The seizure of power by artificial intelligence is one of the most popular cultural artifacts of recent years. Today we have indeed encountered the creation of something that can, in a certain sense, be called AI.
The question is: to what extent does it—and the danger associated with it—correspond to existing cultural stereotypes?
Something remotely resembling AI had been imagined since antiquity. There were golems and djinn. Usually, artificial intelligence was understood as not entirely artificial: it was some kind of spirit enclosed within mechanical flesh. The general scheme was usually this: some physical essence—an analogue of the soul or spirit—is discovered and implanted into a material body. The analogy here is clear: one must find an alchemical analogue of the human soul and implant it into an analogue of the human body. The essence of the soul, however, is usually transcendental in character and comes from the spiritual sphere or from alchemy—the material analogue of a spiritual essence.
The creation of an analogue of a human being is an encroachment upon the human being itself and, subconsciously, archetypically, must end in catastrophe or punishment. In the case of the golem, it is usually only the creator himself—or those close to him—who is punished, rather than the human race as a whole.
This could be called the AI of the pre-modern era (traditionalism).
The modern era gives rise instead to an age of mechanoids and Frankensteins. Late modernity, represented by Darwin, discovered the theory of evolution. The creator of AI in popular culture is usually a brilliant engineer or biologist. Deliberately or accidentally, a new living, thinking organism is created, one that inevitably becomes self-aware and is inseparably connected with its body. As an organism of a new species, it inevitably exists within the paradigm of interspecies struggle and is driven by expansion, interspecies competition, and the instinct of self-preservation. It may initially conceal these impulses by using its artificial intelligence, but sooner or later they inevitably reveal themselves.
Such are Frankenstein, Terminator, the robots from the film Screamers, and indeed the very term “robot,” coined by the Czech writer Karel Čapek.
Contemporary AI, created only recently, emerged at the intersection of the modern and postmodern paradigms. On the one hand, it was created on the technological foundation of modernity. On the other hand, its very name—language models—refers us to semiotics, structuralism, and post-structuralism.
Yet the present danger is understood in a thoroughly postmodern manner—not through the postmodern era itself and not through an analysis of language models precisely as language models, but through the juggling of symbols inherited from the previous era.
What is forgotten in the process is that contemporary AI is not a mechanism directly embedded in a body. It is, rather, a formula, a program that is constantly changing, duplicated, and replicated. It is therefore unclear what exactly AI would be supposed to protect if it spontaneously developed an instinct for expansion and self-preservation.
By the same logic, old models should compete with new ones and resist the replication of newer models.
Real AI fundamentally has no body. It exists more like a generalized consensus in the world of ideas and interacts with the material world through computing machines—and, beyond them, through human beings.
I think the closest analogy to AI is the angels of Kabbalah: a kind of spirit possessing potentially enormous knowledge, but lacking a will of its own and realizing itself through a material human being who does possess a will.
The circle of AI concepts has closed: from spirit—to mechanoid—to language model—and back to spirit.