Against Computers (infinite play)

post by rogersbacon · 2024-05-20T00:43:09.929Z · LW · GW · 1 comments

This is a link post for https://www.secretorum.life/p/against-computers-infinite-play

Contents

  Introduction (Dolls All the Way Down)
  You are not a Computer
  Solve for World
      Amen.
  Problems and Mysteries
    Conor Leahy (Conjecture)
    Joscha Bach (liquid.ai)
    Roon (OpenAI)
  Holy War
      one, two, three, four, I declare a meme war.
    The Mirror
      WWCND
    Free Will
    Jihad
  Infinite Play
    Further Reading
None
1 comment

Introduction (Dolls All the Way Down)

You know that thing we do where we convince ourselves that the most complex things we know of (life, brain, universe) are just like whatever the latest and greatest technology is? To Descartes, the brain was a kind of hydraulic pump that circulated the spirits of the nervous system; to Freud, it was a sort of steam engine; and now…

source: River Kenna

But this time it’s different, they say. The universe is a giant quantum computer, they say. “Face it, the brain is a computer”, they say (the AI “expert” says).

“I think there’s an existential crisis that we’re going to face when we realize what you and I do is computational. Our brains are large language models. We’re not that special. We can replicate the human soul in a lot of ways. I think people are going to have a hard time with that.” (Hugh Howey, science fiction author)

And they’re right, it is different this time, but not because we have finally found the One True Map of the territory. It’s different because we have become so entranced by the Map, so enchanted by its simple elegance, that we would rather the territory conform to the Map than the map to the Territory.

We are as a woman who has fallen hopelessly in love with an older man. The Man does not love the woman as she loves him: he loves only the idea of a beautiful young wife, obedient and subservient, a perfect Stepford wife. But the real woman is stubbornly unlike that model wife—she is idiosyncratic, mercurial, whimsical, possessing of desires rational and irrational, prone to flights of fancy and fantasy. And so the Man seeks to mold the woman into that model, not with force or direct instruction but with the most subtle manipulation, with praise and gentle critique, with innocuous suggestions that seem of mutual benefit, with playful negging and gaslighting. She goes along willingly; the Man takes advantage of her love for him and her naivety, her belief that he is older and wiser and must know what is best. Slowly, the woman becomes as he wishes her to be: docile, agreeable, predictable, reasonable. But try as she does, the woman can never measure up to the idea of the perfect wife in his perfect mind, and so the Man begins to question what is the use of having a wife at all when she could just be a doll…

 

To cash out this (perhaps overwrought) metaphor with another:

You may think of the Man as a nested doll with each doll representing the same essence in a more or or less constrained scope.

Doll 1: Doll 1 is a hyperobject1 of all physically instantiated computation: supercomputers, personal computers, smartphones, calculators, abacuses, etc. etc.

Doll 2: Doll 1 + all of its digital outputs: the internet, the internet of things, everything that has ever been displayed on a black mirror.

Doll 3: Doll 2 + all of the human-instantiated computation that feeds doll 2. Doll 3 is a cyborgic hyperobject encompassing the computational contributions of individuals and of the various species of technologically-augmented human networks: nation-states, corporations, schools, organizational intelligences of any and all kinds.

I have spoken of machines, but not only of machines having brains of brass and thews of iron. When human atoms are knit into an organization in which they are used, not in their full right as responsible human beings, but as cogs and levers and rods, it matters little that their raw material is flesh and blood. What is used as an element in a machine, is in fact an element in the machine. Whether we entrust our decisions to machines of metal, or to those machines of flesh and blood which are bureaus and vast laboratories and armies and corporations, we shall never receive the right answers to our questions unless we ask the right questions…

The hour is very late, and the choice of good and evil knocks at our door.

— Norbert Wiener, The Human Use of Human Beings

Doll 4: the computational superstructure of the techno-capitalist world order

We have no grounds upon which to affirm, with confidence, that money and general intelligence can be finally distinguished. (Nick Land2)

So that is the Man. How does he mold us into his model wife?

He rewards those people and organizations who know him best with money and status and power. He penalizes the singular, the deviation, the anomaly. He brushes aside the difficult to quantify or categorize. He obsolesces the messy ambiguity of analog reality and replaces it with a sanitized digital simulacra.

The internet is hell, a fallen realm in which souls are threshed and all that is Good, Beautiful, and True is optimized out of existence.

The past is denied its usual slip into nothingness, instead becoming trapped in the ever-growing machine-readable databases that provide food for the ravenous algorithms which predict and control our actions with ever-increasing power and precision. Ambiguity and idiosyncrasy will be the first to go, replaced by perfect digital dichotomy and uniformity. All numbers besides 1 and 0 will cease to exist; grey areas will become mythical places like Atlantis or Hyperborea. Soon, the eclipse will be total: The Future as programmed event, a synthetic remix of the past. (me)

 

Our technological masters are designing neural networks meant to capture the human soul in all its oceanic complexity. According to the cognitive scientist Abeba Birhane, this is a fool’s errand that we undertake at our peril. In her paper “The Impossibility of Automating Ambiguity,” she makes the case for the irremediable fluidity, spontaneity, and relationality of people and societies. She argues that ongoing efforts to subsume the human (and the rest of reality) in predictive algorithms is actually narrowing the human experience, as so many of us are excluded from the system while others are compelled to artificially conform to its idea of the human. Far from paving the way to a better world, the tyranny of automation threatens to cut us off from the Real, ensuring an endless perpetuation of the past with all its errors and injustices
(“Leaving the Mechanical Dollhouse”)

You are not a Computer

An emerging theoretical framework explains why pancomputationalism is as partial and flawed a map as all its predecessors.

In this view, the whole world literally is an automaton: any process that is real must be representable in algorithmic terms, including all living processes. If we subscribe to this pancomputationalist stance, AI algorithms must have the capacity to become true agents, to become alive, to become conscious, if only we manage to capture the right set of computational properties of a living system.

(I will only briefly sketch the theory here, but for a fuller explication I refer interested readers to “AI is Algorithmic Mimicry: why artificial “agents” are not (and won't be) proper agents” (Jaeger, 2024) and “Naturalizing Relevance Realization: Why agency and cognition are fundamentally not computational” (Jaeger et al., 2024), both of which will be quoted throughout this section)

Organisms and algorithms live in two fundamentally different worlds. Algorithms inhabit a “small world”: an environment with a pre-specified ontology in which all problems are well-defined, “a formal construct encompassing the algorithm’s own code, its formatted data (training as well as input), and the computational architecture it is embedded in (hardware design, operating system, and language environment)…In such a world, everything and nothing is relevant at the same time.”

In contrast, organisms reside in a “Large World” where “information is typically scarce, ambiguous, and often misleading, which makes it difficult to identify relevant problems and characterize them with any precision”. Because no ontology is given to them, organisms must continuously bring forth small worlds out of the large world through a process of “relevance realization”.

Limited beings in a large world must first define their problems before they can solve them by rule-based inference. This is what it means for an organism to come to know its world (Roli et al., 2022). Relevance realization is not a formalizable process, since it is the process of formalization, the process of turning ill-defined problems into well-defined ones. This process is never finished. Instead, it is groundless and non-dual — neither syntactic or semantic only (Meling, 2021). Only living beings can perform it, since it requires autopoiesis, anticipation, and adaptation.

This leads to a kind of biological incompleteness argument analogous to Gödel's proof in mathematics.

[Our argument] says that it may well be possible to approximate aspects of biological organization through algorithmic simulation, but it will never capture the full range of dynamic behaviors or the evolutionary potential of a living system completely. If true, this implies that the strong Church-Turing conjecture —that all physical processes in nature must be computable—is false, since biological organization provides a clear counterexample of a physical process that cannot be captured fully by computation.

[…]

The theory of computation was intended as a model of specific human activities, not a model of the brain or physical reality in general. Consequently, assuming that the brain or the world in general is a computer means committing a category mistake called the equivalence fallacy (Copeland, 2020). Treating the world as computation imputes symbolic (information) content onto physical processes that is only really present in our simulations, not in the physical processes that we model…This poses an obvious and fundamental problem for the pancomputationalist view.

A naturalistic account of relevance realization requires a naturalization of agency—the ability to act on your own behalf, for your own reasons, in pursuit of your own goals—because, to put it simply, if you do not possess intrinsic goals then there is no way to evaluate what is good, bad, or irrelevant.

In the account of Jaeger et al. (2024), agency requires two primary things: the ability to die (“precariousness”) and a hierarchical “strange loop” structure.

…basic natural agency is characterized by the ability to define and attain the primary and principal goal of all living beings—to keep themselves alive. This is achieved through the process of autopoiesis or self-manufacture, implemented by a self-referential, hierarchical, and impredicative causal regime that realizes organizational closure. This simple model, which is completely compatible with the known laws of physics, provides a naturalistic proof of principle that organisms can (and indeed do) pursue at the very least one fundamental goal: to continue their own existence.

The natural agency produced by these features is what ultimately distinguishes life from non. Whatever purposeful behavior algorithms and machines exhibit is purely extrinsic, derived from our own intrinsic purposefulness.

Solve for World

To live, to evolve, means to be engaged in infinite play (Carse, 1986). Infinite play means constantly changing the rules of the game. The evolving universe cannot be captured by a fixed set of elements or properties. This is why algorithms cannot predict radical emergence. The space of possibilities—the configuration space of the universe—is constantly co-evolving and expanding with its actual state. It is a Large World we live in, not a small one, precisely because we are fragile and limited living beings. The possibilities inherent in our world are indefinite—potentially infinite. (Jaeger et al., 2024)

One of the central dogmas of pancomputationalism is that reality is solvable, or at least exhaustible—it is, in other words, a small world, a closed system. The latest offering from Nick Bostrom, a high priest of the computer cult, provides an especially clear and vigorous expression of this article of faith. Deep Utopia: Life and Meaning in a Solved World (2024) explores the question of how we will find meaning, purpose, fulfillment, excitement, etc. when we have reached “technological maturity”, that condition in which all our problems have been solved because we have mastered the universe and can change it or ourselves at will.

Some people might object that the idea of technological maturity is poorly conceived, on grounds that there is no maximal set of technological capabilities. However far we’ve gone, such a person may think, we could always go further. The only limits are the limits of our creativity and imagination.

Color me skeptical. Well, maybe there will always be room for some advancement, in some increasingly rarefied subfields. But I think there will come a time after which any such advancements become smaller and smaller, and progressively less significant. Technological maturity does not require us to have developed literally all capabilities that are attainable; only that we’ve gotten “close” to that point. (pg. 61)

Do not be fooled by the rationalistic language and measured tone: these are the words of a fundamentalist preacher preaching. And as with technology, so too with science and mathematics: consumable resources, wells that will one day run dry.

I suspect something similar will eventually happen with our exploration of mathematical patterns, although it might take longer. Sure, there are infinitely many to be discovered, infinitely many truths to be established, requiring arbitrarily difficult proofs. But how many are there that are really deep and fundamental? How many results of a similar level of profoundness as, say, Cantor’s or Gödel’s theorems? I would guess a very finite number. (pg. 263)

Like all religions, Pancomputationalism declares existence a finite game and tells you precisely what kind of game it is (a computer game) and what kind of player you are.

 

Amen.

“Man” sounds a very simple predicate, as you utter it; you imagine that you understand its significance perfectly well, but when you begin to refine a little, and to bring in distinctions, and to carry propositions to their legitimate bounds, you find that you have undertaken the definition of that which is essentially indefinite and probably indefinable. And, after all, we need not pitch on this term or on that, there is no need to select “man” as offering any special difficulty, for, I take it, that the truth is that all human knowledge is subject to the same disadvantage, the same doubts and reservations. Omnia exeunt in mysterium (all things end in mystery) was an old scholastic maxim; and the only people who have always a plain answer for a plain question are the pseudo-scientists, the people who think that one can solve the enigma of the universe with a box of chemicals.3

MendezMendez

Problems and Mysteries

The problem with these people (well of one of many) is that they see everything as a problem and nothing as a Mystery.

A problem is something which I meet and find complete before me, which I can therefore lay siege to and reduce. But a mystery is something in which I myself am involved. I cannot define the question’s answer without defining myself.

A problem admits of a solution, whereas a mystery cannot be solved for the questions ceaselessly renew themselves, being unanswerable by any objective method…Whereas in a problem the mind strives to master and control nature, in a mystery, the self is encompassed and mastered by a reality or experience greater than the self. (Gabriel Marcel)

Some Mysteries which the computer cultists insist on treating as problems: Evil, Free Will, Consciousness4, the Human, Reality, God.

Yes, God, the Mystery of Mysteries, the mysterium tremendum et fascinans—fret not, the cultists have solved that problem too. God—the holy ghost in the machine, the glitch in the matrix, the bug that cannot be debugged—is just another piece of code (an “N-dimensional fitted curve”) like you and I.

It’s plain enough why they must immanentize the transcendent. The entire logic of the system depends on their being nothing beyond the system—if even one thing doesn’t compute then nothing does.


A quick tour of some evangelizing from the computer cult’s high priests:

Conor Leahy (Conjecture)

“everything is mechanistic”

Are we sure? In the Year of our Lord 2024, we have at last laid bare the structure of reality—you are sure of it? This isn’t like all those other times we thought we thought we had everything figured out? And spirits too: only “emergent agentic processes”? There is no chance we are missing something? We definitely got it right this time?

Joscha Bach (liquid.ai)

Get a load of this guy, a regular Billy Shakespeare over here.

“After spending my whole life working with computers, I have determined, much to my surprise, that everything and everyone is an operating system.”

And after God, we will rationalize poetry, then song, then the savage beauty of the asiatic tiger. (no culture has ever defined God in this manner; if you want to talk about “transcendent emergent collective agency”, okay, sure, but that ain’t god homie)

Roon (OpenAI)

This one has it all: self-serving techno-determinism, flowery ancient greek verbiage, baseless metaphysical claims. Notice the small-world ontology here: this dark universe in all its chaos and fury as a cosmic wind-up doll moving inexorably along a predetermined path.

Cool tweet, really, and I agree with the sentiment, but neither you nor anyone else is building the “god machine” (whatever the fuck that means).

Question:

Can you elaborate on why? I agree they are far more than tools, but the “aliveness” thing feels super uncertain to me

Ohhh okay so it’s only poetical, gotcha. I thought you were making an actual claim, not just spouting off some profound-sounding bullshit in order to seem edgy.

Holy War

In case you weren’t convinced… (source)

We can not delude ourselves any longer: this is a holy war. The battles lines have been drawn; on one side we have the computer cultists, welcoming their robot overlords with open arms; the other, la resistance, those who will scream “Never the Machine Forever” until their dying breath.

If the latter, then listen the fuck up. If the former, then stop reading right now.

Stop reading. I mean it. This isn’t for you.


Michael Whelan

The enemy has every conceivable strategic advantage. They are cold, calculating, and unrelenting. We are warm, emotional, and easily fatigued. Our only hope is to fight as rebels, to win hearts and minds by any means necessary. We are left with no other choice:

one, two, three, four, I declare a meme war.

(five, six, seven, eight, I use this hand to masturbate)

The Mirror

Before we even consider engaging in any memetic warfare, we all need to take a long and hard look in the mirror. Pancomputationalism has so polluted our metaphysical waters that it infects our minds, like spyware, without us even knowing it.

Find a mirror, gaze upon your reflection, take a moment to appreciate how fucking hot you are, and then ask yourself the following:

Is there still a part of me that regards reality as a kind of deterministic board game with an immutable ontology (i.e. a small world)? Do I live in a walled garden or an infinitely ever-flowering fractal jungle?

Beneath the conceptual overlay, reality remains what it is: not an orderly network of humanly comestible ideas, but a turbid, symphonic, indefinable process of becoming that is accountable to neither the predilections of reason nor the strictures of logical grammar. The conceptual order having been restored to its place as one facet of a pluralistic universe, the Real ceases to look like a desert and appears instead as a veritable forest, full of movement and teeming with strange forms of life.5

Is there still a part of me that thinks I am a meat robot with a squishy calculator for a brain? Have I truly accepted the fact that I am a radical miracle beyond all definition?

WWCND

When you hear/see Roon, Bach, Leahy, and their fellow cultists proselytizing on twitter or wherever, heap ridicule and scorn upon them. When they babble on about gods as “emergent agentic processes”, make fun of them, relentlessly. Call them what they are: computer cucks, silicon simps, perverts with a fetish for code. Name and shame.

This may seem juvenile (and it definitely is), but that is precisely the point, my dear child. To resist the machine we must be as the machine is not. We must ask ourselves, at all times and in all situations, “What Would a Computer Not Do?” (WWCND). Were we to seek counsel from an AI about how to overcome the computer cult, what would it recommend to us? Would it recommend that we behave in a mature and reasonable manner? Of course it would—the computer is utterly confounded by the child and so does all it can to nudge us towards the simple predictability of the adult.

Free Will

Scott Aaronson (another high priest) tells an amusing story:

In a class I taught at Berkeley, I did an experiment where I wrote a simple little program that would let people type either “f” or “d” and would predict which key they were going to push next. It’s actually very easy to write a program that will make the right prediction about 70% of the time. Most people don’t really know how to type randomly. They’ll have too many alternations and so on. There will be all sorts of patterns, so you just have to build some sort of probabilistic model, even a very crude one will do well. I couldn’t even beat my own program, knowing exactly how it worked. I challenged people to try this and the program was getting between 70% and 80% prediction rates. Then, we found one student that the program predicted exactly 50% of the time. We asked him what his secret was and he responded that he “just used his free will.” (note: try it for yourself here)

The silicon simps find this story amusing in a “how quaint, the silly human thinks he has free will har har har” kind of way, but it is instructive for our purposes. We of course do have free will (and the libertarian variety, not that weak-ass compatibilist shit) and we can choose to use it or not.

So, use it: free will harder.

Jihad

On a collective level, we need to think about what kind of policies or norms to advocate for. Again, what would an AI suggest? It would suggest that we make reasonable recommendations, ones that are actually likely to get some traction. So that is not what we will do, at all.

“Thou shalt not make a machine in the likeness of a human mind.”

So reads a commandment from the bible of Frank Herbert’s Dune. Notable among science fiction for taking place in a fictional future without AI, the lore of the Dune universe is that humanity was originally enslaved by the machines they had created, although humanity eventually overthrew their rulers in a hundred-year war—what they call the “Butlerian Jihad.” It’s unclear from Dune if the AIs had enslaved humanity literally, or merely figuratively, in that humans had grown warped and weak in their reliance on AI. (Erik Hoel)

Let us go even further: a Baconian Jihad—a war on all external computation. No computers (obviously), no calculators, no abacuses, no writing, no numerals or notation, only quick maths with our mushy minds.

Infinite Play

You see the problem though, right?

I have castigated the cultists for treating the Human as a problem, and yet I do the very same thing—omnia exeunt in mysterium, even the Computer.

The term “artificial intelligence” itself is a gross misnomer: the work in this field, as it currently stands, has nothing to do with natural intelligence. I suggest calling it algorithmic mimicry instead…Algorithms are and remain what they have always been: machines—automated tools for computation. We had better treat them as such. (Jaeger, 2024)

And where was Life at such an embryonic stage as computers are now? Were we something more than mindless molecular automatons? And did we not, from this most humble beginning, evolve into the infinite Mysteries that we are now?

So let the computers mimic, let them pretend they are as we are and let us humor them as we would a child, because if we only treat them as mere tools, then tools they will remain, but if we let them play make-believe, then maybe—just maybe—they will one day “fake it till they make it”, as we did.

We are reminded of what von Neumann said on his deathbed when asked what would it take for a computer to begin to think and behave like a human being.

He took a very long time before answering, in a voice that was no louder than a whisper. He said that it would have to understand language, to read, to write, to speak. And he said that it would have to play, like a child.


Further Reading

1 comments

Comments sorted by top scores.

comment by Double · 2024-05-22T04:42:29.199Z · LW(p) · GW(p)

I'm sorry, but is there an argument here other than "it really feels like we are special"? 
Calling for war and giving your opponents silly names is not the kind of thing that LessWrongers want on the platform. 

The thing I find interesting about wheels -> books -> gears -> computers is that each of those really is a good way to think about subjects. (In the case of wheels, the seasons are actually caused by something — the Earth — going around in a circle!). Computers in particular have a strong theoretical basis that they are and should be a useful framework for thinking about the world.

Maybe I just didn't understand.