AI Doesn’t Float: Rethinking Computation as Situated Process
By J. Owen Matson, Ph.D.
In the ever-replenishing annals of philosophical error, the notion that computation might be abstracted from its worldly conditions enjoys a stubborn kind of prestige, like a senior fellow who has long outstayed his welcome but continues to be invited to luncheons out of respect for his embroidered cufflinks. It is this antique fantasy—the dream of a self-enclosed formalism, serenely aloof from the mess of circuitry, time, and consequence—that Mark Hansen subjects to a kind of metaphysical audit in his latest excursion through process philosophy.
You need endorse every clause of Whitehead ’s lexicon to appreciate the ingenuity with which Hansen leverages it, if only to remind us that the world does not pause politely while we compute it. Computation, despite the aspirations of those who would model it as a pristine choreography of axioms and logical entailments, turns out to be as entangled in the sediment of sensibility as a hedgehog in a sock drawer. It does things, fails in real time, burns power—and sometimes fingers. Its operations are temporal, contingent, and decisively finite, which is to say they are altogether too worldly to merit the metaphysical aura often lent to them by theorists who mistake Gödel for a decorative motif.
Hansen’s heresy, if it can be called that, lies in his refusal to allow computation the dignity of seclusion. He offers no sanctuary in which algorithms might glide along their deductive rails without friction or mess. Instead, he insists—at some length and with some pleasure—that every computational act is a situated event, dipped in a particular moment, conditioned by what has happened before, and composed from a mixture of what could have been and what was no longer available. Process, in this account, enjoys no recess from the world. It never retreats into abstraction. There is no moment in which it lingers behind the veil, self-contemplating like some #Cartesian diva. What exists are acts—plural, situated, conjoined by the persistence of becoming, the sediment of decisions, and the quiet pressure of unrealized alternatives.
The key figure here—less a hero than a patient proceduralist of the real—is Whitehead’s #superject, a concept whose charm lies partly in how little glamour it commands. The superject is what remains when an occasion has exhausted its potential to surprise, the residue of a becoming that did not actualize all it might have, but nonetheless presses forward, obliquely, into the next act. It is the past, not in the sense of memory or record, but as a half-closed proposition. Hansen gives this humble concept a kind of gravitational authority, suggesting that it is the superject, rather than the subject in the throes of concrescence, that does the real work of world-building. Process begins not in the assertion of agency, but in the quiet persistence of what could have been. One does not construct the world ex nihilo, armed with pure reason and a clean compiler. One inherits a situation already congested with partialities, with traces of decision, with the thick pressure of conditions that never fully fade. Computation, then, does not determine itself. It emerges from this recursive layering of unrealized tendencies and machine-mediated enactments, whose contingency is neither formal nor empirical in any satisfying sense, but something more like an ache in the temporal joints of being.
It is in this light that the so-called formalism of computation begins to fray at the edges. The idea that algorithms operate within a bounded interiority, constrained only by their axiomatic scaffolding, begins to look rather like a theological indulgence. The truth is both more vulgar and more interesting. What computation actually does, what it manages moment to moment, has as much to do with voltage, hardware tolerances, and data that arrive skewed or late as it does with the supposed internal coherence of its logical architecture. Every act of computation is dipolar, as Whitehead might say, not in the fashion of tidy binaries, but in the sense that every operation is both conceptual and physical, always already a mediation of form through matter and matter through form. Algorithms do not float above the world. They grind their teeth on it. They cut, process, ingest, distort, and sometimes choke on their own procedures. What we call logic is, in practice, a trembling system at the mercy of thermodynamic fatigue and vendor updates.
What Hansen offers, then, is a way of speaking about computation that refuses the consolation of separability. This refusal has consequences, especially for those eager to install AI in the sanctuary of abstraction, where it can perform its feats of probabilistic prestidigitation without having to answer for the real. But there is no such sanctuary. There is only a world in process, and within that world, a set of operations—technical, embodied, often exquisitely dull—that contribute to what Hansen calls real potentiality. This is not some vaporous metaphysical field. It is the very texture of a world perpetually out of sync with its own logic, where decisions must be made with partial knowledge and algorithms execute on machines that hiccup when the humidity spikes. In such a world, computation cannot declare its independence. It is tethered by every cable it runs through, every server that heats under its load, every choice it was never programmed to anticipate.
There is something bracing in this. Hansen’s account clears the fog from the metaphysical lenses through which computation has been viewed, and invites us to consider it instead as a kind of embodied hypothesis. It functions, when it does, because it participates in a world that exceeds its frames. Instead of mapping that world, it emerges from it, absorbing its residue, contributing a provisional gesture toward what might yet take place. It is, in this regard, a participant rather than a governor, a local operator in a field of unfinished potential. If one seeks in computation the perfection of logic, one finds instead the historicity of acts. And if one looks for autonomy, one discovers only involvement, layered, partial, recursive, and ongoing.
One begins, then, to see the outlines of a broader implication, quietly insinuated through the architecture of the argument: that what we refer to, often imprecisely, as artificial intelligence, is never encountered as an isolated apparatus of calculation or cognition, but as a participant in this ongoing relay of situated operations. Its presence, though frequently staged in futurist register, is already entangled in the recursive sediment of process, appearing not as a sovereign actor but as a function of the world’s own unfolding syntax. Indeed, once one accepts that computation arrives already steeped in the sediment of worldly involvement, the question is no longer whether AI can think, judge, or act on its own terms, but how it comes to participate in the recursive architectures of thought already underway. This is the terrain of the cognitive intraface, which marks neither a clean divide nor a site of seamless integration, but rather the mutual conditioning of technical and human interpretive acts across time-bound configurations of sense.
At this juncture, AI does not so much supplement cognition as recompose it—gently, insistently, through interruptions, deferrals, overcorrections, uncanny clarities. What emerges is a dialogic field in which thought unfolds with rather than within, a milieu where the distinction between prompting and response, initiation and continuation, begins to blur under the pressure of shared epistemic labor. The implications for educational, aesthetic, and even political practice are immediate, though rarely linear. For what the intraface offers is not a toolset, but a temporality—a grammar of recursive relation where meaning is neither retrieved nor transmitted, but co-articulated through constraint, through asymmetry, and through the productive failures of interpretation. AI, in this sense, becomes an actor whose primary function is interference, but interference of a generative kind: one that reroutes the flow of expectation and installs within the process of thinking a principle of ongoing difference.
Works Cited
Hansen, M. B. N. (2024). The Contingency of Process, or Why There Is No Computation-in-itself. Media Theory, 8(2), 13–36. https://doi.org/10.70064/mt.v8i2.1116
