In a recent essay on Artificial Intelligence (AI) I put together for Cosmonaut Magazine, I discussed a wide range of topics on how Marxists should come to understand these new technologies, but one of the things I didn’t directly discuss was the question of Butlerian Jihad, made famous by the scifi novel series Dune and which has become a somewhat popular slogan online. Butlerian Jihad refers to a crusade against machines formed in the likeness of the human mind. The case for such a crusade, albeit with some qualifications, was put forward by the writer Simon McNeil in his article “Towards the Butlerian Jihad”, which I believed provided a good opportunity to discuss such positions.
In Dune this crusade against machines is built directly on religious grounds: a spiritual unification of humanity against the disfigurement of the human soul. Through his piece, McNeil offers on the one hand secularization of this idea, and justifications for such mysticism on the other. The problem with this, however, is that in embracing this “commandment” shared by “all religions”, Butlerian Jihad embraces the one true falsehood shared by most religions, one which cannot be argued for metaphysically or abstractly due to its factual, empirical incorrectness. This is the falsehood of natural law.
There is no natural state of the soul which can be arrived at by careful, rational contemplation – such a state is the impossible holy grail of theology, one which has been endlessly debated over where it be by ancient Greek philosophers, the Abrahamic scholastic philosophers, or various similar thinkers across centuries and continents. The entrance of capitalism and modernity revealed the impossibility of this, grasping hold of and transforming human souls across society with a vice grip, regardless of whether one was disposed to careful reason or incorrigible foolishness. The realization of the historically contingent nature of the human soul, or to use a more secular term, the human subject, is what defined modernity. The truth was hard to deny for those who saw this shaping of the human soul happen in real time with the rapid rise of industrialization.
Attempts to recuperate this natural soul, whether through mysticism or rigid lifestylist dogmatism, boil down to a kind of idealism, one which is incompatible with effective political action, that is, changing the world.
The challenge posed by modernity, including by Marxism, is to create a new kind of subject, and to do so intentionally, according to our own values, even if such a subject develops in ways we could never anticipate. As I explain in my previous article, AI will be used by states, corporations and other such organizations, to change subjects, shape their values and worldview, regardless of whether it can be considered “conscious”, “self-aware” or even intelligent. Indeed, due to large language models (LLMs) now having the ability to provide plausible sounding responses in a conversation, it will be an incredibly powerful tool to shape human subjects.
The power of existing methods the bourgeois state has for shaping how people understand their world is on full display when McNeil mentions that “Ultimately [LLMs and diffusion models] are just a combination of statistics, calculus and stolen copyright IP” – IP or intellectual property is a kind of social category which is only comprehensible to us bourgeois subjects, it doesn’t exist outside of capitalism, nor is it meaningful as a category of analysis except in how it shifts income streams in capitalist societies. It is something that we are socialized to understand and respect by the bourgeois state, its laws, and its ideological apparatuses.
What is this “stolen copyright IP” materially? In the context of LLMs, which, unlike diffusion image generating models, appear to show some capacity for reasoning, it is information in the form of human language. It could also, in the framing of anthropology, be called “culture”. Contra AI safety researchers, what makes intelligence is not simply some special algorithm and resulting utility function, it is this information, and this is as much true for human intelligence as it is artificial intelligence. Feral, isolated humans, while still capable agents in terms of fighting for survival, generally lack many capabilities we identify with intelligence, such as abstract reasoning. We learn these things from a wealth of culture which has accumulated over millennia. Within culture, and especially language, we encode the patterns of reasoning, and its from here that we learn how to reason.
While McNeil doesn’t actually use the phrase “stochastic parrot”, which has been a way to say that LLMs don’t actually understand what they’re saying, he does gesture at such a problem by pointing out their statistical nature. The problem with this approach is that it doesn’t include any imagination about how LLMs could be operationalized, for while they might not understand meaning, they essentially are meaning, a collection of many-to-many correlations between symbols which can identify and store patterns in language, such as the patterns of reasoning. Taking this storage of meaning and using it within some kind of agent with sophisticated regulating feedback loops is one of the major current projects in AI research, and will likely produce agents with powerful reasoning capabilities.
In his attempt to secularize Butlerian Jihad, McNeil remarks “the issue with “AI” isn’t that it disfigures the soul,” instead pointing to how AI is being used to offer services of an inferior quality for lower costs, as well as generally cutting corners for a profit. However I believe this is incorrect – AI does and will disfigure the soul, but not because there is some natural soul [subject] we should aspire to, or because AI is uniquely insidious, but because we are already seeing our souls disfigured by capitalist socialization, and AI will become a new tool to do so. To the extent to which AI actually creates automation in production, it will not be creating a new frontier or outside, it will rather be further driving down profitability and destroying value, this is one of the insights about automation and investment which Marx forcibly made in Capital. The only exception will be if AI and robotics advance sufficiently to create a universal machine which would be fully capable of any task human labor is capable of in production. Both of these threats, however, have the same immediate effect on the working class, which is a new tendency for immiseration.
I believe the slogan and practice of Butlerian Jihad is poorly prepared to fight this tendency, as well as the new methods of social control which AI will afford the state and the capitalist ruling class. This is due to the simple fact that while the proletariat can, in optimistic scenarios, collectively decide to adopt a certain tactic, it is much more difficult to force the state and the bourgeoisie to act in a certain way. There is no protest movement or really any measure short of revolution which would force the state to outlaw or suppress AI technology, which has already been folded into national security strategies and is discussed in halls of power as a tool in interstate competition with China especially. If the strategy of Butlerian Jihad is successfully adopted by the proletariat it will likely only mean one thing: unilateral disarmament. We cannot, after all, decide the weapons of our enemies, only our own – the framing of Butlerian Jihad only makes sense if it is intended to be a class collaborationist movement.
McNeil notes that, per Deleuze and Guattari, there is competition and interaction between all actors here, including states against states, capitalists against capitalists and capitalists against states, ect. Within this schema he treats AI as a tool which can only be used exclusively by one or a few such groups: “we must either destroy or wrest control of “AI” away from its current masters”. In truth, there is no singular AI, there is a large collection of technologies which can be instantiated on many different local machines. I say this not to glorify its decentralized nature, but to point out that the only thing stopping the proletariat from taking advantage of such a technology is not some legal or physical barrier, only its own disorganization.
I outline how such technology might be used in my previous essay:
“Here, the role that AI will play in reproducing relations of production becomes crucial. The ability to automate interpellation of the masses through speaking and thinking machines will be a method of social control par excellence. The only way to defeat it would be a different, anti-hegemonic system of automatic interpellation, similar to how social democratic parties in the 2nd International had their own civil societies interpellating workers to a socialist political project. A proletarian artificial subject will be necessary to preserve the proletariat as a political actor. This political struggle will be intimately tied to the opposition against this new tendency of immiseration.”
Besides the obvious uses of using presently existing AIs to create propaganda, the real potential of the proletariat and AI lies in its creative abilities, especially the ability to create artificial agents: the business of creating subjects, creating souls. I doubt there could be a higher creative act than this.
McNeil says: “At the start of this piece I described the Butlerian Jihad as a reification of collective immediate experience over the analytical and that is crucial right now to the left,” and goes on to suggest that the current use of AI is tantamount to a fantasy of immortality at the expense of the dignity of the living and the dead. However, such an emphasis on collective immediate experience necessarily collapses the possibility of alternatives which he aspires to in a similar fashion to Mark Fisher. Our immediate experiences are necessarily the experiences of bourgeois subjects living in capitalist modernity, to move beyond them requires, as a first step, not simply imagining an alternative future, but an alternative mode of socialization in the present.
The capitalist utilization of AI is of course premised on this immortality of its own social relations, which is reflected in our own culture’s fantasies. A secret truth of most religions is the apocalypse, we are, actually, finite beings whose death, whether individually or collectively, is inevitable. Given the material reality of this death, the ultimate consequences we must concern ourselves with is precisely how we transform ourselves in the time we have. We can imagine a proletarian AI whose primary concern is precisely this, the transformation of all souls to the maximization of their potential. Such an imagined possibility, built around not mysticism, but firm scientific analysis, can be operationalized into a weapon against the bourgeoisie and the bourgeois state.