Artificial General Intelligence is first of all a science and engineering topic; and as an AGI researcher working on OpenCog and other AGI ideas, I generally think of it in this way.
But of course, AGI is also much more than a technical pursuit. It has big implications about humanity’s place in the universe – both what humanity’s place is logically and scientifically speaking, and what we want it to be normatively. It raises a host of dizzying possibilities regarding our possible futures, including the potential of fusing our minds with AGIs cyborgically, and becoming “more than human” in various senses (some quite strong). It even raises the possibility of understanding the universe and finding answers to the Big Questions of Life, the Universe and Everything – with the help of engineered transhuman minds, to aid us in venturing where the human mind is too weak to travel. And then there are the possibilities highlighted by folks like Hans Moravec, Hugo de Garis and the SIAI, of human extinction, of our replacement by transhuman AGIs who don’t care any more for humans than humans do for flies or bacteria.
One critical aspect of the relation between AGI and the Big Picture has to do with consciousness. If one thinks AGIs can be conscious in the sense that humans are – or maybe even more broadly and richly conscious – then that certainly informs one’s sense of AGI’s place in the Cosmos. On the other hand if one things AGIs are intrinsically just feeling-less, experience-less tools, then the story becomes quite different.
There are many, many ways of fitting AGI into one’s view of Cosmos and consciousness. In this article I'll semi-systematically run through a number of the possibilities, to wit:
Of course this is not intended as a complete list of possible perspectives… And it’s also worth noting that these are not mutually exclusive categories, by any means! Some individuals or schools of thought may adhere to more than one of these perspectives. (And finally -- be aware that the labels in this list are ones I’ve adopted for my own convenience, rather than necessarily because they’re the most standard ones!)
Sometimes I hear people express the naïve thought that AGI research is naturally aligned with a particular philosophical perspective, usually the views I refer to above as physical or informational monism. But I don’t think any such natural alignment exists. I’ve observed AGI researchers and AGI-focused transhumanist thinkers to take a great variety of perspectives.
In this article I’ll briefly run through these views I’ve listed, with more emphasis on breadth than depth. I don’t want to turn this into a treatise on cosmic philosophy! – But I do want to indicate the wonderful diversity of ways in which the notion of AGI is already being interwoven with various human thought and belief systems. One can expect the weaving of AGI into these various human conceptual networks to get even richer once advanced AGI is actually created and interacts with us on a regular basis!
I’ll spend relatively little time on those perspectives that are already extremely well-known due to their commonality, and dwell more on those views I find most interesting! Also, note that two of my own favorite perspectives appear at the end of the article (Cosmism and the Implicate Order… My other favorites of the bunch being Informational Monism and Panpsychism). So if you don't read to the end you’ll miss the good stuff!
What I refer to as “physical monism” is the view that the physical world is in some sense absolutely real and out there and existent – and everything else is a bunch of malarkey. Intelligence, mind and consciousness and experience are viewed as patterns or configurations (or whatever) of physical entities like particles or waves.
To my mind, this is a potentially useful (though rather limited) perspective for getting through everyday life, but somewhat deficient as a real intellectual theory.
After all, how do we know about this supposedly “absolutely existent” physical world? Either we take it on faith, like a prototypical religious person’s belief in God… Or we infer it from various observations, i.e. various pieces of information. But if it’s the latter – then isn’t the more fundamental reality associated with the information, rather than with the physical world whose “existence” we infer from the information?
A classic expression of this view is G.E. Moore’s observation that when you kick a rock, you know it’s real, and philosophical babbling becomes irrelevant in the light of this direct evidence of the rock’s reality. But what is it that you actually know is real, when you kick the rock? The definite experienced reality is attached to the feelings coming to you from your foot (i.e. feelings that you have learned to attach to the concept/percept network called “foot”), and the sensations coming into your eye when you look at your rock and the foot. Actually there might be no rock present in physical reality at all – your brain might be connected to an apparatus that makes you think you’re kicking a rock.
And this leads up to the next perspective in my list, which I find deeper and more interesting…
The view of the world as bottoming out in some absolute physical reality seems hopelessly naïve to me -- but the view of the world as bottoming out in information seems much less so. I don’t quite adhere to this perspective myself, but nor do I know of any rational, scientific arguments against it. AGI researcher Joscha Bach, whom I interviewed for H+ Magazine, described the perspective well:
We grow up with the illusion of a direct access to an outside world, and this intuition is reflected in the correspondence theory of truth: our concepts derive their meaning from their correspondence to facts in a given reality. But how does this correspondence work? According to our current understanding of biology, all access to the world is mediated through a transfer of data, of bits (i.e., discernible differences) expressed by electrical impulses through sensory and motor nerves. The structure of the world, with percepts, concepts, relations and so on, is not part of this data transfer, but is constructed by our minds. It represents encodings over the regularities found in the data patterns at the mind's interface. In theory, all the data entering my mind over my lifetime could be recorded as a finite (but very long) vector of bits, and everything that I consider to be knowledge of the world is a more concise re-encoding of parts of this vector.
Of course, even the concept that the world is mediated through sensory nerves is not directly accessible. It is an encoding, too (it just happens to be the best encoding that we found so far). And because we cannot know the "real" structure behind the data patterns at our mind's systemic interface, the correspondence theory of truth does not work, outside the abstract and pure realm of mathematics.
But what about information itself: how can we know about bits, data and encodings? Fortunately, these are mathematical concepts. They can be defined and operated upon outside of any empirical world. For instance, the theory of Natural Numbers does not need anything material (like piles of apples and oranges) to work: It follows automatically from the completely abstract and theoretical set of Peano's axioms. But Natural Numbers can be used to encode some aspect of apples and oranges--their cardinality, and their behavior with respect to addition and subtraction.
The space of mathematics is self-contained, but to explore it, we need a certain kind of information processing system, which is a mathematical entity too. Minds are part of the class of information processing systems that can perform mathematics (at least to some extend), but minds can do even more: they can conceptualize a world, reflect, plan, imagine, anticipate, decide, dream, interpret themselves as persons, be in emotional states, attach relevance to concepts and so on. The concept of mind is the one we attach to ourselves, we use it to encode that part of the information vector that we consider to be us.
With respect to our understanding of it, the abstract theory of minds is still in its infancy. Our common-sense understanding of what it takes to be a mind is good enough to use it as an encoding concept (i.e., to use it as a conceptual framework that allows the interpretation of parts of the world, as people, self, mental states, emotions, motives, beliefs and so on). Yet, our concept of minds is incomplete, muddled and likely inconsistent. It is possible, however, taken enough time, effort and brain power, to define all the aforementioned capabilities of minds down to the nitty-gritty detail that would make it the mathematical equivalent of a theory of Natural Numbers: a formal and complete theory of what it takes to be a mind. We can express this theory, for instance, as a computer program. And this project is exactly what Artificial Intelligence is about.
I think this makes far more sense than physical realism, or mind/matter dualism. It has led some thinkers to posit that the universe is basically a giant computer. While I don’t quite embrace this myself, I do think the model of the universe as a giant computer has a lot to teach us.
Joscha and I have argued about this extensively face-to-face, and as I told him in our conversations, I agree with him that insofar as it’s investigable using science, the universe may be considered as consisting solely of information. The difference between our perspectives is that I don’t think all aspects of the universe are scientifically investigable.
Science, by its nature, is about gathering finite sets of finite-precision observations – i.e. the whole corpus of scientific knowledge consists of some finite set of bits. But there’s no reason to believe that the whole universe consists of some finite set of bits. We may not be able to measure anything else via the means science, but this just means that only information is “scientifically existent”, not that “scientific existence” is the only kind of existence!
When we last discussed the matter, he was willing to concede this logical point, but also said (to paraphrase more or less loosely) he felt there was no point in attempting to talk or think about forms of existence beyond the scientifically measurable realm. And this difference of perspective between us doesn’t affect our ability to collaborate on AGI work at all, since our AGI engineering efforts do in fact concern the explicit manipulation of scientifically measurable entities!
A variant on the above perspective holds that the universe basically consists of quantum information. This seems to be, for example, the view of Seth Lloyd, quantum computing pioneer and author of the fabulous book Programming the Universe. In the same sense that informational monism deems the world a classical computer, this perspective deems the world a quantum computer.
In the quantum informational monist view, classical information is an approximation to quantum information, relevant in situations of minimal quantum coherence. From the classical informational monist view, on the other hand, quantum theory itself is just a computational explanation of some classical information that’s been gathered from measuring instruments (which are themselves known via information gathered via other measuring instruments, either artificial ones or the biological ones we call our senses). I.e., quantum theory is a classical computer program for getting from some classical information to some other classical information.
Joscha Bach, when I interviewed him, summarized the informational monist view of quantum computation as follows:
Everything that enters the human mind can be reduced to a finite number of discernible differences. In this sense, the most primitive encoding of the universe, as presented to an individual observer, would be an astronomically large, finite vector of bits. Obviously, our minds can do better than that: much of the structure found in that vector conforms to a three-dimensional space, filled with objects. These objects may influence other objects in their vicinity, by touching them, radiating upon them and so on. Thus, we arrive at classical physics as a way of encoding the universe. Classical physics, when expressed rigorously, is a self-contained mathematical theory; objects in classical physics are entities with properties and modes of interaction that are reducible to calculations. As it turns out, classical physics can not only be reduced to computation, it is in turn powerful enough to explain how computation is possible (i.e., information processing in general): using classical physics, we can conceive of a computer that can simulate classical physics.
When we look upon our basic universe input vector more closely, however, we find that the model of classical physics is flawed: it is not totally consistent with the available data. What we take to be objects does not only interact locally, but occasionally over distances in space and time, and what we take to be a defined state of a microscopic object can only be described as a space of possible states, which it occupies all at once. If we care about this discrepancy, we have to abandon the notion of the classical universe, and adopt a theory that accommodates the observations, i.e., quantum mechanics.
Even the quantum mechanical universe is just our way of encoding the finite string of bits we began with, so of course it is computational. But quantum computation allows to accommodate computers that perform some kinds of computation vastly more efficient, which means that a computer designed along classical principles might be too slow to practically simulate a complex quantum mechanical system.
At this point, it is not entirely clear if a computational theory of the mind would have to be formulated along the lines of quantum computation, or if classical computation is sufficient. But there is practically no evidence that the information processing performed by the neurons in our brains would somehow crucially depend on non-local or quantum-superpositional effects, or that people can somehow perform computations that would require quantum computing. Thus, even though classical computers are too limited for detailed low-level simulations of our universe, they are likely probably perfectly adequate for a detailed low-level simulation of a mind.
For Joscha, in other words, it’s bits that are primary. Classical physics is one way of explaining bits, and quantum physics is another way of explaining bits. Either classical or quantum physics is ultimately expressible as a mathematical construct, i.e. as a set of formulas for manipulating symbols, used to predict future bits from past bits. So in Joscha’s informational monist view, the ultimately reality is the bits and not the classical or quantum physical explanation of bit patterns. On the other hand, from a quantum computationalist view, if the assumption of the quantum model of reality is the simplest way to explain the stream of observed bits, then this merits the assignation of some fundamental reality to the quantum constructs involved in the quantum model of reality.
The perspectives I’ve considered so far in this article are basically “scientific” in nature, focused on things like physics and information theory. But there are also some serious AGI researchers who take quite different views, drawing more from the religious or spiritual side of life.
I’m not a religious person by nature or culture – according to facile high-level classifications of religious belief, I’ve veered between “atheist”, “agnostic” and “spiritual but not religious” in various forms at various times of my life. However, I’ve always been fascinated by religious peoples’ beliefs and attitudes, feeling that they do address important aspects of the universe that science (so far, at least) has given us little insight about.
After all, no matter how much you value math and science, they can’t tell you everything. Mathematical derivations start with assumed axioms; and as David Hume was the first to carefully argue) science requires some inductive bias or it can’t infer anything specific from observed data. Everybody’s got to assume something to get by in the world, whether they realize it or not. So it’s interesting to look at the great variety of things that different people and cultures assume.
Even those who, like Zen masters or Sufis, like to talk about assuming nothing – even these folks still implicitly act as if they’re assuming something, as they go about their lives. When the Zen master opens his refrigerator to get some water, he is implicitly assuming there will be water in there – instead of for example, a lion, in which case he might bring his gun with him to the fridge rather than a cup. And the various implicit assumptions he makes throughout his life network together and evolve in complex ways, just like with everybody else. Both the implicit and the explicit assumptions people make are quite interesting.
However, I do find it interesting to understand what truly religious people think about AGI and other transhumanist topics. In an attempt to fulfill my curiosity on this, last year I did some H+ Magazine interviews with a few religious and spiritual types about AGI, transhumanism and religion.
For instance: Changle Zhou, the dean of the Cognitive Science Department at Xiamen University – where I’m an adjunct research professor, and where a team of students are working on some OpenCog-related AI software projects – is also an experienced Zen Buddhist practitioner. He reports with a grin that he stopped his formal Zen study some years back, when his Zen master declared him Enlightened! He’s also the author of a book, in Chinese, about the relation between Zen and Science. It was both a privilege and a source of considerable entertainment for me to interview him about his views on the relation between AI and Zen…. Some quoted from my interview with Changle:
Consciousness, as I understand it, has three properties: self-referential, coherent and qualia. Even if a robot becomes a zombie that acts just like it possesses consciousness, it won’t really possess consciousness – because the three properties of consciousness cannot all be realized by the reductive, analytic approach, whereas AI is based on this sort of approach.
In Zen we say that SUCHNESS is the nature of everything. For a conscious human being, its Suchness is its consciousness; but for material things, their Suchness is their properties such as shape, substance or weight but not consciousness.
The Chinese word for "suchness" is "真如" or "自性", i.e. "buddhahood", also known as "the foundational 识" or "种子识" or "Alayavijnana" etc.... It is the name for the nature of everything.
The awareness (enlightenment) of Zen is beyond all concepts, but all the approaches we use for building robots and AI systems, and all the behaviors of robots and AI systems, are based on concepts
Perhaps these brief excerpts give you the essence of our dialogue, and our disagreement. He believes that super-capable robots and other amazing futurist technologies are likely to exist – but he sees any robots that human build, as essentially extensions of human intelligence. He sees “natural” intelligences like humans, as possessing a certain special quality – Suchness – that mechanical objects like robots will never possess.
I call this sort of perspective “experiential vitalism.” That is: the belief that, while the practical functions of intelligence can be carried out by digital computers or other engineered systems, there’s some essential crux of consciousness that human mind/brains possess that these other systems never will.
A related perspective on AGI and the cosmos, fairly common in the general population though not so much among scientists, is that human brains contain some kind of special quality, which lies outside the domain of empirical science and is responsible for some key aspects of human intelligence.
Perhaps the most articulate advocates of this perspective that I’ve encountered is Selmer Bringsjord, an AI researcher and logician who is also a devout Christian. Where Changle says Suchness, Selmer says Soul – but their perspectives seem closely related. However, there’s a key difference – because Selmer also argues that some functional aspects of human-level intelligence are reliant on the Soul, which digital computers will never possess. This, he believes, is the core reason why the AI field has not yet led to software with human-level general intelligence.
Selmer goes on to associate the soul with non-Turing hypercomputation, a kind of computing that cannot be achieved via physical computers built according to any kind of currently known physics, and that also cannot be measured using empirical science as currently understood. That is: Science as we know it consists of the collection of “scientific data” which consists of finite sets of bits (two other ways to phrase “finite sets of bits” would be “finite amounts of information” or “finite sets of finite-precision numbers”), and then the extrapolation of these to predict the outcomes of future experiments, which will also take the form of finite sets of bits. There is no scientific experiment, conductable within the contemporary understanding or practice of science, that would be able to identify a hypercomputable process, or distinguish it from a conventional computing process. In that sense, hypercomputing is a non-empirical concept – beyond the realm of measurement!
And here the philosophy gets tricky.
If hypercomputers are not measurable, in what sense can we know about them at all? What sense does it make to talk about them? Well, they can be described using mathematics!
But then, what is this mathematics? In practice it’s just the marking-down of mathematical symbols on paper or computer screens – i.e. it’s a game played by humans with finite sets of bits.
But in the Platonist philosophy of mathematics, mathematical entities are envisioned to have some fundamental reality going beyond the notations that we make to indicate them. So if one accepts a Platonist view that math constructs have their own reality beyond empirical science or human communication, then one may say that hypercomputable entities exist in this abstract Platonic math-space, and that Soul also exists in (or at least is better represented in terms of, compared to any representation in terms of empirical science) this abstract Platonic math-space. And digital computers that we build, which manipulate finite bit sets based on scientific theories inferred from scientific data that’s comprised of finite bit sets, live in a much smaller and more impoverished region of abstract Platonic math-space, not touching the region needed to talk meaningfully about human Soul, or human-level intelligence.
One of the more frustrating, and gutsier, things about this perspective is that it ultimately places the understanding of human-level general intelligence outside the domain of science – though not necessarily outside the domain of mathematics. It suggests that we can intuitively apprehend human-level general intelligence using our trans-scientific hypercomputation capabilities, but can never test these apprehensions scientifically. It also raises the possibility that maybe we could somehow build an AGI using intuition rather than science. I.e., if the world has hypercomputable aspects, and our minds do also, then maybe the hypercomputable aspects of our minds could intuitively tell us how to shape the hypercomputable aspects of the world, and we could make a physical AGI system in a manner not determined by any finite bit sets of measurements.
But so far as I know Selmer is not trying to pursue any kind of recondite Soul-guided AGI engineering of this nature. Rather, his concrete AGI work is focused on the implementation of logic systems in digital computers – and on seeing exactly how far one can push this sort of methodology before it runs into the fundamental limits that he believes to exist.
I don’t know of any simple non-technical write-up of these interesting notions, but if you have a bit of math and computing background, you may enjoy Selmer’s book Superminds: People Harness Hypercomputation, and More.
Panpsychism occurs in various forms, but in the broad sense it refers simply to the idea that mind is a fundamental feature of the universe and each of its parts, rather than something that is the exclusive property of specific kinds of systems like humans, other higher animals, intelligent computer programs, etc.
Though not a common view in contemporary Western society, philosophy or science, panpsychism does have a long history in historical Western philosophy, encompassing thinkers like Leibniz, James, Whitehead, Russell, Fechner and Spinoza. A host of recent books treat the topic, including Skrbina’s Mind that Abides: Panpsychism in the New Millienium and Strawson’s Consciousness and its Place in Nature.
Panpsychism also has a long and rich history in Eastern philosophy, e.g. the modern Vedantic thinker Swami Krishnanda observes
The Vedanta philosophy concludes that matter also is a phase of consciousness and objects of knowledge embody in themselves a hidden potential of consciousness which is also the Self of the perceiving subject, enabling experience in the subject. The subject-consciousness (Vishayi-chaitanya) is in a larger dimension of its own being as universality and all-pervadingness beholds itself in the object-consciousness (Vishaya-chaitanya), thereby reducing all possible experience to a degree of universal consciousness. Experience is neither purely subjective nor entirely objective; experience is caused by the universal element inherent in both the subject and the object, linking the two terms of the relation together and yet transcending both the subject and the object because of its universality.
Advocates of panpsychism point out that alternative theories of mind and consciousness are riddled with problems and inconsistencies, whereas panpsychism is simple and coherent, its only "problem" being that it disagrees with the intuition of many modern Western folk. Most current theories of consciousness involve mind and awareness somehow emerging out of non-sentient matter, which is conceptually problematic. Philosopher Galen Strawson has recently lamented the basic senselessness of the notion that mental experience can emerge from a wholly non-mental, non-experiential substrate: "I think it is very, very hard to understand what it is supposed to involve. I think that it is incoherent, in fact..."
Dualist theories in which the mind-realm and the matter-realm are separate but communicating also run into difficulties, e.g. the problem that (put crudely) the mind-realm must be utterly undetectable via science or else in effect it becomes part of the matter-realm. Panpsychism holds that everything in the world has mental extent, similar to how it has spatial and temporal extent, which is a simple proposal that doesn't give rise to any conceptual contradictions.
Some have objected to panpsychism due to the apparent lack of evidence that the fundamental entities of the physical world possess any mentalistic properties. However, this lack of evidence may easily be attributed to our poor observational skills. By analogy, humans cannot directly detect the gravitational properties of small objects, but this doesn't render such properties nonexistent. And in appropriate states of consciousness, humans can directly apprehend the consciousness of objects like rocks, chairs or particles, a fact driven home forcefully by Aldous Huxley in his classic book The Perennial Philosophy.
Panpsychism is not without its difficulties, e.g. the “combination problem,” first raised by William James – which in essence wonders: if everything is conscious, how does the consciousness of a whole relate to the consciousnesses of its parts? How does the brain's consciousness come out of the consciousnessess of its component neurons, for example?
But this doesn't seem a problem on the order of "how does consciousness emerge from non-conscious matter", it seems more a technical issue. A large variety of qualitatively different part-whole relationships may exists, as physicists have noted in the last century. Quantum mechanics has made clear that systems are not simply the sum of their parts but can sometimes exhibit properties that go beyond those of the parts and which cannot be detected by examining the parts in isolation. And black hole physics has shown us the possibility of wholes (black holes) that totally lose most of the properties possessed by their parts and render the parts in accessible (a black hole has only the properties of mass, charge and spin, regardless of the other properties possessed by the objects that combined to form the black hole). The nature of part-whole relationships in panpsychism certainly bears further study, but merely appears subtle, not incoherent. And the emergent, holistic aspect of consciousness is well known in Eastern thought, e.g. Swami Krishananda says that:
The three states of waking, dream and sleep, through which we pass in our daily experience, differ from one another, and yet a single consciousness connects them, enabling the individual to experience an identity even in the otherwise differentiatedness of these states. Since consciousness links the three states into a singleness of experience, it is immanent in them and yet transcends them, not capable of identity with any of them.
In short, the panpsychist view of consciousness has a long history in both Eastern and Western philosophy, and has no glaring conceptual problems associated with it, the main difficulty with it being that most people in contemporary Western cultures find it counterintuitive. At least one of the authors has found it a useful guide for thinking about the mind, perhaps largely because it doesn't contain any confusing inconsistencies or incoherences that "get in the way" of analyzing other issues involved with machine consciousness, such as reflective consciousness, self and will.
Panpsychism holds that everything in the cosmos has at least a spark of mind, of cosciousness, in it. Quite often – though not always – this perspective comes along with a more religious view, which holds that everything in the cosmos has a spark of God in it (in some sense or another). So that ultimately, everything is part of the Mind of God. I call this “spiritual monism.” Of course this may be interpreted in many, many different ways and it would be out of place to review them all here, but I can’t resist giving one particularly fascinating example.
Around the same time I interviewed Changle, I also did a long and fascinating interview with Lincoln Cannon, the leader of the Mormon Transhumanist Association. As well as advocating a sort of Mormon spiritual monism, he advocats the intriguing position that Mormonism is the “most transhumanist religion” – because it explicitly advocates human beings improving themselves incrementally until they effectively become gods. It also contains the idea that God used to be an ordinary being like us, until he self-improved and became, well, transhuman…
What I found fascinating when talking to Lincoln, was how thoroughly he’s integrated transhumanism – AGI, nanotechnology, mind uploading and all the rest – into his Mormon world view. It’s not as though he keeps Mormonism and transhumanism in different parts of his brain, one personal and one intellectual or whatever – rather, for him they’re all parts of one big conceptual complex! And in this conceptual complex, unlike Changle’s, there is plenty of room for computer programs and robots that have the same kind of intelligence, consciousness and spirituality as humans.
A snippet from our dialogue:
Ben: Oh, and one more thing I just can’t resist asking… In your perspective could an intelligent computer program have a soul? Could it have consciousness? Could an intelligent computer program become a God, in the same sense that a person could? Will AIs be able to participate in the collective deification process on the same status as humans?
Lincoln: In Mormonism, "soul" is used to describe the combination of spirit and body, rather than just spirit. … I think computer programs already have spirits, or actually ARE spirits.
In Mormon cosmology, God creates everything spiritually before physically, organizing and reorganizing uncreated spirit and matter toward greater joy and glory. All things have spirits. Humans have spirits. Non-human animals and even the Earth have spirits, and will be glorified according to the measure of their creation, along with us. Many Mormons also anticipate that the day will come when, emulating God, we learn to create our own spirit children. Spirit is in and through all things. Recall, too, that Mormons are philosophical materialists (not dualists), so even spirit is matter, which God organizes as the spiritual creation of all things. So far as I'm concerned, spirit as described by Mormonism is information, and software engineering is spiritual creation. We are already engaged in the early stages of the creation of our spirit children. Taking a step back, consider how this adds perspective to the problem of evil: what justifies our development of artificial intelligence in an evil world?
Another take on the relation of AGI to the cosmos is contained in the broad philosophy of “Cosmism” – a term originated by Konstantin Tsiolokovsky and other Russian Cosmists in the late 1800s, and then borrowed by myself and Giulio Prisco in 2010 to denote a closely related futurist philosophy, more tailored for the modern era. Rather than positing a fundamental theory about the stuff of the universe or the makeup of mind, Cosmism posits an attitude toward life, technology and the world, which includes an attitude toward AGI and the cosmos and refinement of our understanding of their nature and relationship.
I wrote a short book in 2010 called A Cosmist Manifesto, presenting my views on Life, the Universe and Everything, and drawing together thoughts about AGI and other advanced technologies with my take on Zen Buddhism and other spiritual philosophies. As I state at the start of the book,
By Cosmism I mean: a practical philosophy focused on enthusiastically and thoroughly exploring, understanding and enjoying the cosmos, in its inner, outer and social aspects
Near the beginning of the Cosmist Manifesto is the following list of high-level principles, which was initially written by Giulio Prisco and then edited by myself:
Ten Cosmist Convictions
This is followed by a longer list of a few dozen principles and hypotheses, which are then elaborated in more detail – but, well, if you want to know all that, just read the Cosmist Manifesto book!
In the Cosmist, view AGI is just one (important) part of the ongoing process of intelligence spreading itself through the Cosmos – a process that we and our intelligent software creations are products of, agents of, and passengers upon.
In late 2010 Giulio asked me to give a talk at an online “Turing Church” workshop he organized, on the “Cosmist Manifesto” theme… I spoke off the cuff as usual, but beforehand I wrote down some notes so I’d have them there to look at, just in case I drew a blank during the actual workshop (which didn’t happen, thankfully!). Here are the notes I made, which are somewhat similar to what I actually said:
The relation between transhumanism and spirituality is a big topic, which I've thought about a lot -- right now I'll just make a few short comments. Sorry that I won't be able to stick around for this whole meeting today, I have some family stuff I need to do, but I'm happy to be able to participate at least briefly by saying a few remarks.
Earlier this year I wrote a book touching on some of these comments, called "A Cosmist Manifesto" -- I'm not going to reiterate all that material now, just touch on a few key points.
The individual human mind has a tendency to tie itself in what the psychologist Stanislaw Grof calls "knots" – intricate webs of self-contradiction and fear, that cause emotional pain and cognitive confusion and serve as traps for mental energy. Ultimately these knots are largely rooted in the human self's fear of losing itself – the self's fear of realizing that it lacks fundamental reality, and is basically a construct whose main goals are to keep the body going and reproducing and to preserve itself. These are some complicated words for describing something pretty basic, but I guess we all know what I'm talking about.
And then there are the social knots, going beyond the individual ones… the knots we tie each other up in…
These knots are serious problems for all of us – and they're an even more serious problem when you think about the potential consequences of advanced technology in the next decade. We're on the verge of creating superhuman AI and molecular nanotech and brain-computer interfacing and so forth – but we're still pretty much fucked up with psychological and social confusions! As Freud pointed out in Civilization and its Discontents, we're largely operating with motivational systems evolved for being hunter-gatherers in the African savannah, but the world we're creating for ourselves is dramatically different from that.
Human society has come up with a bunch of different ways to get past these knots.
One of them is religion – which opens a doorway to transpersonal experience, going beyond self and society, opening things up to a broader domain of perceiving, being, understanding and acting. If you're not familiar with more philosophical side of the traditional religions you should look at Aldous Huxley's classic book "The Perennial Philosophy" – it was really an eye-opener for me.
Another method for getting past the knots is science. By focusing on empirical data, collectively perceived and understood, science lets us go beyond our preconceptions and emotions and biases and ideas. Science, with its focus on data and collective rational understanding, provides a powerful engine for growth of understanding. There's a saying that "science advances one funeral at a time" – i.e. old scientific ideas only die when their proponents die. But the remarkable thing is, this isn't entirely true. Science has an amazing capability to push people to give up their closely held ideas, when these ideas don't mesh well with the evidence.
What I see in the transhumanism-meets-spirituality connection is the possibility of somehow bringing together these two great ways of getting beyond the knots. If science and spirituality can come together somehow, we may have a much more powerful way of getting past the individual and social knots that bind us. If we could somehow combine the rigorous data focus of science with the personal and collective mind-purification of spiritual traditions, then we'd have something pretty new and pretty interesting – and maybe something that could help us grapple with the complex issues modern technology is going to bring us in the next few decades
One specific area of science that seems very relevant to these considerations is consciousness studies. Science is having a hard time grappling with consciousness, though it's discovering a lot about neural and cognitive correlates of consciousness. Spiritual traditions have discovered a lot about consciousness, though a lot of this knowledge is expressed in language that's hard for modern people to deal with. I wonder if some kind of science plus spirituality hybrid could provide a new way for groups of people to understand consciousness, combining scientific data and spiritual understanding.
One idea I mentioned in the Cosmist Manifesto book is some sort of "Confederation of Cosmists", and Giulio asked me to say a little bit about that here. The core idea is obvious – some kind of social group of individuals interested in both advanced technology and its implications, and personal growth and mind-expansion. The specific manifestation of the idea isn't too clear. But I wonder if one useful approach might be to focus on the cross-disciplinary understanding of consciousness – using science and spirituality, and also advanced technologies like neuroscience and BCI and AGI. My thinking is that consciousness studies is one concrete area that truly seems to demand some kind of fusion of scientific and spiritual ideas… So maybe focusing on that in a truly broad, cross-tradition, Cosmist way could help us come together more and over help us work together to overcome our various personal and collective knots, and build a better future, and all that good stuff…
Anyway there are just some preliminary thoughts, these are things I'm thinking about a lot these days, and I look forward to sharing my ideas more with you as my thoughts develop
Cosmism doesn’t ask us to commit to viewing the universe as information or quantum information or hypercomputation or God-stuff or whatever – it asks for a broader sort of commitment, to an attitude of joy, growth, choice and open-mindedness. Adoption of this attitude may lead to a variety of different perspectives on AGI as it and related technologies evolve. Science in its current form, and religion and philosophy in their current forms, may turn out to be overly limited for the task of understanding (human or artificial) mind; if so, by actively engaging with the world and studying and engineering things, and by reflecting on ourselves carefully and intelligently, we will likely be able to discover the next stage in the evolution of collective thinking…
Finally, at risk of leaving you thinking I’m totally nuts, I’m going to share with you some more recent thinking I’ve been doing, going beyond the ideas in the Cosmist Manifesto in a sense. These are half-baked ideas at this stage -- but who knows, maybe some reader will encounter them and have and publish an idea that will help my own thinking along.
The great quantum physicist David Bohm, when he turned more philosophical in his later years, posited the notion of the “implicate order” – i.e. an aspect of the universe that implicitly underlies all things, but isn’t in itself scientifically measurable or sensorially perceptible. The explicate order that we can see and measure, in some sense emerges from the implicate order (and then folds back into it, contributing to it).
There are relations between this notion and panpsychism, in that the “spark of mind” implicit in something may be equated to (or at very least associated with) the “implicate aspect” of that thing. Bohm also connected the implicate order with quantum mechanics, though in a manner that I never fully understood from his writings. Sometimes it seemed he wanted us to look at quantum logic as a sort of interface between the implicate and explicate orders. Not that the explicate order uses classical logic and the implicate order uses quantum logic; but rather that quantum logic captures some aspects of the explicate/implicate interaction that classical logic misses.
Recently I’ve begun thinking about the implicate order from a different perspective, and looking at models of the implicate order as a “logic of questions” rather than a logic of answers. I’m experimenting with modeling the implicate order as something I call QP, a questioning process – not a process of questioning anything in the everyday world, but rather a process of questioning itself. To quote a manuscript I wrote on this a while ago (tentatively titled “?”),
If I had to summarize QP in a brief phrase of (almost) ordinary English, I suppose I’d go with something like “the process of a complex, autopoietic pattern/process-system growing and developing via self-referentially (and joyfully, autonomously and interconnectedly) self-questioning.” Sorry if that sounds like gobbledygook! It makes a lot of sense to me, and I hope it will make a lot of sense to you after you finish the book!
QP… The process of questioning. Questioning everything, including the process of questioning everything – and so forth! What I’ve been studying is how one might model the universe as something that fundamentally emerges from this kind of self-questioning process.
Another famous quantum physicist, John Wheeler, speculated about the possibility of deriving quantum mechanics and general relativity theory (two great physics theories that remain un-unified, leaving modern physics in a state of unacceptable contradiction) from some sort of statistical analysis of the space of logical propositions. So that physics would emerge from a “pre-geometry” made of logic. My QP approach is actually somewhat similar, except that I’m looking at a logic of questions rather than a logic of answers… And thinking about the emergence of the mental as well as physical universes.
Science, after all, is a questioning process – it’s about questioning the universe, and about questioning science itself. Every scientific theory brings new ideas and tools used to question its own self, which is why every scientific theory ultimately leads to its own destruction/transcension. The scientific method itself is not a constant, it’s constantly being revised, because it invariably leads to relentless questioning which undermines its own foundations.
And while institutionalized religions may seem to have more to do with obeying than with questioning, Gnostic Christianity was precisely about trying to question and understand everything for oneself, to know “God” directly. Jainist Buddhism and Sufism shared the same aspect – Jainists were trained to greet their every thought, belief or concern with the response “Not this! Not this!”
In the QP perspective I elaborate in “?”, the questioning processes of science and gnostic religion are examples and instantiations of the broader questioning process woven into the fabric of existence.
As I write this, I’m reminded of my Hong Kong friend Gino Yu, who likes to throw up his hands and grin and ask his friends, gesturing at the world around, and at himself and the rest of us: “What is this??” When I mentioned QP to Gino he simply said “Oh, that’s just the Socratic method.”
What is this, indeed??
AGI will not tell us “what is this”, in any definitive way. AGI will get rid of some of our old questions, and replace them with new questions!
The Ten Cosmist Convictions describe the future of humanity – the future of the ongoing growth and unfolding of the universe, the ramification of the Cosmos into configurations as far beyond current human life as we are beyond the proto-life in the oceans of early Earth, or the lifeless molecules spinning through the void of space prior to the formation of planets. The QP train of thought, tries to dig deeper into this growth and unfolding process – looking at the ongoing unfolding and growth of the universe as a process of relentless self-questioning. It looks at the universe as a big mind, which is constantly asking itself “What is this? What am I doing?” and in doing so is changing itself, bringing about new forms like planets and proto-life and humans and AGI systems.
But please be assured: You don’t have to follow me into these peculiar thought-domains to appreciate my AGI work, or my projections about the future of technology! Any more than you have to agree with my panpsychist view of consciousness to think a completed OpenCog system will be conscious – you may have your own conception of consciousness and your own way of applying it to OpenCog.
One thing I’m quite sure of is: None of us humans really knows what’s going on!
So, at the end of this romp through strange ideas and perspectives, what’s the take-home point about AGI and the Cosmos?
I’m not a Mormon – I’m not religious at all… I’m not even really a Jew, in terms of beliefs, in spite of my Jewish racial background. And unlike my colleague Selmer, I don’t place much stock in the hypercomputable Soul.
I feel like existing religious and spiritual notions of “God” and “Soul” are getting at important aspects of the universe which science misses (and probably will always miss, due to its foundation in data-sets comprising finite sets of bits); yet I also feel like they’re tangled up with a lot of superstitious beliefs and historical “cruft”, which makes me want to consider them more as general inspiration than as foundations for my thought and understanding.
I don’t know whether classical or quantum information is ultimately a better model of the universe; I think there are a lot of unknowns thereabouts, which will be resolved as science, math and philosophy unfold.
I strongly gravitate toward some form of panpsychism, but I’m not exactly sure what kind. When consciousness theorist Stuart Hameroff said to me “I don’t think an electron has consciousness but I think it has some kind of proto-consciousness”, I felt like I probably disagreed -- but also wondered whether we were just getting tangled up in webs of language. I tend to agree with Charles Pierce, Spinoza, Galen Strawson and others that drawing a rigid distinction between mind and matter is ultimately logically incoherent.
I find Cosmism an agreeable attitude toward myself and the universe, but I’m also aware of its limitations. I can’t help questing toward a more fundamental understanding of the Cosmos, even though I suspect my limited human brain isn’t going to be able to understand the Cosmos too well, and that AGIs, brain enhancement and the like will give us much deeper and better perspectives. I’m currently somewhat bemused by modeling the universe as a process of self-questioning, but unsure how far I’ll be able to take the idea.
Often I find myself holding back the part of my mind that wants to spend a lot of time theorizing about such things, because I feel it’s more important to work on building AGI! Understanding the world is important to me, but the choice is between trying to understand it directly using my human brain, or building an AGI mind that will be able to help me understand it far better than my human brain is capable of!
One thing that struck my mind when writing this article, was the old saw: People can adapt to almost anything… It occurred to me that it makes sense to generalize this to: Robust conceptual frameworks can adapt to almost anything. For example, perspectives like Mormonism and Buddhism and panpsychism can adapt themselves to make meaningful statements about AGI and Singularity, notions that were completely unanticipated (and would have been largely incomprehensible) at the time these religions and philosophies were founded. This says a lot about the adaptable nature of human mind and culture (which emanates, of course, from the adaptable nature of the biological world from which these emerged).
Ultimately, it seems, human conceptual and belief systems are able to adapt to all sorts of new realities, including the Internet and mobile phones and birth control, and soon including AGI and molecular assemblers and cyborgs and what-not. Spirituality and religion embody key aspects of human nature, such as the quest for deep fundamental understanding of the universe, and deep communion with the universe and with other minds – and these quests will keep on going as technology advances, not so much colliding with scientific and technological growth as synergizing with it. Technological revolution will foster a continuous unfolding and expansion of spiritual understanding and experience, which may end up taking a tremendous diversity of different forms – far stranger (and perhaps far deeper) than wild ride of ideas we’ve rolled through in this article.
And so, as I already emphasized, I can make one statement about the contents of this article with great confidence: Every single idea posited here will appear to us rather silly and limited, once we have expanded our world view via intense communication and possible fusion with trans-human AGI systems. In other words: Where the cosmos is concerned, we humans don’t understand much! We understand more than non-human animals do, and more than pre-civilized humans did… But the AGIs we create will understand much, much more. Which is not to say that any mind will ever come to a complete and thorough understanding of intelligence and the cosmos – maybe it will, maybe it won’t; that’s among the many things humanity currently has no clue about.
Just as the contradiction between quantum theory and general relativity (gravity theory) tells us that current physics MUST be seriously incomplete in some ways… So, I feel, does the human race’s wild diversity of confusingly contradictory and complementary views on the cosmos tell us that we really don’t understand this world we’re living in, and our intelligence’s place in it, and the roles and properties that will be played by AGIs we create. We don’t necessarily understand what our current activities are going to lead to, any better than the “cavemen” who first created language understood that their invention was going to lead to Dostoevsky, differential calculus, Prolog, Google Talk and “Neuroscience for Dummies.”
The various concepts and perspectives we’re currently experimenting with (and in some cases pouring our hearts and minds into) – classical and quantum information theory, hypercomputing, implicate orders, Mormonism and Buddhism and Cosmism and what-not – all exist within the limited scope of current human individual and collective mind. These perspectives may help us cope with the changes that we are now wreaking in ourselves and the world, via creating AGI and a host of other technologies; and they may well affect the particular nature of the technologies we create and the future world these technologies help create. But these perspectives will then be subverted, and made to seem quaint and ridiculous, by the greater intelligence to which they will lead. And yet, we have to work on refining and elaborating our current perspectives on the world – knowing they will seem absurd and limited in hindsight, from the perspective of greater intelligence – because this refinement and elaboration is part of the process of bringing about said greater intelligence.
Sign up for the Humanity+ newsletter:
Joining Humanity+ as a Full, Plus or Sponsor Member enables you to participate in Humanity+ governance and decision-making - an important role in the growing Transhumanist movement. It also, of course, gives you the opportunity to support us in the work Humanity+ does!