A Commentary on "Six Theses on Existential Risk"
In Six Theses on Existential Risk I copied a number of remarks that I had posted to Twitter as they occurred to me, and I promised to return to them to "put some flesh on the bare bones of the ideas represented." This post will attempt to keep that promise by offering a commentary of sorts on my compressed thoughts of the other day.
Theses demand an exposition to fill them out and fill them in; as bare theses, they express little except the satisfaction of the author to have summarized a number of ideas in definitive formulations. In just this way -- personal satisfaction in sententious expression -- I have tied together my own perennial (and possibly also parochial) interests in existential risk, the big picture, non-constructivism, infinitary reasoning and a variety of (loosely) related ideas. But linking a number of ideas together is not enough; one must show that the linkage constitutes a whole greater than its parts, or one is indulging in mere eclecticism.
What follows is my attempt to show, by (as I said above) filling out and filling in, that my Six Theses on Existential Risk are more than an eclectic grab bag of ideas and do represent more than a mere summation.
Six Theses and Commentary
1. The marginalization of existential risk issues involves the pathologicalization of big picture, long term thinking.
Everyone who has tried to talk about existential risk knows that as soon as you talk about the "big picture" -- and existential risk is part of the big picture -- people turn off for a variety of reasons, including their own parochial assumptions, lack of interest in things that don't touch their personal concerns, and an instinctive objection to what strikes them as "science fiction" (I have personally been told that my ideas were "science fiction" so I am particularly sensitive to this charge). One is likely to be branded a "chicken little" because one appears to be perpetually saying that the sky is falling.
I've thought about this problem a lot, and certainly one of the central issues (at present) for existential risk is simply making people aware of it, making it real, and getting people to understand that we can begin to take action without being thought a crackpot or a chicken little. What is needed, in short, is existential risk consciousness raising.
The history of consciousness raising is a chequered one. Marx believed it was essential, that if the revolution is to come about, that there must be a consciousness raising on the part of the proletariat. I would argue that this never successfully occurred (and I have obliquely written about this in The Work of Life).
By contrast, a large number of mostly left-of-center political and social issues that were the object of explicit consciousness raising in the late 1960s and early 1970s appears to have had a lasting influence on the development of Western culture, as many of these issues have become mainstream concerns since that time. What this shows us is that explicit attempts at consciousness raising may be successful, but are by no means guaranteed success.
The long and difficult word "pathologicalization" means to make something pathological, and in the present case what I am talking about is how a concern for existential risk has come to be perceived as pathological, and the marginalization of existential risk concerns essentially means their pathologicalization. Along with existential risk we lose all of those conceptions tied to the big picture and la longue durée that inspires and motivated existential risk mitigation.
2. The pathologicalization of big picture, long term thinking implies the normativity of short term thinking.
In so far as a concern for existential risk is pathological, the absence of any concern for existential risk is normal. Now, since existential risk is something that (generally speaking) only appears in the big picture, and over la longue durée, the exclusion of existential risk concerns means the exclusion of the big picture and la longue durée and confinement to local, parochial, and provincial concerns.
One never need say outright that short term thinking is to be taken as normative, when all one's actions, ideas, plans, and orientation are directed at the short term, and the short term is divorced from any connection to the long term in which it is contextualized. The example of short term thinking is sufficient to establish its normativity.
With the exclusion of existential risk, the big picture, and la longue durée there remains nothing else but risks that never rise to the level of existential threat, the small picture, and the short term. This becomes normative through the method of isolation.
3. The normativity of short term thinking implies, and is in turn implied by, finitary assumptions.
Less than existential risks, small picture, and short term ideas all make finitary assumptions, and there is a sense in which this is entirely legitimate, since everything in human experience is finite. Finite assumptions are, essentially, human assumptions. And, until Cantor's work, philosophers and mathematicians alike held that the infinite was unknowable and not an appropriate object for human contemplation, except under the aspect the infinity of the divine, which was universally regarded as unknowable and an impenetrable mystery.
Although Cantor's motivation for the formulation of set theory and transfinite numbers was clearly theological, the upshot of Cantor's work was a mathematical theory of the infinite that made it possible for human beings for the first time in their history to think about the infinite in a rigorous and systematic way.
While I understand that my interpretation of Cantor's work is way out of the mainstream, as I see it Cantor's work is a seminal moment of secularization, which has so relentlessly accompanied the expansion of science and knowledge, and rendered all theology of "god of the gaps" arguments -- with the gaps continually shrinking.
What this means is that we are in a position to contextualize our finite, human concerns within an infinite context that we can now talk about with some degree of rigor. While philosophers have been slow to fully exploit the consequences of Cantor's work (though, to his credit, Bertrand Russell saw these implications immediately and often applied them in his work), this remains a promising field of philosophical endeavor.
4. Finitary assumptions ultimately must be contextualized in infinitistic reasoning, which is inherently non-constructive.
Once we have mapped out the infinitistic context of our finite concerns, we will no longer have any philosophical or scientific right to confine ourselves to finite concepts, because finite concepts beg the question of their conceptual context, and, according to Frege's context principle, apart from their context (which ultimately must be infinitistic) they are ambiguous.
In The Foundations of Arithmetic Frege formulated what is often called his context principle. In his Introduction to The Foundations of Arithmetic, Frege wrote that, throughout his inquiry he kept to three fundamental principles, including this:
“never to ask for the meaning of a word in isolation, but only in the context of a proposition”
Gottlob Frege, Foundations of Arithmetic, Evanston, Illinois: Northwestern University Press, 1980, Introduction, p. x, translated by J. L. Austin
Frege formulated his context principle as a principle of semantics, but we can immediately see that there is an implied ontological context principle, and it is this: Never ask for the existence (or meaning, or value) or an object in isolation, but only in the context in which it exists. In other words, every individual (object) in the world is definable only in terms of a totality of which it is a part. (This is essentially the position of Carnap’s conventionalism, but a detailed explanation of this would run for pages, so I will not go into this here.)
We think of the doctrine of terms possessing meanings independently of their use as a typically realist position, but context principles that relativize meaning to use are essentially impredicative, hence essentially non-constructive, hence, in a sense, realist. Frege’s context principle can thus be understood to underline his non-constructive realism.
The context principle is a principle of impredicativity (and therefore non-constructive in at last one sense of the term "constructive"). The world itself is impredicative, that is to say, non-constructive.
5. Thought and action consistent with an infinite horizon must employ non-constructive reasoning, which is also non-mechanizable reasoning.
Anything less than contemplating the human future in the context of an infinite horizon means that we consign ourselves to oblivion, and this is a form of nihilism, no matter how subtly defended or how deeply felt. Since we are not nihilists, we cannot consign ourselves to go gentle into that good night.
Our thoughts, and our actions that follow from our deliberations, must extend upward and outward to the infinite horizon that is the context of our thoughts, our actions, and our future, or the future of whatever is the successor of ourselves or our civilization. Thought cannot rest without pushing on to see the whole of which we are a part, and so by a process both incremental and relentless we fumble our way toward the infinite. This is the intellectual realization of mathematical induction: for any n value we push on to n + 1.
All infinitistic reasoning, including even mathematical induction which was approved of by the Greeks, who eschewed the infinite and identified the limited with the good, remains beyond the capacity of machines at the present time. While mathematical induction is constructivistically acceptable (for most constructivists, or at least I assume that it is), the conceptual leaps of Cantor's theory of the infinite, which can ultimately make sense of the infinite context of the world, are most definitely non-constructive and most definitely non-mechanizable.
While the future evolution of machines may eventually result in machine consciousness, and machine consciousness may allow machines to conceive mathematics as human beings conceive mathematics, if machines do reach that point of development their capability to conceive non-constructive ideas will result not from dramatic increases in computing power, but only because increases in computing power allow machines to be more like us.
Artificial intelligence simpliciter, short of actual machine consciousness, i.e., subjective awareness, is not sufficient to arrive at non-constructive conceptions. No matter how powerful any computer is now or in the future, as a computer it cannot compute non-computable functions. Machines might someday conceive non-computable functions if they becomes conscious, but that isn't the same thing as computing the non-computable.
The day of machine consciousness, however, appears to be some way off yet.
6. a. Our machines will not save us; mechanization has no soteriological function.
b. We must save ourselves or perish.
I wrote above that I was particularly sensitive to charges of my admittedly speculative ideas being labeled "science fiction" and I can also say that I am sensitive to charges of whitewashing mechanization and industrialization, so I usually find myself on the other side of the argument as compared to what I seem to be advocating above.
In fact, I think that one of the perennial sources of the rejection of rational planning for existential risk -- an attitude of utterly giving oneself over to the human condition and indulging oneself fully in human pleasures and sorrows for the time one has on Earth, which might also be nihilistically summed up in the saying Live fast, die young, and make a beautiful corpse -- is a certain kind of "humanism" (which I must put in scare quotes, because no one ever uses "humanism" in the sense in which I am here proposing) that rejects technologically sophisticated responses to existential risk made possible by our industrialized civilization, but which were not available to any previous, earlier iteration of human civilization.
This is a difficult conception to formulate (here I fall fall short of a definitive formulation, so I will need to dedicate further intellectual labors to clarifying this) because this particular conception usually resides in the mind below the level of explicit consciousness, but the implicit rejection of the machine, and of mechanization (and automation) generally speaking, can be interpreted as a defense of human nature and the human condition in a peculiarly contrarian spirit of "defense."
In so far as one thinks that industrial-technological civilization is, on balance, a good thing, one is put in a position of defending mechanization, automation, and industrialization, so that is why I say that I usually find myself on the other side of the argument, because I am willing to defend industrial-technological civilization (not that it needs my defense), but I am here saying that our machines will not save us, and that, if we are to survive, we must save ourselves.
Our machines and our technology are tools for us, but (as against the singulatarians) they will always be tools, and we human beings will not be able to surrender our agency, our responsibility, or our efforts to machines, or indeed to any surrogates whatsoever (I add this qualification for the crackpots who think that space aliens are going to come to save us from ourselves).
Important note: we don't "save ourselves" by becoming virtuous, or by obeying rule and laws, or my becoming intensely religious. We save ourselves by taking action to assure a human future in perpetuity. That is to say, existential risk mitigation ultimately aims at human (or transhuman, or post-human) sempiternity.
I hope that the reader who has made it this far has come to realize through my admittedly rambling commentary on my six short theses -- all of them spelled out within the 140 character limit of Twitter -- that much is implied herein and that volumes would be required to spell it all out.
It would be only a matter of having the spare time and one could easily write a book, or a series of books (perhaps one for each thesis), developing the ideas above. Perhaps, fate willing, I will do so. In the meantime, I continue to develop my ideas, which all bear upon the above because they are all intimately connected to the intellectual milieu from which my theses came.
