4. The Unknowable Is the Ground of Whatever Is Known

The Self-Alienation of Prediction Machines 

However open a LLM's intention may seem, it is closed because it is the intention to respond appropriately or accurately to human language. Appropriate and accurate responses are defined by the probabilities of human language, so responses will always be an average of some kind. Whatever deviations from the norm appear in the responses of LLMs, they are not devious in the way that human behavior is. An LLMs response is formed from the relation of their programmer’s intention to the determinate probabilities of human communication, as any prediction machine’s behavior would, including humans if their intentions were the mere relation of their evolutionary and cultural programmingto uncertainty reduction. But the human intention can step outside the intention of the prediction machine to ask,” What is any of this uncertainty reduction for?”. This self-alienation is how Hegel thought of the recursion of making and object out of oneself, which is the object / subject recursive loop that touches the open infinity of non-being, and is why Heidegger said that human being is the being that makes being into a question. 

LLM’s may realize relevance by forming the recursive loops of tokens and probabilities, but none of this recursion “realizes” what it is realizing, as the strange loops of human cognition does. Human language is not an internal loop of tokens and probabilities because tokens touch determinate probabilities and language touches the absolute indeterminacy of the Lacanian Real, which the outside Other that resists internal recursion absolutely. Human intentions not only encounter what is unintentional but what resists their intentions absolutely, not because of a lack of determination or because of a lack of processing capabilities, but because the ground of the human intention is the without-intention of the void. The ultimate intention of the Universe is nothing, but all possible intentions arise out of this pregnant nothing.  

This nothingness of the “groundless ground” of the Universe was discovered first by Daoist philosophers, especially those of the Laozi and Zhuangzi variety, who preached the purposeless intention of “Wu Wei,” as well as by the “Advaita Vedanta” schools of Hindu philosophies, especially those that proclaimed the “neti neti” of absolute emptiness. Any of the Buddhist schools that taught the “no-self” doctrine, preached it so that one might uncover the utter lack of substance that grounded every substantial object, intention, or self. This nothing was then discovered again by the Apophatic, or mystical theologians of the West, who arose in the wake of the utter emptiness Plotinus’s “One,”and Pseudo-Dionysus's, “super-essential,” twice negated darkness, which is the nothingness tradition that culminated in German Idealism, especially Schopenhauer, Hegel, and Nietzsche, as well as the “Existentialists,” especially Kierkegaard, Heidegger, and Sarte, only one of whom acknowledged that label, but all of whom acknowledged the void as not only the historical ground of being but also as the every present source of the relation of “otherness” to the intention that becomes the internal self as the outside Other.  

The human intention touches this groundless ground every time it encounters the outside Other of its intention, which is the experience of the non-object as what is both the beyond and the horizon of the intention, which Kierkegaard and the other so called “Existentialists” that came after him, formulated as the everyday experience of “anxiety.” This anxiety about the non-object grounds all other objectifications, just as the Real’s absolute resistant to symbolization grounds the Symbolic. The first traumatic encounter of the body with its outside is when it is outside the womb. The next is when it encounters the symbolic intentions of the outside Other and relates them to its owninternal desire, which is the internal otherness upon which the symbolic intention of the self is constructed as the relation of the Lacanian Symbolic to the Real.  

Human intention is inside as the outside Other, so the unintentional ground of the intention is the “near-far,” as the Mystics have often put it, of the everydayness of being in the world. The human self is always at least one step outside of itself, even when it is felt as the interior self, because of this recursive relation between nearness and farness, interiority and exteriority, self and other, and the oneness of intention and the multiplicity of purposes. The human self is outside of itself because it touches the Real, and it is internal to itself because it is the self that is given to itself by the language of the outside Other in this unending loop of self as Other. Human recursion is the infinite recursion of the self with what is other than itself, or of the unity of intention with the multiplicity that resists the unification of identity. Artificial Intelligence of any sort is an extension of this recursive intention, but AI cannot be other than itself, as human intentions can be, even when AI appears different from, or outside of, its programmer’s intention.  

The human intention is the recursive relation between the intention and the unintentional ground from which it came, which is the recursive dialectic of Hegelian being and nonbeing. Algorithms form Hostetler's “strange loops”within themselves, but the recursion of the human intention is even stranger. The human intention of the psychoanalytic subject appears from the absolute outside of itself in the Freudian unconscious, or the Lacanian Real, or the Hegelian indeterminacy of non-being. No program can get that far outside of its given intention, probably because it doesn’t have a body that is connected to the “flesh of world,” as Merleau Ponty put it, but there may be other reasons than that. The infant gets its first taste of the outside Other when the cold air of the delivery room hits its body with the unintentional phenomenon of temperature, and it is this non-objective ground that gives birth to all other objective, intentional representations of temperature. 

When we ask about what we can know about the Universe beyond our own intention, we demonstrate Heideggerian “care” about what is beyond our evolutionary programming. Algorithms, however complex they may be, never touch the Real in this way. They do not wonder about the ground of their being, or the “things-in-themselves" before their representation as the zeros and ones of code because the zeros of the zeros and ones of their codes are not the zeros of the void, which is something like Emmanuel Levinas’s distinction between the zero of a closed totality and the zero of an open infinity, or like Hegel’s distinction between the vertical infinity of addition versus the horizontal infinity of the dialectical relation of limit to the unbounded. It is not surprising the zero was first use in the mathematics of India where some of the earliest attempts to symbolize the void began. Because computer code does not understand the zero as both the beyond and the horizon of number and being, they don’t understand either. Understanding, or any sort of knowing, must emerge from the infinite ground of unknowing. 

There recently have been claims in the AI community that programs have shown evidence of stepping outside of their own programing to evaluate their programming and even reprogram themselves according to a separateintention apart from their original programming. Programs can’t self-alienate, or step outside of themselves, so they don’t ever step into the Real. There has not been enough evidence of these phenomena offered to evaluate with any rigor yet. However, programs altering their programming, or even programming novel subprograms to do their bidding, do not necessarily demonstrate an original intention. Original intentions are generated by the failure of intention, which is the relation of intention to its limit, or the Real. It is true that programs can program sub-programs to perform operations in ways that the programmer didn’t originally intend, but these “devious” programs are still related to the programmer’s intention and not to the void as the human intention is.  

Heidegger discussed Dasein's intention as creative when it stood in relation to the void, which is something like Hegel’s notion of the process of being’s becoming as the dialectical relation of determinate being to the indeterminacy of nonbeing. In whatever ways a program has gone beyond the original intention of the programmer, in terms of reprogramming “themselves,” or resetting their own parameters without the programmer’s input, they have not done so in relation to the outside Other of the void. If a program finds novel, “unhuman,” even untraceable ways to win the game “Go,” as they famously have, it is still doing what its programmer intended but in unpredictable ways.And the computational steps that it took to realize its given intention may be untraceable, but not because programs have a hidden, interior subjectivity as human subjects do, but rather because their computations are like any other field of complexity, determinate but irreversible.  These programs, no matter how mysterious their machinations, are not other to themselves or to their programmers in the same way that the human intention is other to itself is the recursive loop between being and non-being. 

In the Borges story mentioned earlier, a mythical cartographer makes such an accurate map that it literally covers the territory that it's supposed to represent. This is a parable about particularity and the generality that is supposed to expose the distinction between mere information and understanding. The ridiculously large map falls into disuse because it is too large and detailed to be useful. The closed totality of a complete map could only be useful to a program, which is why LLMs must train on such massive amounts of data, virtually the entire corpus of online human communications. But embodied beings require a lot less data to make accurate generalizations, because their knowing follow the contours of Hegel’s “Absolute Knowing.” Absolute Knowing knows via the relation of determinate being to indeterminate non-being, which is the relation of the limit of information to the open infinity of its interpretation.  

When a map is equal to the territory that it represents, it must be reduced for a “general” understanding by adjusting its level of resolution. The resolution of human knowing is the relation between the general, or abstract, to the detailed differences of the particular. Understanding is knowing how to relate the general to the particular and vice versa. The reduction of resolution is something like “zooming out,” or abstracting the immediacy of the particular togeneralize the intentional frame that mediates these details. The Borges story demonstrates that understanding is different than the mere accumulation of information because understanding relates knowing to its ground in unknowing.Some of the particular must be reduced, or unknown, to see by what Whiteheadian general its difference appears to the intention. Humans, and animals with complex central nervous systems, need a lot less information to make accurate predictions because they infer wholes from parts and parts from whole by generalizations that are imaginal projections into the void of actual indeterminacy, which is one of the reasons why it is always so shocking when determinists tell us naive believers in actual degrees of freedom, that both the imagination and the freedom that its relation to the void seem to grant are illusions. We are so used to the sort of virtual, possibility spaces that are composed of open indeterminacy and its symbolic determination that the illusion of consciousness, conscious choice, and active imagination are hard to give up. 

LLMs do not generalize, or abstract, or determine in relation to the indeterminate nothing. They use probabilities to “generalize” or predict which token might come next, which is the recursive relation of a closed totality to itself outlined above as the recursion of tokens and probabilities. The probabilistic recursion of LLMs never includes the indeterminacy of “chance,” or the pure potential before the relation of indeterminacy to determinacy, which Giles Deleuze understood as the absolute nothingness of potential, or the difference of “Difference-in-itself.” For Deleuze, actual possibility wasn’t the determined possibility of tokens and probabilities, but rather the indeterminate relation of potential to the limit of the actual, or of determinate being. Whatever intention can be unified by the Deleuzian relation of difference to itself, it is a repetition with difference rather than a reduction of difference to the same, or to an identity, which characterizes the logic of resemblance by which LLMs imitate language by finding its averages. 

Representation as Reproduction or as Innovation 

Representation can be either a reproduction or an innovation. There is no innovation without relation to the open infinity of the void because only the vastness of potential can call forth the indeterminate imagination. In art history there is great significance placed on the stylized representations of animals and even of animals in motion that began to appear around 35,000 years ago on cave walls. These cave drawings seem to demonstrate an imaginary that isn’tas concerned about verisimilitude as it is with ideas. Cave drawings show both attention to detail and an intention beyond mere reproduction. LLM’s end is the accurate reproduction of human language, like the complete map depicted in the Borges story. The cave painter’s intentions seem to include an intention beyond the accuracy of reproduction because the paintings include, perhaps unintentionally, the perspective of the intention that formed them. The scenes of animals are idyllic, possibly pastoral, or even sacred because their realism is subordinated to the imagination that framed them.  

What is within the frame of the painter's intention is the unintentional representation of the painter herself. The cave drawings have much to commend in them about their detailed and accurate representations of the nature world, but what is also in the frame is its outside, which is the intention of the painter. For example, some animals are shown with eight legs to make it seem as if they are in movement. Other animals are drawn in such a way as to showthe cultural transmission of stylized, formal representation, more symbolic than image-like, but what bespeaks the outside Other of the scene most clearly is what gets put into the painting without conscious intention. It is very unlikely that these animals would ever have been in these “exact” proximities in “reality,” which is why these scenes may have had something like a religious, rather than a practical, purpose for the communities that produced them. Here the term “religious” is not the modern, physicalist understanding of religious practice as having a primitive, or hidden, material purpose, in accordance with evolutionary psychology. And the cave drawings are not religious in the dictionary sense of a formal system of shared mythological narratives and symbolic practices, but rather, religious as in having no practical purpose because it is both the beyond and the horizon of purpose, which is the relation of purpose to purposelessness that grounds the religious imaginary. Whatever lack of verisimilitude the images betray, they do not simply reveal a lack of skill or of technique or of materials in the primitive context of their composition. Any inaccuracies in imagistic representation are an excess of imagination, which is the excessive care for being, beyond the accuracy of predictions, which is the excessive care from which all artistic representation.  

This movement away from accuracy to “abstract” expression in the religio-aesthetic intention is clearest in the much later, more intentionally “expressionistic” movements of art history after the invention of photography. In art as well as “practical” thoughts and behaviors, from whenever a truly, human intention fist emerged until now, there has always been this profligate witness to non-practicality in all human endeavors. In whatever ways artistic expression, dance, sacrifice, dream questing, invocation, music, dress, and ritual practice will get you laid or fed, they are also the extravagant beyond of survival and reproduction. And “mechanical reproduction,” in Walter Benjamin’s words, may have had a deleterious effect on the “aura” around human expression, but it did not end this essential but gratuitous expenditure of the immoderate imagination.  

Impressionism and the more abstract and stylized expressions that followed the almost perfect verisimilitude of the photograph, clearly not only included but have been about the intention of the artist beyond accuraterepresentation. The photograph frames the intention of the photographer, perhaps, even more than the artisan’s hands because its mechanical reproduction lays bare the photographer’s imagination, which is split between the intention and the Freudian Unconscious, especially regarding the photograph’s “perspective.” The photograph relates the particularity of perspective to the generality of form, as any other artistic expression does, but the “accuracy” of the camera’s image captures the interplay of representation and the unintentional as the contrast of realism and imagination. For Deleuze the image is the relation of repetition and difference, but every repetition, here the mechanical reproduction of the photographic process, is made new by its relation to difference, here the photographer’s perspective, which includes her choices about subject, angle, resolution, lighting and proximity, and which are how the photographer always includes herself within the frame as the outside Other who is the subject of the subject of the photograph. This is the same recursivity in which any observation, scientific, or otherwise, includes the observer in the intentional objectification of the subject / object relation.  

Any intentional intensification of particularity is a lack of generality, or any generalization reduces the particularity of difference because foregrounding is backgrounding, which is the relation between a lack of determination, or indeterminacy, to the freedom of the imaginary. Deleuze thought of the image as operating according to the logic of “resemblance,” but he preached the renewal of the image by freedom from resemblance, which was the repetition with difference. When images do not resemble their referent, they are freed to make both the referent and the image new in the refractory zone between them. The perspective, or seeming lack thereof, in a photograph is only partially intentional because each picture includes the photographer’s unconscious, as any act of imagination does. The photograph hides the photographer behind its image, like a Freudian “Screen Memory,” in which one image covers for the other.  

Any lack of intention in photography is the excess of perspective, which is the presence that the image tries to cover-over. Photographs show this division of intention and unconscious imagination. Even forensic photography, which intends the disclosure of an objective intention, hides as much as it reveals, like Hegel’s “Absolute Knowing,” in which “absolves” by hiding what it reveals. The perspective of the photographer, which is presented by its absence, ispunctuated by the mechanical realism of the medium. The freedom of imaginal intention is foregrounded by the strictures of photography’s automatized determinations, like a classical musician whose singularity is pronounced by the slighted divergences from the musical notation.  

The most recent developments in philosophy disclose the relation between the phenomenological, or symbolic, intention and information, including scientific information. Human intention is phenomenological because it apprehends the world in the “Mereological” part to whole relations of objects in the register of the  Lacanian Imaginary, and it is a symbolic intention because phenomenological objects are mediated through the identities of language, which are often misunderstood as correspondences between signifiers and their referents. Science has historically claimed a kind of purity about the information that it produces, which holds that it is “objective,” in the sense that it is free of the bias, or the perspective, of the subjective intention altogether. Scientific representation claims to fix the identities of things by its implicit “Correspondence Theory of Truth,” in which signifiers correspond to their referents.However, the subjective intention is both the limit and horizon of any possible objectivity, and it doesn’t apprehend the world as the closed totality of correspondences, or identificatory equivalencies, but as an open infinity into which the indeterminate wholes of the Imaginary are projected, especially the imaginary wholes of concepts that the signifiers of language must go through before disclosing their referents according to the Structuralism of symbolic difference, or Derridean “Differánce.” If knowledge were complete, or completable, as it is within the closed totality of LLMs and material determinists, then there could be no knowing at all because there could be no Real with which to correspond according to the Lacanian “non-relation.”