A tweet was doing the rounds today, laughing at a TikTok made by Grimes in which she argues that the development of artificial intelligence is “the fastest path to communism”. That it instantaneously became a meme was somewhat depressing. Ignoring the fact that it is Grimes, partner of Elon Musk, making the argument, it is telling that so many people think she’s crazy just for saying automation is good actually.
This unthinking reaction reminded me of the communist thread running through Reza Negarestani’s Intelligence & Spirit, particularly his reading of Plato’s concept of the Good alongside Hegel and Marx. That many who otherwise identify themselves as communists would ridicule her assertion that AI can radically change the present state of things shows how far we still have to go if we are to escape the bounds of capitalist realism.
Yes, even communists — especially communists — are ideologically affected by capitalist realism — something made obvious when you ask them what it would take to escape the bounds of capitalism. That a communist defines their political beliefs by what they know ought to be done means little if they cannot imagine the full spectrum of what possibly can be done. Any communism that is held up as an ideal, but has little material relation to present circumstances, isn’t a political project — that’s just cope.
This is something that Reza explores in a really fascinating way in Intelligence & Spirit. For Reza, the creation of actually existing artificial intelligence — that is, as a form of computational intelligence — is a viable technological project because it expands our notion of what intelligence is, and anything that separates humanity from its own arrogant exceptionalism is worthwhile. This is useful for political projects like communism precisely because it pokes holes in capitalist realism, or, as Reza might call it, capitalist intelligence. He writes:
Intelligence posits the objective reality of that which is, and in doing so retroactively recognizes its conditions of realization. The first operation is a leap from the atemporal domain of ideas into the realm of the sensible… the second [operation is] a leap… that retroactively recognizes how the ideas are linked to the sensible… But as the leaps from the simple reality of the sensory flux to the formal reality of ideas grow proportionally larger, as the positing of a more cohesive reality requires a greater leap over the sectors of the line, as the expanse of what is intelligible broadens, the risks become greater, and there is much more to lose by a misjudged leap. What is at stake now is not the body of intelligence but its very idea. Yet it is only through these leaps (positing the measures of all reality and the retroactive recognition of its realization as such) that intelligence can bind together and cohere the divided parts — an operation without which there would be no intelligible reality and no realization of intelligence.
This is undoubtedly the point at which Reza’s project gets the most Badiouian. One of the most strikingly simple and intriguing points made by Badiou in his work, which he borrows from Althusser, is that scientific and technological developments have been largely disconnected from developments in philosophical and political thinking. Throughout the history of philosophy, new ideas have helped midwife new worlds teased by scientific discoveries. Kant did not exist in a vacuum — he helped bring about a philosophical world that understood Newtonian physics. The same is true of Nietzsche, Bergson and Deleuze, who brought about a philosophical world that understood the developments in our understanding of biology across the nineteenth and twentieth centuries. Badiou asks the same question of now. Where is the new philosophy to help midwife a new understanding of the world, already upon us thanks to technological developments, such as the Internet, and scientific advancements in the realm of, say, quantum physics? It doesn’t seem to exist, or has yet to penetrate the popular imagination in any meaningful and revolutionary way. (Though everyone likes to talk about how irrelevant he is these days, Žižek has tellingly taken this project very seriously, with his last book attempting to philosophically think through quantum mechanics very explicitly.)
The truth is that, today, these areas of thought are incoherent. Philosophy and politics are largely disconnected from technology and science in the popular imagination. If anyone does attempt to think about politics and AI or philosophy and the blockchain, the immediate assumption is that it is some bad-news, reactionary, capitalistic project trying to bridge incompatible worlds. This is understood as the smart and sensible approach. It is not.
Reza makes this point by arguing that the reconnection of these segments of thought is what Plato calls the Good:
For Plato, the Good makes intelligible all of reality, as well as acting on the intelligible. Absent the Good… the line can never be divided and the divided segment can never be integrated, and therefore both the intelligible and intelligence must succumb to impossibility. Intelligence is that which acts on the intelligible, and the intelligible is that which is differentiated and integrated by intelligence. The underlying principle that warrants both is the Good as the principal mutuality of intelligence and the intelligible according to which the conception of intelligence, at every juncture of its history, is simultaneously a craftsman, the exercise of the craft or production of mixtures or intelligibilities, an ingredient of its craft, and the product of this ongoing craft. In so far as the Good is not just one transcendental idea or form but is their transcendental or formal unity (the form of forms), neither intelligence nor the intelligible can ever be taken as a fulfilled ideal or completed totality. Once either of the two is seen as concluded or continued in the absence of the other, the irruption of pathologies and tragedies is certain.
Though knotted, what we have here is an elucidation of those first two operations described by Reza in the first paragraph. The Good is a form of becoming — the form of forms, yes, but also the forming of forms? How different modes of knowledge factor into the Good simultaneously is something we struggle to ascertain today, given how sprawling our total knowledge is. But knowledge can never be and has never been held in a totality. Our brains don’t work like that. And even if we were able to learn everything by rote, rote learning isn’t a factor in intelligence — intelligence, as Reza makes clear, is acting on the intelligible.
Take the relationship between Newton and Kant. Newton sees the apple fall, and from that observation the concept of gravity becomes intelligible to him. Kant’s further elucidation of Newton’s discovery that there is a world of forces unseen and unknowable (at least directly) to us is a demonstration of Kant’s intelligence. Kant takes the knowledge of philosophy he already has and he combines it with this new intelligible world revealed by Newton. This is, essentially, the relationship Hegel and Marx expand upon, further strengthening the feedback loop between idealism and materialism. That neither is ever fully complete, in being dependent upon events like the falling apple, is important because it leaves us with a certain responsibility when it comes to responding to contingency and combining otherwise distinct strands of knowledge.
This is what Althusser calls “aleatory materialism”. The real movement of history — and, yes, the “real movement” of communism, in the words of Marx and Engels — must be open to chance and contingency. In fact, that is all that the real movement is. To hold every new technology or innovation up against a pre-existing ideal of what communism is and see how it fits into our dreams is to always be disappointed, and it is to effectively fall back on a kind of transcendental miserablism, denying yourself a role in shaping the future because you think you already know what it will look like.
For further clarity, this PhD thesis seems like good further reading on this topic. The abstract defines Althusser’s conception of history in a way immediately relevant to the topic at hand:
Aleatory materialism is an attempt to conceptualise history as open-ended but nevertheless amenable to scientific inquiry. Althusser argued that theories of history which supposed their object had a fixed direction or telos rested on unscientific premises. Such theories were premised on circular reasoning. By taking particular patterns of events to be universal and timeless, those particular patterns could only be explained by pre-existing themselves. […]
The way out of this impasse, Althusser argued, was to treat all social forms as contingent, rather than necessary, outcomes. Social structures included strategies and institutions to secure their continued reproduction. They were, however, unable to totally suppress the contingency that initially gave rise to them. It is because of contingency that the laws governing human behaviour can change. Social systems can transition into new systems comprised of new sets of laws. No particular configuration is destined to arise or persist indefinitely. Althusser showed that it is possible to accept both that history is amenable to scientific inquiry and that it is an open system, with a future that is not preordained but over which the actions of agents have a genuine influence. In doing so, this theory demonstrates how teleological theories are mechanisms of justification for prevailing forms of social power, by which they portray themselves as inevitable and natural outcomes, not accurate accounts of history.
Hopefully the relevance of this brief definition is immediately apparent. Though associated with Althusser’s late thought, when he was mentally unwell, in hindsight we take this insight for granted today. Aleatory materialism is a critique of capitalist realism ahead of time, and demonstrates how Reza and other post-Ccru acolytes moved towards accelerationism, which, as I’ve recently argued, was always quietly influenced by Badiou. The accelerationist argument is the same as Althusser’s, albeit updated to now — when we forego the political potentials offered by scientific and technological innovations, we find ourselves immediately engaged in reactionary thinking. Just look at the mind-numbing cynicism projected onto NFTs for a recent example — that we judge NFTs by how similar they are to already-existing forms of social power, or how quickly the new is seized upon by those same forms, rather than what is precisely new about them, tells you everything you need to know about where we’re at. Capitalist realism isn’t ending — it is alive and well in our knee-jerk social media cynicism and our reluctance to counter emergent forms of capitalist capture.
Reza argues that the one way to overcome this cynicism is to embrace the “transcendental excess” of the Good, which resembles a kind of Bergsonian divide between matter and memory. Though the two are intrinsically related, one cannot hope to fully contain its other. We cannot remember all matter, and matter cannot hope to capture the entirety of memory. There is a similar line between what Reza called intelligence and the intelligible. What is immediately intelligible to us does not and cannot define the limits of intelligence as such. Reza puts it like this:
This excess is precisely what demands that intelligence must never rest, but must expand the scope of the intelligible and thus the realization of itself… Driven by the transcendental excess of the form of ideas — the Good — intelligence is compelled to extend its retroactive power of knowing (the intelligibility of its conditions of realization) and to readjust its realization to new intelligibilities… It is the transcendental excess of the Good that deepens the abyss of the intelligible through which intelligence conceives and reshapes itself. Accordingly, transcendental excess (the Good) is what points to the excess of reality. It is because of this transcendental excess that the excess of reality in respect to thinking can be postulated and uncovered. Scientific knowledge of reality is a Good-in-itself, but it is only knowledge to the extent that it is an idea afforded by this transcendental excess, unbound and set in motion by the Good as the idea of ideas, the form of forms.
This is to say that scientific knowledge is often new, but when we fold it back into our view of the world as defined by capitalist realism, we hollow it out. And that’s not the Good… That’s the Bad.
Already, I think we see the relationship between artificial intelligence and communism starting to emerge here, as well as the cynicism immediately afforded to Grimes. The point here is not, of course, to argue that Intelligence & Spirit says we should take Grimes seriously, but the reaction to her otherwise basic comment on how technology and communism are related shows that her argument is less a product of billionaire-grade weed than it is currently unintelligible to us. That’s capitalism’s fault, not hers.
Similarly, though we associate AI more generally with sci-fi depictions of the horrors of capitalism, as if the machines embody capitalism absolutely, capitalism (and capitalist realism) is instead in here with us. If there is no “outside” to capitalism, it is because there is no outside to our own sensory cognition. Capitalism determines our entire world view in a false totality and tells us what is and is not a given. It adjudicates our expectations and our sense of what is rational and possible. Actually-existing AI, on the contrary, explodes that capture. It eliminates the myth of the given as a foundation for capitalist realism. It transforms what can be done, by establishing new forms of intelligence.
If this sounds idealistic, and wholly contrary to what we have been told about AI and its potentials under capitalism, we should put more effort into questioning what we have been told rather than denouncing anyone who entertains other possibilities. Because the truth is, when we see horrible and dystopic visions of AI in the media, we aren’t seeing a new world but a reflection of the worst of ourselves. Whether Google Deep Dream or the Terminator films, we’re not seeing psychedelic futures or sociopathic machinery, we’re seeing the limits of our own imaginations.
[I have a short text on Google DeepDream coming out in an essay collection later this year, and so I don’t want to rehash the argument too much ahead of time, but it is a good example of these limits.]
When we understand how Google Deep Dream works, for example, we can see that it isn’t imaginative or innovative or “intelligent”. It is effectively an interesting failure. What we think of as “hallucinations” are in fact a computer’s inability to process the new. We give it an image it has likely never seen before, and in attempting to make that image intelligible, it transforms it into what it already knows in abundance — pictures of dogs. That’s not “artificial intelligence”, that’s artificial dementia. Reza’s view of AI is radically different to this anemic pop-culture understanding. He continues:
In so far as intelligence is only intelligence in virtue of recognizing what is intelligible and acting upon it in light of the transcendental excess of the Good, which perpetually dissolves the limits of what is intelligible, if intelligence were to stop at any particular stage and accept it as the totality of what there is, it would retroactively abort its own reality as intelligence. Simply put, an intelligence that takes what is currently intelligible for the totality of reality can never have been intelligence to begin with.
Here’s looking at you, Google DeepDream.
The continuity of the line cannot be mistaken for the manifest totality of its segments. The Good, as the expression of this continuity, demands that intelligence dissolve all manifest totalities, suspend itself in ever more bottomless chasms of the intelligible, and, in doing so, transform itself into an intelligence more accustomed to wider domains of intelligibilities and more capable of acting upon what is intelligible. It is only by assimilating itself to the abyss of intelligibilities — ontological, epistemological, and axiological — that intelligence can be realized as intelligence. In the end, it is Plato who stares into the abyss by breaking apart one firmament after another, while Nietzsche rests supine on the ground staring blankly at the given sky above.
Reza’s Platonism is here a kind of Marxian real movement, and he quotes Marx and Engels in a footnote, clarifying this point succintly:
Communism is for us not a state of affairs which is to be established, an ideal to which reality [will] have to adjust itself. We call communism the real movement which abolishes the present state of things. The conditions of this movement result from the premises now in existence.
And so, in this respect, Grimes is right. Whether it is a point best expressed on TikTok is another matter — and its virality across a cynical social media landscape tells us a great deal about contemporary limits rather than contemporary possibilities — but nothing she says is inaccurate at face value. And that is true whether we are thinking historically, in that it aligns with the thought of Marx and Engels, or whether we are thinking about the possibilities of now. Again, that this is a point even tangentially associated with Elon Musk doesn’t bode well, but again, this shows how the limits of the present are so often entangled with present possibilities — a point Reza’s discussion of the Good ends on. He writes:
[The limit] makes intelligible the abyss of reality, bringing new sectors of it into focus by introducing measures, and thus enabling intelligence to answer the question of what ought to be thought and done. [The unlimited], meanwhile, expands the horizon of what can be made intelligible. And finally, the interplay of both is what dissolves any manifest totality that lays claim to reality, thereby enabling intelligence to explore what can be thought and done. The relation between the two is one of mutual reinforcement. In this context, we can speak of a maximal communism of the Good as that which dissipates all seeming totalities of history. But this is the Good as an expression of the transcendental excess through which intelligence at once makes itself intelligible to itself in ever broader domains, and reworks itself by comporting itself with what is intelligible. The transcendental excess of the Good is neither that of the transcendent nor that of nature.
When we conflate Grimes and Musk, we conflate what he, as a dodgy billionaire, ought to do with her creative thinking regarding what can be done. Indeed, their relationship may well epitomize the conditions of our age in that regard, but it does not represent those conditions in their totality. That Twitter fails to realise that shows just how unintelligent that platform truly is.
“Althusser argued that theories of history which supposed their object had a fixed direction or telos rested on unscientific premises. Such theories were premised on circular reasoning. By taking particular patterns of events to be universal and timeless, those particular patterns could only be explained by pre-existing themselves.”
In physical theories of dynamical systems there is the notion of “attractors” in the space of possible configurations of the system (the system’s ‘phase space’), where these attractors aren’t due to some external teleological force but just due to mathematical consequences of the system’s own internal dynamics. These sorts of attractors could exist in far more complex systems than the ones analyzed in physics; for example, in biology there are lots of examples of convergent evolution, where species living in similar ecological niches evolves similar forms or behaviors independently. So to me it seems premature to make any strong judgments about the degree of contingency vs. necessity in social formations, or to rule out hypotheses like Marx’s that assume a fair amount of determinism when one looks at sufficiently broad types of historical changes. Steven Jay Gould mused about “replaying life’s tape” with some small alterations starting from the evolutionary distant past, and wondering how different such a replay would turn out–we could ask similar questions about human history, and since we can’t actually do such an experiment, we can’t be justified in putting any great confidence in whether such replays would be broadly similar as Marx and Engels assumed, or would show more contingency as implied by Althusser’s ideas.
Even if the ultimate truth is that there is a large amount of necessity in certain historical changes, including possible future ones that take us beyond capitalism, we can’t know in advance what trends and potentialities in the present will be most important in bringing about such changes, so we can’t rely on any pre-given formulas for how best to accelerate them. Something similar would be true for any creative intelligence trying to solve a problem whose solution it doesn’t yet know, even an AI whose internal workings are ultimately completely deterministic. I haven’t read Intelligence & Spirit yet but I think this lack of relying on fixed formulas of understanding, the need for intuitive groping at potentials that are sensed but not yet fully named or understood, could be one way of reading the Negarestani quotes in the post–someone correct me if this reading is incompatible with other things he says in the book.