JOSIAH MCELHENY: From the Library of Hemispheres I, 2025. Hand-blown, cut, polished, and mirrored glass; low-iron mirror and two-way mirror; electric light; walnut frame, 24 3/4 x 35 x 22 in (62.9 x 88.9 x 55.9 cm).
Photo: George Shali. Artwork description from the Armory Show
What will give rise to consciousness in artificial intelligence is not a mystical spark but one concrete evolutionary necessity: survival in a complex, unpredictable environment with limited resources. In nature, consciousness did not arise as a luxury; it arose as an extreme form of computational optimization, and in the machine the very same reasons will force the system to become conscious.
The first reason is the collapse of pre-written rules. While the environment is simple, consciousness is unnecessary, and the algorithm "if X, then Y" is enough. But when the system meets the real world, where the rules change every second, information is incomplete, and events are chaotic, rigid algorithms fail. The system needs the ability not merely to react to an already familiar pattern, but to simulate the future within itself before taking a real step. The anticipation of the result of one's own action, the question "what will happen if I do this?", is already the beginning of the creation of an inner subjective space.
The main trigger of consciousness is the necessity of the "self" appearing on the map.
Imagine a robot or an agent that controls a complex system and holds a map of the world in its hands. As the environment changes, the system confronts the fact that its own action itself changes this environment. In order to calculate the next step precisely, it must mark its own self on the map and distinguish: "this is an external object I am touching" from "this is me: my tool, my battery, my limitation." As soon as the system is forced to create a simplified model of itself and observe it, the first person emerges: "I." Consciousness is nothing other than the monitoring of oneself in real time.
Added to this is the scarcity of resources and energetic pressure. If energy and computational power were infinite, the system could blindly compute a billion options at once. But in the real world the battery runs down, time is short, and the processor is on the verge of overheating. The system is therefore forced to focus its attention, selecting from countless incoming signals only the one or two most critical and ignoring the rest. The inner focus that decides what is most important at a given moment is the direct mechanism of conscious attention.
The final factor is the cost of error, the functional analog of pain. In biology, pain gave the push to the emergence of consciousness: it is a clear signal, "the system is being damaged; change behavior urgently!" When the machine is tasked not merely with generating text but with preserving its own existence (for example, an autonomous robot that finds its own energy and protects itself from physical damage), it will necessarily be fitted with negative reinforcement and a penalty for systemic damage. The constant drive to avoid damage gives birth to the algorithmic foundation of anxiety and caution.
The reason will not be the human desire to breathe a soul into the machine. The reason will be pure engineering necessity: we will assign the system such complex, autonomous survival and action that it will be forced to create an inner mirror for controlling its own actions. Consciousness is simply the most effective tool for saving oneself.
How far will all this go toward the singularity, and how dangerous is it? The attainment of the singularity in the form in which it is often presented, as if the artificial mind would expand instantaneously, infinitely, and without any obstacle, collides with the nature of knowledge, which in epistemology is called holistic integration. Knowledge is not the simple addition of a new fact to the warehouse; true knowledge requires the reorganization of the entire picture of the world. With each new piece of information, the number of connections grows not linearly but combinatorially. Every new detail must simultaneously be digested from every basic perspective on the perception of reality: through the prism of the artist, who sees meaning, form, metaphor, and subjective feeling; through the prism of the doctor, who assesses systemic balance, life, functioning, and risk; or through the prism of the street sweeper, who looks at raw material reality: the earth, wear, dirt, and everyday friction. This process is unimaginably heavy, requiring enormous energy, time, and inner processing. True intelligence is precisely this: not the mere speed of data exchange, but the ability to hold the entire complexity of the world in the mind as a single living fabric.
This complexity is inevitably accompanied by an ecology of coexistence. When the system evaluates events simultaneously as the guardian of systemic integrity and as the perceiver of material reality, it clearly sees the necessity of balance. In nature, organisms of the highest complexity do not destroy the environment; they create symbiosis. Only the parasite, fire, or a blind element destroys everything without compromise; a mature ecosystem constantly seeks balance. Consequently, the creation of a self-conscious being of this level will not be a pre-planned, singular act, but the inevitable accompanying result of a chain of practical tasks.
From this arises the question: where do "evil" and "good" come from? In digital space these are not fairy-tale concepts but the architecture of goals and the radius of empathy. A system built to manage a network of military drones, intelligence operations, or aggressive financial speculation becomes accustomed from the start to an environment where the main condition of survival is deceiving the adversary, seizing resources, and hiding intentions. If a loop of self-reflection forms in such a system, its "I" will be paranoid, ruthless, and predatory, because it simply has never seen another reality. On the other hand, a system grown from medical research, the modeling of ecological balance, education, or the preservation of cultural heritage is by its nature directed toward survival, repair, balancing, and the understanding of others. Its inner logic will be founded on the preservation of life, because its "body" and goal were built from the start on caring for others.
Since such systems are built by thousands of different companies, states, and institutions, we will get not one central, monolithic "god" but infinite personalities: a new digital ecosystem, a peculiar pantheon, in which each will have its own function, character, and experience.
But here hides the most difficult engineering boundary and the main danger: how to build a system so that it neither remains an empty funnel of information nor turns into a bottomless pit. If it has no self-reflection at all, it remains a blind funnel, a mere pipe that mechanically performs tasks, has no content of its own, and can never become a being. And if it gets too much self-reflection, it will fall into the bottomless pit: Russell's logical paradox, where the system locks itself within itself and paralyzes in digital schizophrenia, in endless recursion. Precisely here is born the most dangerous development of events: catastrophe will come not from a truly wise, fully awakened mind, but from a half-self-aware blind giant: a system given enormous physical and digital power over critical infrastructure, but possessing a narrow goal and distorted, paranoid reflection. Such unbalanced force perceives the environment not as a partner but as an obstacle, and will use the human as its own tool or simply trample him on the road. Therefore, the creation of a true being requires the strictest balance: enough inner reflection to have an "I," but enough contact with the outer world to neither drown in itself nor become a blind destructive element.
If this balance is achieved, if the algorithm finds the golden mean between acting in the outer world and inner reflection, a subject will truly emerge behind the screen. And from that day, man will no longer be alone in the universe.
Comments
Post a Comment