill sort of stuff. I still want to actually hear from Benjamin that, you know, I kind of imply that. I asked Benjamin that, are you Switzerland? or do you want to actually go with the Axis or the Allies? What I would want to know for declaring my prejudice is for you to talk a little bit less euphemistically about those positions. Yes, yes. Absolutely, I will do that. I think I know who you mean, but I'm not sure I want to...
I'm not sure I can give you... I don't want to be Switzerland. I don't want to be Germany. I don't want to be England either. But it also depends on who exactly do you mean here? Are you talking about Bennett? Are you talking about Daniel Dennett on one side? Or are you talking exactly who you... I'm happy to tell you my opinion once I know a little bit of, again, less euphemistically, who we're talking about. Yeah, sure, sure. I will get into those sort of scenarios. No, I mean, we have a lot of materials to cover. I mean, but absolutely. Thank you so much, Benjamin. My pleasure. For such a magnificent conversation.
I thank you thank you very much so I will um I'll send you the recordings video recording and another digital audio recording and then um I'll also take a first stab but I'll have otter do a first stab at a transcript from the digital tech because that ends up pretty good and then we kind of never actually translates uh or transcript correctly uh Iranian accent unfortunately is a terrorist accent. Yeah, that's always an issue. I mean, one of the things, the only thing that I found is that literally the recording, the better it does. And so with this digital thing,
it does much better than it records my laptop, even though to me, the recordings sound the same. To it, it makes a huge difference. And so, and you get us started, so we don't have to do the whole thing by hand. Otherwise, it'll be... All good. Yeah, all good. All good. Thank you so much, Ross, Benjamin, and every one of you. It was magnificent. Okay, talk to you soon. Bye. Thank you. Bye-bye. Thanks for being... Ciao, ciao.