AI art

On the mortal soul

/ Simon / 1 Comment

When I published my recent piece Toward the Butlerian Jihad one of the concepts I brought in was the mortal soul. This was largely in service, as others have noted, of a secularization of the concept of the Butlerian Jihad – a holy war against “thinking machines” that occupies the position of a considerable historical event in the background of the science fiction novel Dune and its sequels.

However, as has been rightly pointed out by others, playing around with the idea of a soul which could be disfigured raises the risk of reintroducing natural law into our metaphysics. This is, of course, something we should avoid. I had been thinking about expanding on the concept of the mortal soul regardless as a part of my overall project on materialism and magic however, in light of this well-received response, I thought it’d be a good idea to get this explanation out a bit faster than I otherwise might have.

A brief genealogy might be a good place to start. The idea of the mortal soul is something most directly encountered in Nietzsche’s Beyond Good and Evil where he says, “Let it be permitted to designate by this expression the belief which regards the soul as something indestructible, eternal, indivisible, as a monad, as an atomon: this belief ought to be expelled from science! Between ourselves, it is not at all necessary to get rid of “the soul” thereby, and thus renounce one of the oldest and most venerated hypotheses—as happens frequently to the clumsiness of naturalists, who can hardly touch on the soul without immediately losing it. But the way is open for new acceptations and refinements of the soul-hypothesis; and such conceptions as ‘mortal soul,’ and ‘soul of subjective multiplicity,’ and ‘soul as social structure of the instincts and passions,’ want henceforth to have legitimate rights in science.”

Nietzsche attempts to walk a razor’s edge here between a materialist account of the world which abandons the concept of the soul and what he describes as the Christian atomism of the soul which he treats as the last vestiges of a belief in a rigid substance.

Nietzsch’s soul of subjective multiplicity is, instead, a process of transformation that occurs within a person. We cannot treat a mortal soul as a substantial object: a ghost within a shell. Instead it represents the ever-transforming flow of subjectivities, affects and material effects created by a person. However this psychological frame does not discount a theological reading as we can find reflections of this in the far-older concept of Anattā.

This is a thorny issue within Buddhism so I will provide my interpretation which is largely drawn from the Chán school however I do anticipate lively disagreement here. At an etymological level, Anattā means no-self. This is a challenging concept because Buddhism upholds reincarnation as a part of its metaphysical universe and if there is, in fact, no self then what is there to reincarnate?

There are two possible solutions here. The first is to suggest that there is no unchanging self while there may be a continuous stream of being, a flow of moral development and consequence, we cannot point to it as an eternal and unchanging self. After all, how could a person develop toward Nirvāna if change was impossible? As such we see a soul that exists but is, much as Nietzsche would later propose, a “soul of subjective multiplicity.” From this sense it doesn’t matter much if the soul passing between bodies is substantial because it will be caught in a constant transformation brought about by the accumulated weight of its past lives and the condition of its latest birth.

Another solution would be to treat a soul much as a candle flame used to light a second candle. In this case the original flame will eventually burn out but the new flame will remain. The origin of this new flame carries from the extinguished flame before but it is not of a substance at all. Rather it is a spark, a blend of flow and event, which ignites a novel being. The treatment of a soul as a flame is valuable here for creating an account of soul as process. A flame is never still, never static, it consumes fuel it produces waste. Early metaphysicians such as the stoics also associated fire with an elementary vital principle throughout the universe. In this sense fire stands as a form of life and even now that we understand these phenomena to be separate from each other fire remans a valuable metaphor for describing the process of a life as process, as flow and event.

And so what is a mortal soul but the account of the changes brought about through a life. As such the soul extends past the body of any given subject and into the socius that forms around them. A subject is a process of transformation. I am, at 44, not the same person I was at 22 or at 11. And tomorrow I will be somebody different still. The very act of putting pen to paper on this essay transforms me in that it will change, subtly or suddenly, how others see me. This act dissolves the body of the subject into the field of being because it is equally true that subjective changes within me – the idea of what I, as a subject, am is constantly reassessed. Being is contingent and there is no essential character to a being.

This is ultimately my interpretation of Anattā: a being that exists as process and absent substance, absent essence. This, then, gets to my later criticism regarding AI and death. These necromantic objects operate from the assumption of an essence. In order for a podcast with Plato to have any meaning whatsoever there must be an essential Plato who can be conjured back out of his texts.

That the idea of Plato, the soul of the man, is entirely different now as he has become the commentaries of philosophy and counter-philosophy passing through Aristotle and Plotinus, a worm through time all the way to Kant, Hegel and all the rest, makes the idea of a podcast that returns him to a single essential figure who could be interrogated or who could interrogate in some meaningful way absurd.

It is disfiguring of the soul because it wants to fix this process of transformation back into a substance. In the Jean Leflambeur trilogy by Hannu Rajaniemi we see such a mission taken to its absolute extreme as the Sobornost seek to do away with death itself. Their mission, to roll back time and do what Benjamin’s angel of history could not, restoring all the dead souls lost to history is a threat to others in this book’s universe precisely because of the terrifying impact such a deed would have on the ability of (post)humanity to continue to change.

Growth and change depends on the elimination of essence. Once we allow essence into our metaphysics we are trapped by the idea of Platonic remembrance and everything becomes nothing but an emanation of an essential elsewhere.

In the Theses on the Philosophy of History Walther Benjamin describes Historical Materialism as a Mechanical Turk (not unlike these “AI” tools) which must be animated by a hidden theology to become puissant. Atheist Marxists often interpret this as a critique of the failures of Marxism, like Marxism in the 1930s was insufficiently anti-theological, but this depends, to a certain extent of continuing the mistaken reading of “opiate of the masses” to mean “drug, bad, avoid” rather than historicizing it as meaning, “something to ease pain.” If we read the first thesis in a straightforward way we can instead suggest that a theology (I hesitate to say secular theology here as Benjamin was not a person of secular spirit) is needed. Regardless Benjamin’s theological interpretation of Marxism serves to target the very idea of history as a process of progress. Instead history is the wind which blows Angelus Novus into the future as the debris and dead of past eras heap up at his feet. AI technologies then attempt to do what the angel of history cannot and return these dead to us in some essential form.

I know it is a frightening concept to deny that the unknown future will be redemptive and then to insist we must fight to go there anyway. This is why I briefly invoked Kierkegaard at the end of my piece on the Butlerian Jihad because embracing the danger of an irredemptive and unknown future requires a leap past extreme anxiety. We do not leap toward God for his throne is, by now, thoroughly vacated but this increases the urgency by which we must strike down those people who would raise up a mechanical god to redeem the dead of history.

“It was but a step to the illusion that the factory work which was supposed to tend toward technological progress constituted a political achievement,” Benjamin says and I see a similar critique in those who say that “AI” is necessarily a tool that could be meaningfully wielded by a leftist project. When I say that “AI” must be stolen from the Bourgeoisie what I mean to say is that it is insufficient that Proletarian hands wield this technology for Proletarian aims. This is falling for the same progressivist view of history Benjamin rightly criticizes. Rather I am saying that it is a technology that must be denied from the Bourgeoisie. We don’t take it like Prometheus taking fire from the gods but rather to deny other hands the use of it. I see this as a moral imperative because the resurrection of an essential and immortal soul clogs the path to an open and liberatory future. Effectively the leftist project we can trace through Spinoza and Marx to Beauvoir and others depends on us disregarding the rubble of the past. We cannot redeem the dead. There is no past to return to. If we are to be free we must be mortal: we must be subject to absolute contingency and transformation. Meillassoux describes contingency as being necessary if we are to “get out of ourselves, to grasp the in-itself, to know what is whether we are or not,” and, again, this circles back to the anti-facial consciousness raising of Fisher and Foucault. I raise up this spontaneously insurrectionary desire against specters of the social democrats of the Second Internationale as they were the self-same people Benjamin critiqued for mistaking change for progress. Certainly AI tools represent a change; it does not follow they represent a progression.

Unlike Marx I do call for a revolution with a specific moral character – one which I think is clear from the citations of Deleuze and Guattari, Beauvoir, Benjamin and Kierkegaard. This moral universe is one that is necessarily toward spontaneous liberation, the potential of which is as evident as the spontaneous enlightenment of Chán Buddhism. As such its character is necessarily a mass character but one that will not allow for the possibility of redemption. It is, however, still a (secular) theological proposition. We must overcome an overwhelming anxiety that we will not bring about a future that is free and act a if that liberation were assured. In this regard, by putting the debris of history before us, “AI” is an obstacle at best. While contingency allows that a tactical use of “AI” might be valuable in this or that moment we must recognize that any such tactic will be counter to our own ethics; only holding up contingency as the supreme absolute opens the door at all to wielding such a tool.

I do want to temper this statement a bit to suggest I am not making of “AI” a Ruling Ring. This would depend on an absolute and essential understanding of evil which would go against all the contingent and transformative metaphysics I champion. But we should recognize that these uses, even if effective, are not moral. This is not because it violates some natural law. If we follow Meillassoux on contingency then we must vacate every absolute and this includes the absoluteness of laws as fundamental as the Planck Length (as Rajaniemi speculated). If we cannot even say with certainty that:

represents a limit in all places and in all times then how could we possibly say with certainty that there is any sort of social absolute? We must vacate any natural law and treat law with as much contempt as Benjamin did in the Critique of Violence. But we should also accept that even contingency is contingent and that this may lead to the creation of (contingent) fields of consistency. In such a case we can say, barring some transformation heretofore unseen, it’s right for us to do away with these tools as serving only our enemy.

Ghost of Ned Ludd in the Shell

/ Simon / 1 Comment
“Ned Ludd Smashes a Loom” via an AI Art platform.

With the total collapse of the NFT market the financiers whose grift involves the full financialization of art has had to look to different tactics. Happily they have found just such a rhetorical tool in the emerging field of, “AI Art.”

AI Art, much like NFTs, has been around for a while but has had a recent influx of attention and cash from the tech sector. Google Deep Dream was likely the first exposure people had to this medium and it has been around since 2015. However recent iterations of the software have become more controllable than Deep Dream. The training sets have “improved” as long as one’s yard stick for improvement excludes exploitation. The result is that it’s easier to get aesthetically unified results from a prompt than it had been previously where you’d mostly just get animal chimera jammed into input images like distortion patterns.

There is currently a debate ongoing regarding AI art which asks a few questions:

  1. Is AI art actually art at all?
  2. Is AI art theft?
  3. Should AI art be resisted.

I will principally be discussing the third point here but I do want to address the first and second points to say the proponents of AI art are mostly correct in that what I’ve previously called Will Toward Art can be found in the cycles of prompt and iteration undertaken by an AI Artist. The automation and mediation by machinery present in AI art is just as present in photography. One is shot framing and selection from a field of material objects. The other is shot framing and selection from an iteration of an algorithm. As such it would be disingenuous to say that AI Art is not art.

Now that doesn’t mean it’s any good and the majority of AI art is at best, by the very nature of its iterative selection process, parodic and derivative. The algorithmic basis of AI art is to take a catalog of extant works related to the prompt keywords and to shuffle through them seeking out similarities in order to output a result. You cannot but create a parody of extant works when you are using such a basis for creation.

But parodic art is still art and insofar as difference can arise out of the affective change brought about by repetition this art can, in theory, lead to the arising of the new via that process.

This then brings about the question of whether AI art is theft and I don’t think it’s possible to say anything other than that it is. As AI art is entirely predicated upon the iterative sampling of extant images it is, fundamentally, a theft. But then I’ve been clear in the past that such iterative cycles are a part of art and that this criminality is inseparable from the artistic process. What’s the issue here is that AI art automates this theft.

A counter-example of art being theft in a non-automated manner would be to look at the upcoming Zach Snyder film Rebel Moon. Snyder’s project started off as a Star Wars film but, when that fell through, he went on trucking, iterating upon the ground Star Wars laid. I suspect the parodic character of the final product will be effectively self-evident. Certainly everything I’ve seen about it anticipates this likelihood.

However, in order to do this act of replication, Snyder had to produce a whole $83 million film project employing a few hundred people, including many, many artists, each of whom will be bringing their own ideas and influences into the fold. An AI art program does this with the literal push of the button.

We can make similar statements regarding iteration and the use of samples in music. While music that samples other songs clearly is taking from that art it requires labour to do so. This then is the crux of the problem with the automation of AI Art: the complicated and organic process of iteration has been handed over to a machine that automates it, making it far easier for artists and non-artists alike to produce a result that is, at the very least, reminiscent of artwork.

And that raises the third question: Should this be resisted?

Now I have seen some proponents of AI Art conjuring the specter of the Luddites to argue against resisting the arising of AI art. However most of them couch this within the idea that automation was inevitable and Luddites were fools to resist. “AI art is coming for your job regardless so you better be prepared.” And of course this is nonsense.

Let’s start by looking at one of the most rigorous nearly-contemporary accounts of the Luddites.
“Factory legislation, that first conscious and methodical reaction of society against the spontaneously developed form of the process of production, is, as we have seen, just as much the necessary product of modern industry as cotton yarn, self-actors, and the electric telegraph. Before passing to the consideration of the extension of that legislation in England, we shall shortly notice certain clauses contained in the Factory Acts, and not relating to the hours of work. Apart from their wording, which makes it easy for the capitalist to evade them, the sanitary clauses are extremely meagre, and, in fact, limited to provisions for whitewashing the walls, for insuring cleanliness in some other matters, for ventilation, and for protection against dangerous machinery. In the third book we shall return again to the fanatical opposition of the masters to those clauses which imposed upon them a slight expenditure on appliances for protecting the limbs of their workpeople, an opposition that throws a fresh and glaring light on the Free-trade dogma, according to which, in a society with conflicting interests, each individual necessarily furthers the common weal by seeking nothing but his own personal advantage! One example is enough. The reader knows that during the last 20 years, the flax industry has very much extended, and that, with that extension, the number of scutching mills in Ireland has increased. In 1864 there were in that country 1,800 of these mills. Regularly in autumn and winter women and “young persons,” the wives, sons, and daughters of the neighbouring small farmers, a class of people totally unaccustomed to machinery, are taken from field labour to feed the rollers of the scutching mills with flax. The accidents, both as regards number and kind, are wholly unexampled in the history of machinery. In one scutching mill, at Kildinan, near Cork, there occurred between 1852 and 1856, six fatal accidents and sixty mutilations; every one of which might have been prevented by the simplest appliances, at the cost of a few shillings.” Marx says at the start of Capital, Vol. 1, Ch. 15, Part 9 – framing the conflict between milling machinery and workers like the Luddites not in the abstract realm of the dangers of automation but rather in the physical toll these factories put to workers and, this being important, the power relations that allowed for that toll. Marx is clear that it is, in fact, the vague wording of laws and the penurious behaviour of factory owners that led to factory casualties rather than the intrinsic character of the factory.

Marx pivots to discussing technological change more directly, saying, “The only thing, that here and there causes a change, besides new raw material supplied by commerce, is the gradual alteration of the instruments of labour. But their form, too, once definitely settled by experience, petrifies, as is proved by their being in many cases handed down in the same form by one generation to another during thousands of years. A characteristic feature is, that, even down into the eighteenth century, the different trades were called “mysteries” (mystères); into their secrets none but those duly initiated could penetrate. modern industry rent the veil that concealed from men their own social process of production, and that turned the various, spontaneously divided branches of production into so many riddles, not only to outsiders, but even to the initiated. The principle which it pursued, of resolving each process into its constituent movements, without any regard to their possible execution by the hand of man, created the new modern science of technology.”

And of course it’s immediately evident to see the process by which automation is now doing to the mysteries of the arts what Marx was demonstrating in his discussion of potters and weavers. As such we have to re-situate the Luddite movement, even based on the strength of these establishing statements alone, as not one of a class against machines but rather as a battlefield of antagonisms between two classes: the craftsmen who were undergoing a process of proletarianization and the owners of machines who wished to suck their blood. As Marx says, “We have seen how this absolute contradiction between the technical necessities of modern industry, and the social character inherent in its capitalistic form, dispels all fixity and security in the situation of the labourer; how it constantly threatens, by taking away the instruments of labour, to snatch from his hands his means of subsistence, and, by suppressing his detail-function, to make him superfluous.” This is precisely the ‘inevitable’ future, brought about solely by technology, that these advocates of AI demand artists content themselves with. Marx’s final word on the Luddites comes down to this, “It took both time and experience before the workpeople learnt to distinguish between machinery and its employment by capital, and to direct their attacks, not against the material instruments of production, but against the mode in which they are used.” And it’s necessary, when deciding if AI art is to be resisted, to ask this same question: is the problem the machine or the hand that controls it?

Certainly this automated art stealing from and iterating upon a vast catalog of images posted online, has the capability to supplant illustrators, advertisers and other such artists. But this supplanting is not a matter of the tool but rather the mode in which it is used.

And this, then, is where we must begin asking for whom these tools have been made and to what ends. There is a tendency, within capitalism, to attempt to mystify the machinery of it. If the problem is that the eternal system of capitalism creates externalities it’s easy enough to shrug it away. It wasn’t on purpose that the machine crushed illustrators; it was merely their time to be automated into obsolescence.

But, of course, this assumes far too much. Who owns this machine is a far more pressing question and, in the case of OpenAI whose Dall-E tool is one of the most popular, the ownership question points back to Elon Musk and Sam Altman. Musk eventually departed leaving the “capped profit” limited partnership, registered in the tax haven state of Delaware (natch) under the control of Altman and Greg Brockman. This is not a tool owned by artists nor for artists. It’s a commercial asset of the financial class. And this, then, demystifies the nature of the struggle. Altman, Brockman and the rest of the tech-startup-venture-capital crowd would prefer that they be paid for illustration instead of little artists. Craftsmen find their work copied by a black-box machine and their jobs supplanted by an AI that can produce ugly illustrations on demand for the low-low price of $15 for 115 prompts. So much more efficient than hiring a craftsperson.

So, yes, AI art should be resisted. It shouldn’t be resisted because it copies images and iterates on them but rather because its application is yet another attempt of tiresome tech bros, the self-same ones who tried to sell the world on NFTs, to suck the blood of working artists. Smash the fucking things to the ground.