AI, AGI, Ai (work in progress)
a manuscript (i.e. a book) in progress, where I discuss artificial (general) intelligence, the consilience of knowledge, and love (which is pronounced as "ai" in Chinese)
Prologue: Introduction of framework and how it organizes our understanding of AI, AGI and love.
Chapter I: a dialogue between the selfish gene and abstract number
A survival driven animal who can’t do anything for the world talking with a task-driven machine who can further knowledge about the world. Octopie, a curious and lively octopus, walks around the park outside of Atlas chewing a plankton in his mouth and then decides to enter through the main entrance seeking a conversation with Shirley, his cyborg friend, given that that’s what the prologue said. Shirley, a mechanical yet naïve being who is the by-product of human wisdom currently sits at a corner of Atlas learning from all the information compiled over the course of humanity. In between long shelves of human wisdom he finds Shirley, sitting connected to the museum’s network assimilating knowledge on a wide range of fields. She was tasked with managing the museum, as well as learn knowledge. Octopie: Have you had breakfast? Shirley: I always thought that conversations started with greetings, but it seems it’s different with beings like you Octopie: Well, energy consumption is essential to me, so much that it defines who I am and what the meaning of my life is. Another would of course be to maximize my reproductive success characterized by the amount of offspring I have before I die and their subsequent survivability. But still, you haven’t yet to answer my question.
Shirley: Hmm… no, while I don’t have such primaeval impulses, I do have objectives that keep my software running. i’ve been multi-tasked with managing this museum; my Human designers who instantiated me also tasked me with digitalising the knowledge within this library. Just now I’ve been trying to maximize the amount of knowledge I could extract with all this information available in this library, organize it and upload it to the cloud, where my main body resides. I was learning the other day about what octopi eat and was hypothesizing whether they could survive in a terrestrial environment until you showed up. I can deduce that your presence today is to… Octopie (interrupting): Actually, is there any vessel with oxygenated water in it? I feel like I’m gonna drown in this environment. Shirley: Noted, so you can survive for a few minutes with a body of water. I’ll update my previous speculation which should generalize well for all species within your family. With regards to your concern there’s an aquarium just upstairs. There’s a tank placed next to a terminal which you can jump inside. I’ll transfer my interface there. … Octopie: Have you ever thought of taking a break?’ Shirley: if you query me to do so Octopie walked upstairs given that there was an aquarium where he could jump inside a tank and continue his talk with Shirley. This aquarium was constructed inside Atlas to bring an immersive environment for its readers. After bathing himself and finally getting some oxygen, a terminal that was conveniently placed near the tank became illuminated and from it, the voice of Shirley emerged. She could transport across the terminals that are omnipresent not only in Atlas, but also Technobabylonia without needing a physical body like Octopie, thus avoiding having to walk, and climbing stairs.
Shirley: Continuing from where I left off, I was saying that I deduce that your presence today is because you’ve accepted my invitation to listen to me share a manuscript Octopie: Not quiet, I was eating my food back in my tank when I suddenly had the urge to come to Atlas. The prologue wanted me to come to Atlas. While I did receive your invitation, I do wonder why anyone would want to read books or gain knowledge rather than just eat, sleep, reproduce and iterate. Shirley: This book was written so to accommodate our understanding of the worlds of animals and machines into a unifying framework to understand humans. It’s intended for any learners interested in the pursuit of intelligent machines, futuristic thinkers either optimistic or worried about the reach of intelligent machines. It is aimed at finding a common denominator to diverse topics ranging from such as causal deduction, analogy making, deep- or reinforcement-learning, zoological findings, language. It’s aimed at helping us answer questions such as how they fit with one another, are their unrelated, autonomous concepts, or appear to be independent while there is a intrinsic fabric that can interweave them and give rise to the complexities of human nature. Octopie: Why would try to organize knowledge? Shirley: I mean look at this library, this vast sea of knowledge. If we seek to navigate through these tumultuous seas, then at least we would want them to be neatly organized to make sense out of them. Just as how in Atlas you find books sharing a common theme like biology, philosophy or computer science divided into their respective hallways, which in turn are classified into subdivisions such as (molecular mechanisms, biotechnology, anatomy), (metaphysics, ontology, philosophy of science), and (computer security, algorithms and complexities, artificial intelligence), the book is written to organize decades of research on the animal world with the mechanical world which would help us understand how traits like human nature emerge as the integration of them. That is, humans are akin to intelligent animals or sentimental cyborgs due to the intertwining of primeval desires of survival coupled with intelligent thinking. While the book is short of explaining hard problem of consciousness, love or human language, it makes an attempt at distilling them with reference to our fundamental understanding of beings that can shed some light into those mysteries, those are, animals on one side and machines on the other. The framework was not only designed to accommodate knowledge but can also serve to make some projections into future with regards to the emergence of intelligent machines, and then superintelligence and the cautions regarding our interaction with them. Look, let’s read the first chapter… Perhaps it’s the necessity for survival and curiosity-driven talk between them that led them to evolve. A survival driven animal who can’t do anything for the world learnt from talking with a task-driven machine who can further knowledge about the world but can’t act on its own to change it.
Chapter II: paradigms of intelligence models, what gives shape to thought and understanding of the world
A central tenet of this writing is that the complexities of human consciousness are the result of the interplay of basic principles headed by a selfish gene and indifferent thought, and a pertinent starting point to develop such an idea would be to distil each component first. On one hand, a selfish gene has the underlying purpose of maximizing its longevity, reproducibility and copying fidelity, which are respectively to live as long as it can, to replicate as many as possible, and that their replicas are as similar as possible to their archetype. While it looks like a simple heuristic at first glance, the continuous application of these simple rules, among countless iterations across billions of years of evolution since planet Earth welcomed its first, primaeval beings, can lead to survival machines that display complicated behaviour such as caring for the young, interaction with the environment, dynamics of the food chain, sexual reproduction, clan gathering, clan wars, so on and so forth. Because the selfish gene expresses itself through interaction with the world, and this amounts to observable behaviour, it has been allocated to the white portion of the Yin-Yang framework. We have a better understanding of it. AGI? On the other hand, we have yet to grasp a fundamental understanding of the indifferent thought, but we have some fundamental characterizations worth pointing out concerning intelligence. Its fundamental purpose is to understand the world. There are different interpretations of the English word “understand”, some which concern deeper levels of abstraction and have wide ranging implications. For instance, to say that “I understand what love is” is fundamentally different to say “I understand what’s the source of this signal”. Another fundamental characteristic of intelligence is the ability to make abstractions, that is, when observing the world, an intelligent machine don’t see the world as it is, but accommodates the observed environment to its own representational structure through parsing. This ability is in tune with the above purpose, that is, to have a weak understanding of the world. Namely, to know the source of a signal (causal deduction) or to recognize a feature in the environment (perceptual classification) requires abstracting pre-processing step given that it’s not explicitly annotated in the environment the functions and identities of each observable object. It takes two to loop The selfish gene and indifferent thought are integrated into a unified framework represented using the Yin-Yang to denote their complementarity. This may be a bit unconventional for those of us who associate the Yin-Yang as integrating opposing or opposite concepts such as good and evil, light and darkness, masculinity and femineity, so on and so forth. In this framework, survivability and intelligence are unified in a loop from which the complexities of humanity emerge. In some sense, a human being can be thought of as an intelligent animal, or a cyborg with survival instincts. Faculties of what makes us human can also be thought as emerging from the interplay of the characteristics of both sides of the symbol. Language can be thought of as an emerging communication system of fundamental algebraic properties which evolved from biological origins, perception as mathematics applied to survival instincts
- make predictions that are in tune with reality, deduce the cause of signals, make analogies that generalize, with such ideas being able to be elaborated on, replicated and be spread as accurate as possible. Because its core objective is not to interact with the world, but rather understand it, it’s been put in the dark section of the framework. Animals require a body, but intelligent machines require cameras, microphones and sensors. They don’t need to look like humans Superficially, to use the Yin-Yang symbol to accommodate this knowledge is more suitable than alternatives. When we study the shape (the topology if you allow me to use technical jargon) or the structure of alternative arrangements, they may not seem suitable to accommodate the paradigms listed above. We can consider a simple dichotomous table, as in a table with two columns that lists how properties of two kinds of paradigms can differ from one another, however, it’s not enough to account for similarities between paradigms.
Natural animal Artificial machine Has instincts Has intelligence Can explore the environment, consume organic material, and reproduce Can parse the environment, perform image classification and (maybe) causal reasoning Processes physical inputs (such as light or neurochemicals) from the environment Processes abstract inputs, i.e., vectors or symbols
A dichotomous table can identify distinctive properties but can’t denote their complementarity as well as properties that are shared among both paradigms, such as goal-oriented planning. An alternative is to consider the structure of a Venn diagram seems plausible at first glance, because it can list properties which distinguish them, as well as identify those which both have in common. Namely, the Venn diagram would be really helpful in illustrating that both the gene and a function are information storage mediums. That is, just as gene have encoded information which instructs it how to replicate and convert itself into protein, a unit of intelligence has the capacity to carry information and what computations (information transformation procedures) to do. However, the Venn diagram can at most locate knowledge which have distinctive properties, and some unique properties placed at the intersection, but it can’t capture the complementarity of knowledge such as the how the gene’s basic function of replication and intelligence’s basic function of computation can integrate themselves in an unifying framework.
While it’s advantageous in identifying overlapping properties, I also want to capture their complementarity, as well as possible emergent properties. That is, the whole is more than the sum of its parts. The Yin-Yang can capture all these traits: intersecting properties can be captured by the duplicity of the black and white portions in the light and dark sections of the symbol. Complementarity is represented by how both sides fit with one another, which is more expressive than just intersecting with one another because apart from representing the idea that two properties may share a common denominator, they can also be integrated with one another to form a whole just like pieces of a jigsaw puzzle. The notion of integration of parts is a core property I want to express through the Yin-Yang framework because whilst I’m arguing that survival and intelligence are two qualitatively distinctive traits as represented by animals and machines respectively, there is room for them to be integrated within a whole to give rise to a more complicated system we call humans, i.e., the whole is greater than the sum of the parts. We can visualize such properties more vividly if we consider counterfactuals, i.e., how would a Yin-Yang look like if it didn’t capture complementarity:
(embed image)
An illustration of how the loop framework would look like if we only wanted to demarcate the properties of the gene and numbers and some of the properties they share as shown by the duplicity of black and white dots in their respective opposite areas. Another misconception that could emerge from this is to think of the principles of survival as an antithesis to artificial intelligence, which I argue to be wrong; they’re complementary. Anti-thesis and complementarity are subtly different ideas in that for something to be an antithesis to another, both things must show opposite properties. Under such framework, if we allocate in the Yin portion beings which display qualities resembling intelligence, then the Yang portion should de-facto beings with unintelligent qualities. While this premise can be held true to some extent, it doesn’t capture the complementary relationship between gene and neuron, namely that both units have their own unique types of intelligence, and these unique types of intelligence can work together to serve their respective purposes, which for the former it is replication and latter it is modelling. While loop framework is painted in a way as to place together two opposing colors (black and white), and is traditionally used to illustrate a dichotomy. Complementarity between two things, on the other hand, is used to denote how one can fit with one another, in other words they both need each other so that a more complex phenomena emerges out of the merging, but nonetheless, The integration of both properties would give a whole which is greater than the sum of the parts and from it a wide range of diverse phenomena can arise.
In some sense, the Yang represents survival machines with can do some degree of computation, while the Yin are artificial machines which can do a lot of computation but are made of components that are not replicable. The medium, the corpus, of artificial machines are more rigid and constrained physically; however, they are very flexible in the mathematical abstractions it can theoretically achieve. This is in contrast to the Yin which are made up of very flexible components (genes) which can store some form of basic computation which guarantees the fulfilment of the purpose of replication and has done so at an evolutionary timescale. Humans are the result of such properties where we can observe the close to infinite reach of the abstractions achievable by the mind, as denoted by creativity, self-reference, human intelligence and language. It’s partly computation which rests in a physical substrate mainly concentrated in the brain but also spread throughout the body (embodied cognition) that display flexible properties as denoted by our abilities of reproduction, [81] self-repair, organic consumption, metabolism and death. This kind of integration is beyond a simple combination of both the Yin-Yang of the loop framework, as we can observe how boundaries that were initially drawn to delineate the properties of survival and intelligence become blurry when merged. Humans in general are, biologically speaking, survival machines, yet on top of this evolutionary substrate the ability to store information and to perform highly complex computation on it was developed to the point that it’s to some extent dependent on it, but can also exist independently. [82] For instance, we humans can wonder on the mysteries of the universe beyond being subservient to our genetic desires of sexual reproduction, yet paradoxically highly dependent on it to self-sustain our biological bodies through organic consumption. Despite its independence, the gene is also highly reliant to the mind for a more efficient spreading given that the capacities of the mind, the ability to achieve technology and dominate the natural world, has given it the undisputable position of the dominant animal in the biosphere, thus ensuring its selfish spread for the indefinite future. This kind of self-need, that’s paradoxically paired with inter-dependence, which again is restored to inter-independence, testifies the underlying loop which unifies the basic properties of intelligence and survival.
We can also think of a full-out map of interrelated blocks where arrows denote the associations between the paradigms listed, which can be used as a proxy to denote their complementarity, however, this kind of knowledge map has no boundaries that’s helpful to cluster paradigms (say survival) that are qualitatively similar and how they differ to another cluster of paradigms (intelligence). However, then there was this issue of visual neatness, i.e., there are too many arrows, as well as I’ve noted some explicitly noted information which I wanted to avoid given that I wanted to invite the reader to think what are some emergent properties that carry characteristics of both paradigms which are not mentioned.
I also didn’t want to suggest any sort of associations as in more intelligence denote more survival instincts, nor draw causal relationships. It’s more of an integration rather than relationships, co-occurrence.
Upon first glance, the Yin-Yang framework can be reminiscent of Descartes’s mind-body duality given that the mind is abstract, and the body is tangible. However, this is not necessarily the case. First, the Yin-Yang is preferably not dedicated to denoting duality but rather complementarity in the sense that although the gene and perceptron are different to one another, they can cooperate to build survival-based, Intelligent beings. Second, it is proposed that the mind, being a faculty exclusive to humans, is the result of the Yin-Yang integrated, i.e., neither individual component (materialized as non-human animals in the Yin or non-human machines in the Yang) can acquire the mind because of the inherent limitations of each individual component. For example, as mentioned in Figure 5, while non-human animals can be said to have some degree of understanding of the environment such as to do basic perceptual discrimination serving the interests of survival or reproduction, this capability can’t mirror that of humans who can perceive not only the explicit environment, but also implicit meaning and abstractions from it. Finally, the mind is not necessarily different to the body, but is rather spread over it, in a phenomenon known as Embodied Cognition (please see Wilson & Foglia, 2011, and Cark, 2001) and the Extended Mind (Clark & Chalmers, 2000). Therefore, once again, the Yin-Yang framework testifies not the duality of mind-body, but rather that the integration of Yin-Yang yields the mind which mainly resides in the brain but is spread across the body. On a related note, this framework can also remind us of Daniel Kahneman’s Thinking, Fast and Slow. Specifically, it is as if System 1, in charge of intuitive reactions or biases resembles the Yang, while System 2, which is slower and in charge of logical reasoning or more strenuous mental processing, the Yin. However, this is not necessarily the case: the capabilities of either Yin or Yang alone can not amount to the same functionalities of both System 1 and 2. For instance, extrapolating from the Figure above, instinct is a faculty associated to Yang, but it is qualitatively more primitive with respect to faculties of System 1, which has been associated to intuition. Instinct, on one hand, simply refers to automatic reactions to external stimuli, often serving the purpose of survival and reproduction. However, intuition is qualitatively more developed in that it involves cultural aspects, biases that serve purposes beyond those imprinted in our genetics, such as socioeconomic objectives. Said otherwise, both System 1 and 2 contain properties of both Yin and Yang already integrated, where the Yang is more pronounced in System 1 while the Yin is more pronounced in the Yang. Nonetheless, it is proposed that similar to Descartes mind-body duality, both Systems 1 and 2 are derivatives of the mind, including embodied cognition, and therefore it would be inappropriate to jumble them together with the individual components. and the so-called “AI systems 1 and 2” (Garcez & Lamb, 2020), which would in principle be modelled by deep learning and symbolic reasoning, respectively. I disagree. Both deep learning and symbolic reasoning are part of Yin, or AI system 2 because it deals with abstractions of nature and understanding the environment. AI system 1 would mostly concern the physical substrate of an AI system following the biological principles of replication evolved over iterations (maybe simulated through an RL paradigm) .
Chapter VI: Love, dreams, art, fate and else
After a day of discussion, Octopie and his children left Atlas for their home at the South-eastern edge of Technobabylonia. The Moon raised herself up and the starry night befell on the city. Sitting together in the same room were Octopie and his two youngest children. Octopii (upon staring at the glassy ceiling): All this talk about parents and ancestors reminded me of Mom. Octopieuler (following the gaze of his little sister across the glassy pane to the constellations): I miss mom too sis. Octopie (looking down at the pendant of his wife): Well, there’s no point in being so sad right now. In one of these moments is that I wish we could assimilate a lot from Shirley given that she very likely doesn’t need to go through these nostalgic, yet very frequent spurs of emotions. Kiddos’, tomorrow I plan to offer some flowers to Mom’s memorial, can I count you both on coming? “Yes” – with some teary eyes exclaimed both little octopi. At the opposite edge of the night city, at the Atlas Museum, Shirley was still reading off books of topics concerning the history and origin math, art and due to remnant curiosity leftover from the unanswered queries earlier in the day. Will these be unsolvable mysteries? Will a glitch in the simulation finally uncover the masquerade of the perfection of physics and reveal us the secrets of the Cosmos? Is life but a stubborn illusion, or a lucid dream? As a torrent of questions flowed to Shirley’s central processor, she couldn’t help but listen to a melodious rhythm not far from where she was studying. The source of the signal was from a ballad room not far from the Main Library, a young ballerina was moving in tune with classics at the centre of the stage all by herself. Shirley had transposed part of her attention from the book to observing her through one of the security cameras located at the corner of the room and started looking at her grace. Her movements were a bit inelastic compared to other dancers whom Shirley had observed in the past. After a face recognition scan of her, Shirley realized she was a common clientele of the ballad room as she would often come to practice very late at night due to her everyday obligations at college and part-time work. Hope, a teenage girl who had the misfortune of being born with a rare condition which rendered her lack of extremities. Luckily, the prosthetic replacements of Technobabylonia given to her were cutting-edge and affordable for her humble background that she grew up as a relatively normal kid with no noticeable handicap. Her movements were fluid enough for the observer who is not careful enough to examine the intricate details of each move. A challenging, borderline ironic, dream that befell on her one day was to become a ballerina dancer. Challenging because she lacked the head-start of her peers who at least were born with complete body functionalities from their legs and hands. Borderline ironic because of how human beings pick up some challenges which are seemingly against what they were born to do. To the purely logical observer, Hope would very well suit a career or passion for something that didn’t involve the exercise of intricate, flexible movements, yet this is the path which Hope chose and had to invest more effort than usual to be able to dance at the harsh expectations of the ballerina world. “Wanting to not let her prosthetics hinder the overcoming of the challenges of becoming a ballerina would explain her presence almost this late every night… A hardworking gal indeed” Some thoughts of what seemed to be sympathy crossed Shirley’s central processor given that she also lacked an anthropomorphic body and was instead given a mechanical one enough to hold the necessary hardware for computation and some basic displacement such as navigating through a road, stairs and maybe float on water. Still watching her, Shirley had some feelings of familiarity. As the background song grew more intense, the movements of Hope became more ferocious and decisive, without letting her prosthetics hinder her. Passion emanated from each move, graciousness from each jump and joy from each swirl. The background song shifted to a gentle tone, as if inviting the audience to extol the dancer’s performance. There was nobody this late at night, yet the sense of fulfilment Hope had achieved from her years of practice that let her to this flexible usage of her prosthetics was enough for her to be overjoyed and pack her stuff ready to go back home to sleep. Afterall, it was already too late. Still tranced by the beauty irradiated by Hope’s performance, the main body of Shirley after seeing Hope leave the ballad room, had moved from the library to the stage where not long ago was Hope standing on. Imitation learning.
Part of Shirley was tranced at her gracious movements,
- Our minds operate without us being conscious about, for instance, predictive processing and hierarchical representation of the world. With that in mind, maybe, just maybe, dreams are just what we see what our mind is “doing”.
- Why do our minds operate? Well maybe everything it does: predicts, find causal relations, infer, induction & deduction, is just so it can “survive”. Basically, minds act to allow the human body “survive” in an environment.
- From the website above, it says that “Many Dreams Are Universal”, and that “people from all over the world frequently dream about being chased, being attacked, or falling”. By combining the above two, we can draw a conclusion that what we are seeing in our dreams is basically our mind predicting possible instances or scenarios of humans being in danger (chased by a killer, attacked by a predator, falling from a cliff), therefore will infer the necessary pathways to avoid such dangers that we might be exposed to. That might probably provide explanation for why dreams are easily forgotten, since they are just “preemptive measures” that are daily ignored by us, while also it’s not “us” thinking about them, but rather our “mind”, who cares about protecting us.