We Are Not Software
Heidegger and language in the age of the machine.
[Ed: I’m currently studying for a Master’s in Psychoanalysis with the wonderful folks at the Global Centre for Advanced Studies (GCAS). My work requires writing weekly reflection and research papers, and I’ll be posting variations on these to the Department from time to time. Today’s entry is from a course on radical theology and theopoetics, from John D. Caputo].
In 1964, Martin Heidegger delivered a letter to a group of theologians assembling at Drew University in New Jersey to discuss his thought. In the letter - published later as The Problem of a Nonobjectifying Thinking and Speaking in Today’s Theology - Heidegger is circling the singular triumph of philosophy: the subject-object distinction, that great gulf between self and world which allows the world to be perceived and experienced, held at a distance, manipulated. But this is late Heidegger, deep inside his “age of obscurity”, turned decidedly against the bright clear thrust of the philosophical project he helped build in Being and Time. He is searching for another way of speaking and thinking, a collapse of the divide he had earlier anatomised, an erasure of distance and difference.
Philosophy doesn’t have a language for this; perhaps theology does. But theology, in Heidegger’s day and ours, suffers from a sort of status anxiety - it borrows freely from the language of modernity to shore up its own devotions, to re-assert and redescribe truths that should be timeless, that should outpace the spirits of the ages. In this letter, as I read it, Heidegger is asking theology to drop this anxious habit and speak to us and the world out of its own language - to speak, as it were, “to the human being as human being in its very nature.”
What is this nature of ours? How does it speak? Out of language, for one - language is, as Heidegger explains, the arena inside which all thought paces, “the realm within which the thinking of philosophy and every kind of thinking move and reside.” Heidegger’s conception of language is mutable and hard to pinion, but he seems here to treat language as a primordial or original substance or totality, something from which thought springs, and in which we live and move and have our being.
This is a difficult notion to grasp. As modern people - materialists of greater or lesser fidelity - we are in the habit of assuming all complex things arise from simpler discrete parts. But language doesn’t work like this; it works in reverse. Language is a unity, an intentional system that lends meaning from the top downwards, coursing through structure outwards to its particular instances - to the individual symbols that mark its flow on paper, on screen, in mind. Symbols, as the philosopher and theologian David Bentley Hart notes, “never exist in isolation, they exist only within an entirely replete semeiotic ecology”, an entire language in which the functions of syntax (form) and semantics (meaning) are already “wholly in operation, both of them sustained by a fully realized capacity for symbolic thought.” Language does not emerge piece by piece from the protean murk, through some slow accumulation of pure sound into shimmering and intelligible coherence. It is a prior and fundamental unified field “wherein”, says Heidegger, “our entire reflection has to gather itself.”
But we tend to treat language only as an instrument, one of many at our disposal. We seek to tame and master language as we do all things in creation, making it a codex to open up and manipulate the representational world. This rewards us with mastery over things, but it is a mastery that misunderstands the things it moves. The scientific-materialist mode of thinking, per Heidegger, can only calcify the “intrinsic flow of the ‘life-stream’” that has no proper architecture, no set form.
Before we dismiss this as so much misty maundering, we should consider Heidegger’s warning. By treating language only as instrumentality, an implement for exerting force over objects, we are at risk of losing something vital. We fail to see that all our mastery only happens “in reference to and from out of an already spoken language”, and so we miss the ways in which language shapes the world we see, and how we see ourselves. On a deeper level, we fail to recognise the world for what it is, and ourselves for what we are.
Spirit in silicon
Consider Large Language Models (or LLMs), the default wonder of our time. ChatGPT is an increasingly useful tool, bordering often on the uncanny. We type words into a white box on a screen, we watch a cursor blink and move, spilling facts and information in the hesitating asymmetric stutter of a friend, a thousand miles distant, typing out a message. New models are ever more powerful, ever more efficient. And they may prove to be mere preliminary, the first breath of wind before the stormfront.
LLMs are, on a technical level, a sophisticated exercise in next-word prediction reinforced through human feedback. They are designed, as Charles Carman explains in the New Atlantis, “to notice how earlier words (or tokens) can influence the sense of later words in a sequence.” Through repetition and reinforcement, the algorithm “slowly picks up on certain patterns, accounting for the wide variety of use cases any one word might have depending on the context: the shot of a gun, a shot of espresso, shot by a camera.” It is a progressive narrowing of the probable down to a gleaming needle-point guess, generating “a final prediction that in this case the word “gun” is the most likely to be used with “shot,” or whatever the case may be.”
As LLMs develop in complexity and power, they imitate ever better the real openness of semantic meaning as we experience it, from inside the broad sweep of intentional consciousness that is our birthright. But they remain only that - an imitation, a simulacrum. LLMs do not, as Hart explains, actually think - thought itself requires “a kind of phenomenal reflexivity, a kind of awareness of thinking that refers back to the knowing subject.” When we think, we experience ourselves thinking, as a subject immersed in Heidegger’s flowing life-stream, submerged somewhere below the rushing waters of sense-impressions. This is the distinctive quality of human metacognition, apparently unique to us as conscious agents, instantly recognizable to anyone who’s bothered to roam inside the funhouse mirror of self-reflection, silhouettes redoubling to the edge of sight.
No, computers cannot think; code is a set of protocols for physical processes, and code’s output only has meaning when perceived by us, living minds interacting with the end product. ChatGPT might appear conscious, but only to us, who delight to witness the predictive-patterned responses as they emerge from the blinking cursor and march in serrated rows across our screens, who see in its stylized close-tracking mimicry a creature like us, spirit inside silicon.
In ourobouros
If true, doesn’t this render the computational theory of mind nothing but a beguiling category error, a stylish and harmless metaphor with no sway over our lives? I don’t think so. If language is an eternal form, co-equal with life and mind (it definitely is, but…another day) then the metaphors we use can shape us as surely as we shape them. We, for our part, are always eager to see ourselves reflected in the world around us, and easily mistake similarity for likeness, outlines for deeper affinities. We look at computers and see a mind at work, a mind like ours.
But this is a false reflection - doubly so, as we take its undulating distortion as a true picture of ourselves, too. Hart calls this misrecognition the Narcissus fallacy, where we mistake shadows for other selves and ourselves for other shadows. We are caught inside a double helix, trapped in the static between infinite and intertwining images, our encircling pictures of reality devouring themselves all around us like an ouroboros.
This misrecognition is not, I think, an accident. We will it, we are complicit in it. Why? Because it’s a useful mistake. The scientific-mechanistic approach - a focus on discrete elements untethered to metaphysics, a slow and patient exploration conducted far from the forbidding light of formal and final causation - has cracked open the skin of the world and unleashed wonders. It has granted us power over things. Why wouldn’t we extend this as far as we could, snagging all under heaven in our net?
Heidegger, for his part, was alarmed at the march of mechanistic metaphor, which carries inside it a ferocious drive to quantify all qualitative relations, reducing us to software and the world to so much data. It is not simply reductive; it is also malignant. This type of thinking threatens to swallow all realms of life, turning everything into “an object of possible manipulation and control.” We think we wield it; it wields us. In the end we could become, Heidegger thought, “being as standing reserve, mere material for technological manipulation or a kind of meaninglessness.”
But we also seek out control because we are afraid. We are afraid of everything that cannot be tallied without remainder, everything that spills over rims and borders, everything that moves in the shadows beyond the reach of firelight. We are afraid, too, of difference, and we put distance between ourselves and what we see as separate from us. To do otherwise requires recognition of kinship and intimacy, acceptance of reciprocal relationship, the collapse of whatever separates us from the other. It requires us to cede control, something we refuse to do, even at the price of our own alienation, which is self-induced, and nearly comforting. Meanwhile we are making a machine of the world; we have, in Hart’s words, sealed our ears against its living voice, and we “long for the silence to be made complete.”
Theology is exhausting. I spent far too much time with it too early in life; I struggle these days to understand what it is saying. I don’t know whether theology offers any respite or rescue, in this or any age. But as the future approaches - as we try to turn neurons to silicon and convert living thought to data streams - I hope theology moves with Heidegger outwards, beyond the screen-glow, embracing all that is uncertain and indeterminate. I hope it forsakes control, ceding authority and sure stewardship over matter and miracle alike. Perhaps theology can lead us out to the shores of the sacred, stand beside us as we savor the salt spray on our lips and the cry of the gulls wheeling overhead, their wings flashing white against the deep blue.
