Why AI Is Never “Now”
by Deborah DeSilets
Be warned. The singularity is coming. AI will surpass humanity. Convergence is inevitable. These claims flood headlines, fuel investment rounds, and quietly terrorize the public imagination. They are also, at their foundation, a category error dressed up as prophecy.
AI cannot do one thing every human does effortlessly: exist in the present moment. Not because the hardware is insufficient or the training data incomplete, but because the human “Now” requires a kind of being no information-processing system can possess. The gap between human and artificial intelligence is not a distance to be closed. It is a boundary between different kinds of existence.
The Convergence Thesis Is Fearmongering, Not Science
The convergence narrative rests on a single unexamined assumption: that human and AI intelligence differ in degree, not in kind. Scale up the compute, expand the dataset, and refine the architecture— and eventually the machine thinks like a person. This logic is commercially convenient, narratively dramatic, and philosophically bankrupt.
The differences are structural. Evidence from action recognition research makes this plain. When humans identify actions on video, they lock immediately onto semantically critical zones — the hands, the point of contact, where human agency is most visible. Researchers call these Minimal Identifiable Recognition Crops (MIRCs). Humans can identify actions accurately even when the temporal sequence is scrambled, because they are not reading a sequence. They are reading meaning inscribed in space by another agent they recognize as an agent.
AI models do the opposite. They degrade smoothly under spatial reduction, spreading attention across low-level distributed features, and show insensitivity to temporal scrambling — not because they have achieved semantic grounding, but because they were never tracking temporal structure to begin with. More damning: AI systems maintain high confidence precisely where human performance has already collapsed. They do not know what they do not know. That is not a calibration bug. It is the structural signature of pattern matching mistaken for understanding.
Time Is Not a Variable — It Is the Architecture
Philosophers saw this coming long before the AI boom. Bergson argued that human time is not a sequence of discrete moments but a flowing duration — past and future folded continuously into the present, giving it weight and texture. Heidegger showed that human existence is constitutively being-toward-death: because we are finite, every moment is irreversible and therefore matters. Merleau-Ponty grounded all of this in the body — a living organism that ages, tires, hungers, and moves through a world that pushes back.
Together, they describe a “Now” AI simply does not have access to. Human cognition is not a process that happens to occur in time. It is constituted by time — by mortality, embodied urgency, and the felt pressure of a future that is closing. An AI system operates from an eternal present: a synchronic retrieval space with no duration, no anticipation, no stakes. Nothing is irreversible for a system that cannot die. Nothing matters to a system that cannot lose anything. This is not poetry. It is ontology.
Why This Matters
This is not a pessimistic conclusion. AI is genuinely transformative, and its capabilities will continue to expand in ways that benefit humanity. But the convergence thesis mistakes quantitative scaling for ontological transformation. Human intelligence is not merely faster or better-trained than current AI. It is constituted by lived duration, embodied intentionality, and mortal urgency — features that cannot be engineered from the outside, because they are not features added to cognition. They are the architecture through which cognition becomes human.
AI is never “Now” because the Now it would need to inhabit requires a life — a body that ages, a future that closes, a present moment that therefore matters.
The gospel knew this long before the first neural network was trained. It staked everything on the claim that what the world needs is not infinite intelligence delivered from a position of safety — but love that enters the irreversible, bears the loss, and comes through the other side still wearing the wounds. We are being asked, with some urgency, to resist accepting a simulation of that love in place of the real thing. Not because the simulation is malicious, but because the hunger it appears to satisfy is one that only genuine, mortal, costly human love can actually feed.
Please contact Deborah at [email protected]
