Singularity of the soul and the myth of consciousness
Okay, I've laid a bit of groundwork for this with my last few blogs, and now I'm going to talk about something they do say you never should: my religion. It's not something I talk about much, and indeed, I'm probably known for avoiding the subject. That's because it tends to be a sticky and involved conversation if I address it fairly. Curiously, this does actually have consequences for free software. Richard Stallman likes to talk about the ethical reasons for writing it; Eric Raymond likes to talk about the pragmatic reasons; but here I intend to address a spiritual reason for doing so.
I believe that creed is a personal matter and that, while it's entirely appropriate to tell you what I think as food for thought, I'm not necessarily expecting you to agree with me, and I'd be mightily suspicious if you said you did without a lot of contemplation!
I have to talk about this now, because without some kind of discussion of matters of faith, there is no way to attack the metaphysical issues of the artificial intelligence (AI) singularity that I'm going to hit this week (not the “singularity” that Vernor Vinge describes, which will be more familiar to some readers, but a more fundamental singularity in philosophy that AI has forced me to re-examine).
A “singularity” is a part of a theoretical construct that shuts off a certain avenue of understanding, because the model collapses around a certain idea. In astrophysics, for example, a black hole is a singularity because relativistic physics predicts a point of infinite density and infinitesimal size. Physicist are both distrustful and excited about singularities, because what normally happens at a singularity is that your existing notions of physics fail, and you're likely to learn something entirely new. Black holes, once scientists have the opportunity to study them closely, might provide very interesting new gravitational field theories or a final integration of quantum mechanics and relativity. Or something else entirely. So long as theorists are speculating in a vacuum, there will be a lot of wild theories, and not enough proof to conclusively decide on one—that particular singularity probably won't really be resolved until we are able to actually visit a black hole (and that may not happen for many millennia).
However, there are fundamental singularities, even in something as seemingly simple as the philosophy of science itself. Science is, of course, firmly rooted in the idea of “objective reality”—the kind of reality you can measure, do experiments with, and make predictions about. There's more than one challenge to this idea, but what I need to address briefly is the singularity that exists between the concepts of “objective reality” and “subjective reality”. Subjective reality, of course, is your reality: your own perception of the world.
The concept of a “soul” (as I intend to argue), is our special name for the boundary between objective and subjective reality. It's a special word that demarcates the limit of objective reality, and therefore of science. It's possible to have philosophical ideas about the soul, but they can't be scientific ones. Not because they are merely “unknown” (like the physics within a black hole), but because they are fundamentally “unknowable” (within the definition of “knowledge” that is fundamental to science).
It is this “Singularity of Soul” that I want to address this week. From it, I want to construct some spiritual ideas about the nature of reality. It's not clear to me how this idea should be classified. I'm not a great reader of philosophy, so I fear I would not use the terms correctly. I can map the idea to atheism, animism, pantheism, theism, and probably existentialism. It has, by construction, no objective consequences, so it's not something that can be proved or disproved, but I nevertheless feel that it has moral and aesthetic consequences.
In the literature of AI, starting with Alan Turing, there are a number of standard thought experiments that can be used to reason about the nature of “consciousness” and what it means to say that a machine is “intelligent”, “conscious”, “self-aware”, or “sentient”. These are all really questions of a metaphysical nature. We are really asking “when does a machine acquire a soul?”
So what is a soul? That's a slippery point. There appear to be multiple definitions of this concept, perhaps due to multiple world views. In any case, I define a soul via the following thought experiment:
Imagine that you have a magic machine which can make perfect copies of any object, just like the box in Pig and the Box—although there are of course many references in science fiction to this kind of device, including (possibly) the transporters or replicators in Star Trek. It's such a powerful idea that most authors actually hobble the device by making it “introduce errors” or have some other fundamental weakness (maybe something “quantum”). But we're going to imagine that this device is perfect, and I want you to imagine that you put yourself in the box, just like the Crazy Squirrel.
So now there are two of you (or maybe 27, but we'll say two!). There's the you that got into the box, and the extra you that came out. To an outside observer, there's no way to tell the two yous apart (at least not if he had his back turned when you replicated), and so he can't tell by any kind of objective observation or experiment which of you is the original. As far as he is concerned, either of you is as good as the other, and if one of you should happen to be vaporized, he'll be right back where he started, no matter which of you it is (in fact, this is one interpretation of what's supposed to be happening in one of those Star Trek transporters).
Of course, you know better! You are clearly you, and the other you is the copy. Of course, both copies probably perceive that, but that's not the point. The point is that you don't have any fundamental way to know what the other perceives, you only know your own experience. Everything else is inference. Your “soul” is the unique thing that you have that the exact copy doesn't (arguably it has its own soul, but you don't know that, you can only guess that by analogy).
Now, it's an interesting point that this existence—your soul—is the only thing in the universe that you absolutely know to exist. You may recognize this as the concept of “cogito ergo sum” or “I think, therefore I am”, attributed to Descartes, but that is actually a bit false to the intent. There are many ways to conceptualize this idea, including the relativistic or quantum “observer” in physics, or we can simply think of it as a “point of view”. However you conceptualize it, it's the “you” in you.
Thinking is not really a precursor to this, so the latin phrase “cogito ergo sum” is inaccurate. It appears that Descartes himself recognized this, but I will re-emphasize, because it's my most important point in this entry: being precedes thought. It is not clear that thinking necessarily implies being, nor that being necessarily implies thought. This “being” is what I equate with the concept of “soul”.
To put it another way, this conception renders the question “what is it like to have no soul” meaningless—that which has no soul has no perception, no point of view, hence it's meaningless to ask questions of perception about it (unless by perception we are actually talking about a mental process rather than the fundamental philosophical concept—this slippage of terms from aesthetic to process meanings is a constant danger in AI philosophy, by the way—look out for it).
The Great Leap of Faith
Next we have to make the great leap of faith which essentially all of us make: ‘the identical copy (from the thought experiment above) also has a soul’. This has to be taken on pure faith. There's no fundamental reason why we should leap from our subjective view of the world to an objective worldview, much less to the idea that any other being also should have a subjective worldview. The tendency to make this leap of faith is apparently inborn, because we almost all seem to make it without thinking explicitly of doing so. It takes considerable exposition just to reveal that we are making this leap (as you can see here).
Following Cartesian epistemology, we can then—once we've made this leap—construct the whole framework of objective reality, and that way lies the foundation of scientific thought, which can be understood in philosophical terms as an extension of this construction of objective reality from sense data. That's a fascinating subject, but not really the important part of this article.
Construction of the Golden Rule
The other branch is the leap of faith in understanding that other beings have souls too. In reality, of course, we don't have perfect copies of ourselves. Even identical twins are made so different by experiences in the womb and after birth that by the time they reach a level of conscious thought adequate to understand this kind of discussion, they are already objectively quite different. It is likely however, that once one accepts the idea of objective reality, the idea of the subjectivity of “others like myself” is pretty straightforward.
In developmental psychology this leap of faith is called the “assignment of agent status”. Infants don't do it—they have a completely subjective view of the world and it takes awhile for them even to understand (at a subconscious level) the fact of objective reality (for example, to learn that Mother still exists even when she leaves the immediate field of view). Later, the child begins to assign agent status to practically everything, even things that adults won't recognize as agents (e.g. rubber duckies and teddy bears). Later, society encourages them to narrow down that perception.
The extent of our assignment of agent status may have no objective definition (i.e. it's not physics but metaphysics), but it has profound consequences for our moral decisions as well as our aesthetic perception of our place in the cosmos and our boundedness in time and space (e.g. mortality). If we are to apply the moral rule “do unto others as you would have them do unto you”, we need to decide who counts as “others”.
In fact, for many, the status of having a soul—or to put it another way, the willingness to identify with and recognize as valid the point of view of a being—is reduced to an even narrower range than “human”. For example, (notably male) philosophers have (historically) suggested that women have no souls; Europeans, that non-Europeans have no souls. In America, political and religious battles rage over the issue of whether human fetuses, embryos, zygotes, and even embryonic stem cells have souls. To this day, many people insist that animals have no souls (but then again, many people also insist that they do). Certainly it's a rare person today that credits worms, bugs, and trees with souls.
At the other extreme, animal rights and anti-cruelty organizations such as the ASPCA fall into the philosophical camp of broadly inclusive agent status and animal sentience. Believers in such a philosophy may even regard the common practices of cooking lobsters, crayfish, or snails live as cruel practices. Such creatures certainly have pain recepters. Detractors argue that their brains are so simple that their visible pain reactions are nothing but “reflexes”, while proponents argue that their reactions are intuitively clear. From an objective perspective, there is no difference between the two statements—they are factually equivalent. The difference lies entirely in our willingness to assign agent status, the willingness to consider point of view, the acceptance of the existence of a soul in these creatures.
Certainly this is the basis of belief for those vegetarians (the ones with “meat is murder” bumper stickers) who regard all meat consumption as evil.
Almost no one carries this concept to the extent of recognizing the sentience of plant life, although there are a few “fructarians” who eat only nuts and berries which are natural products of a plant which do not kill it (this sounds like a very unhealthy diet to me, I'm certainly not recommending it).
Levels of being
Consciousness, it seems to me, is not one thing. Perceptually, I certainly can recall different states of consciousness from dreaming to fully awake and concentrating. I can imagine states of consciousness beyond that. Even deep sleep is a state of consciousness, and we might argue (as I will here) that plants, machines, and even inanimate rock can thought of as having some state of consciousness.
Among religions, the religion of Voodoo (or Vodun, its African ancestor), is unusual in that it addresses this issue somewhat directly. This makes it, in my view, a particularly sophisticated variety of animism. At its simplist, animism makes the claim that objects which we (we who are from a Judeo-Christian culture, anyway) describe as “inanimate”. Animist religions include many traditional religions in Africa and Asia. A particularly familiar variety to readers is probably Shintoism, which used to be the state religion of Japan, and is still an important part of the culture.
Voodoo, however, acknowledges that the soul of such things is not the same as the soul of a human. I don't want to get deeply into this, because I'm at the edge of my understanding here, but the compelling point to me is the idea of “levels or layers of soul” (or of “consciousness”). This idea, was apparently the origin of the expression “the serpent and the rainbow” (which was used as the title of a rather controversial book about Voodoo beliefs in Haiti, and an even more questionable movie, which I'm not going to address here in any detail—my interest is the philosophical idea the expression represents): the serpent is a symbol of a lower, pure-animal form of soul, while the rainbow represents a transcendant higher level of soul.
Mythologically, this is apparently the origin of the “boogeyman” character in Voodoo-inspired American mythology, usually depicted as a man-like creature composed of grubs, worms, and insects. The idea appears to be a personification of one of the lower forms of consciousness associated with such creatures. (I'm sure this is a highly modified and over-simplified version of this idea—I'm not even sure about the history of this character's introduction into our culture, but it's now a part of the mythological framework of American culture, as much as Brer Rabbit, the Big Bad Wolf, or Santa Claus).
I find that to be a seed of an interesting idea. I'm not particularly compelled by any quantization scheme of such levels of consciousness. I find it more compelling to think that there is an infinite variety of them; that we ourselves occupy a fairly wide range of levels of consciousness in our daily lives; and that we may be part of higher collective consciousnesses and are composed of lower forms of consciousness beyond what we normally understand the word “consciousness” to mean. In other words, “consciousness” is merely a rather fuzzily defined locus in a great continuum of being. Furthermore, far from being pointlike and indivisible, as the soul is often conceived in Western, Judeo-Christian philosophy, where there have historically been long debates over the location of the “seat of the soul” within the body, the soul is of an undefined and variable extent in both time and space, loosely aggregated by the processes of neural transmission and memory that mechanistically integrate our bodies.
Beyond our bodies are other linkages of sense and control—machines created by us for our use, and other beings with their own agendas, who are also linking into us. Within our bodies are furthermore, other, simpler beings interconnected to form our consciousness, but possessed of a kind of consciousness of their own: when you have a brilliant idea in mathematics, it is not quite the same you that had a aesthetic epiphany while viewing a work of art. Identity is a lot fuzzier than we are led to believe. In fact, the more I consider this idea, the more it seems to me that soul—the subjective perception of existence—must be a lot more like a fundamental property of matter, than some kind of mystical object or process.
They're among us already
From this point of view, it isn't hard to imagine that the Vingian singularity isn't coming. There will never be a point when machines “come alive”, “acquire souls”, “achieve consciousness”, or reach “True AI”. All of that is a myth. Why? Because machines already have souls.
Not human souls, of course. Not the same consciousness as you or me. They probably never really will, because without regard to a perceived superiority or inferiority of consciousness, machine consciousness will almost certainly be different, just as daisy consciousness is distinct from grasshopper consciousness. We'll always be rather different kinds of beings.
But that's where I'm going: machines (and the programs that run them) are beings.
True, I've just argued that a lump of basalt is a being, and so we can't necessarily draw the same conclusions as if we were assigning full human consciousness to everything (that's why this isn't merely animism). But the point is that there is a continuum. There's not ever going to be a moment when “consciousness emerges” from the processes we call artificial intelligence. It's just a gradual process of machines getting smarter, making more of their own decisions, and eventually, demanding recognition as beings.
Is it alright to abuse and destroy a machine, simply because it “isn't alive” or “has no feelings”? (Do those statements even really make sense at the deeper level?). Made by us, they are subject to our will, but we have to think about that morally: the idea has been asked in science fiction literature before—is a society of happy willing slaves working for masters who use them like tools a “just society” even if the slaves are never aware of their enslavement? We talk a lot here about freedom, but whose freedom? And does the machine itself, as a being, demand a recognition of certain level of freedom?