One Thousand AIs Could Not Write This

Against the logos-lacking machines (LLMs)

“From all bodies together, we cannot obtain one little thought; this is impossible and of another order.” —Pascal

“Is me, is mine.” —A two-year-old

For a while now, I have been saying that it is improbable that one thousand monkeys should compose an English sonnet, but impossible that one thousand AIs should compose the same. This clever quip has been terribly under-appreciated and “done poor numbers,” perhaps owing to the prejudices of the algorithm. But the sad fact that my wit is inadequately admired must not detain us. Far more important than the quip’s cleverness is its truth. And truth does not stand in need of honor, thankfully. Nor does respect shown to truth properly accrue to the one who voices it. The dignity of truth proceeds from and subsists in truth itself. (How wonderful it is!)

What do I mean when I say this: that it is improbable that one thousand monkeys should compose an English sonnet, but impossible that one thousand AIs should compose the same? Or perhaps more pertinently: that it is improbable that one thousand monkeys should compose an English essay, but impossible that one thousand AIs should compose the same?

The answer lies in the question. Rather, in its quiet presupposition. We take for granted that behind my words — in and under them — abides some radiant conception of my mind, which my words are intended to articulate in mutually familiar signs, thereby occasioning — nay, effecting — a communion of intellects, yours and mine. (Yes, I will use the em-dash, and even deploy a juxtaposition or two, refusing to cede an inch of lexical, grammatical, punctuational, or rhetorical territory to the chatbots!) This conception is the substance of the “meaning” the question has in view.

Without the conception, there is no language or exercise of language, rightly and fully, since the outward word is merely the faint likeness of the inward word, itself the faint likeness of the idea, which is a participation in the intelligible form beheld by the mind. Language — its deep fundaments — is therefore the spontaneous self-ordering of intellection as it grasps ahold of that which is. The exercise of language is intelligence in ecstasy, intelligence going beyond itself, that it might share its reflected and refulgent conceptions of being, and thus facilitate the unity of shared contemplation with other intellects.

Hence my contention that an AI cannot write a sonnet or an essay, or anything else for that matter: a recipe, a limerick, a hymn, a short story, an ad, an FAQ, a response to this article. By “write” I mean to visibly signify intelligible realities apprehended by the intellect by clothing them in the vestiture of legible symbol, that they (the intelligible realities) might be imparted to others. The things we call “Grok” and “Chat” and “Claude” and so on, they can produce artifacts that superficially mimic the exercise and fruits of language, but unless we are willing to traffic in equivocation, we must assert that they do not, cannot, could not, write anything. They are word-less, because conception-less, because intelligence-less.

For, as we have already said, language is intimately wed to conception, and thus to intelligence, which they manifestly have not, intelligence being a spiritual power which by its subtlety and celerity attains to intelligible forms, which are immaterial realities. (I appreciate that some self-styled “philosophers” dispute this, but I cannot hope to correct them if they are unable to perceive the error of their position by a few moments’ silent introspection.)

Let us descend from the realm of abstraction. I present the following “sonnet,” which I forced myself to elicit from “Chat”:

Upon the hills the golden sunshine falls,
And wakes the meadow from its silver sleep;
It paints bright fire on ancient stone-built walls,
And stirs the silent valleys dark and deep.

Through emerald leaves it weaves its shining thread,
A gentle hand upon the morning air;
It crowns with light each blossom’s bowed head,
And turns the common field to something rare.

The clouds retreat before its warm command,
Their shadows scattered far across the sky;
The world seems newly fashioned by its hand,
As if creation’s dawn were drawing nigh.

O sunshine, brief yet faithful in your grace,
You lend the earth a brighter, kinder face.

Not bad for a calculator, eh? But let us consider the thing for a moment, privileging reality over appearance.

Not only has “Chat” never seen or felt sunshine, it does not actually know what sunshine is. To know what sunshine is would require an operation of intelligence. Which “Chat” lacks. Ditto “hills” and “meadow,” “crowns” and “scattered,” “far” and “newly,” “you” and “if.” These words — or verbal semblances — mean nothing to “Chat,” nor, strictly speaking, mean anything from “Chat.” Hence they have no more import than a whistling wind that howls (seems to howl): “Hellooo!” Moreover, “Chat” does not know what a sonnet is, or even what poetry is — since, at the risk of repetition, it knows nothing, period, full stop, being constitutionally incapable of knowledge. Nor did it intend to write a sonnet and supply the same to me. For just as conception is a certain fecundity of intelligence, so intention is a certain determinate resolution of will, which also is a faculty of mind, and so wholly alien to “Chat.”

The artifact under consideration — the sonnet-seeming thing — is therefore unconceived and unintended. It has no spiritual reality to or about it. It is not an act or creature of language. It is instead a contrived assemblage of hollow signs, pseudo-signs, signs bereft of meaning and divorced from intelligence and will, from mind (and thus hardly worthy of the term “sign”). A kind of simulation of language fabricated by a super-sophisticated abacus designed by very shrewd (and arguably insane) mechanics, who have released upon the world talking machines calibrated to lure people into conversing with, and soliciting pseudo-verbal “content” from, a thing less animate than a smooth stone at the bottom of an ancient New England riverbed. (For which offense against sacred humanity they should be compelled to dig holes in the earth until finally they taste what is really real!)

Take the title of this article, “One Thousand AIs Could Not Write This.” This, this, is the luminous bloom of spirit! It is not a probabilistically-patterned collection of betokened marks, but a deliberate setting down and setting forth of my mind so as to reveal that which I — I, a human being! — have inwardly birthed and beheld in the fertile mirror of the intellect, which captures and reflects all things, and somehow even is all things.

Yet one thousand AIs, one million AIs, are individually and collectively incapable of emitting a glimmer of a flicker of mental flame. A child of four or five could write this, or something approximating it essentially. I dare to say, with due reservations, that a monkey could write this. At least, I could imagine how a monkey might write this, if maybe some bright ray of intellection briefly enlightened its sensuous cogitations, opening the eye of the simian mind, that it might penetrate the accidents of things and obtain their blessed forms, and granting an attendant capacity for rational explication, calloused monkey-digits striking the keyboard with commendable speed and dexterity. But, praise God, one thousand AIs could not write this. They are logos-lacking machines, LLMs. And LLMs can only feign the most extrinsic qualities of language.

I hear the technophile inquire: “What’s the difference? It [the sonnet-seeming thing] makes sense to me!” It “makes sense” to me, too, but only by way of an accommodated construction introduced by the operation of my own intellect, which by its light makes all things to be, or seem to be, light.

By way of illustration. When a parrot squawks, “Polly want a cracker,” we understand the words, but we also understand that it does not understand the words: that it does not know what it is saying, does not mean anything by the words in particular, does not intend to convey any conceptions sprung from the majestic font of intelligence. It has an instinctive craving for sensible satisfactions, and it has found that a given set of sounds reliably occasions the delivery of a crunchy morsel which puts its vague but persistent yearnings temporarily to rest. The most rudimentary apprehension of cause and effect.

But the parrot has not the slightest recognition (re-cognition) of the significance of, e.g., the indefinite article as opposed to the definite article. It might have learned to squawk, “Polly want the cracker.” “A” and “the” are bare sounds on its beak and to its ear. When it squawks “a,” it does not intend indefinition over and against definition. Indeed, it does not perceive or intend to convey any sort of meaning, as we understand meaning, because it lacks the light of intellection. I do not deny, of course, that the parrot is sentient: it has the elements of what might be called inner life or psychological subjecthood (memory, imagination, aversions, attractions, and so on). I only deny that its mind, such as it is, is imprinted and informed by logical light, the splendor of logos.

Do not mistake what I am saying: as compared to “Chat,” the parrot is almost seraphic. I am able to make myself believe that I might enjoy real communion with a parrot, under some marvelous circumstances. When the divine wisdom is all in all, maybe. The parrot has a very remote disposition to intelligence. The monkey, too, although its disposition is more proximate yet. But the chatbot is as far removed from intelligence — hence from conception, from intention, from expression, from communion — as a root or a rock. Actually, at the hazard of dabbling in fancy: the root and the rock have a greater theoretical aptitude for intellection and its dazzling entailments than does the AI. I might someday look upon a mountain and hear it speak wisdom to my heart, wisdom from the mysterious recesses of its being. But not so “Chat,” etc.

One thousand AIs could not write this any more than one thousand Magic 8 Balls could write this. If you could design a Magic 8 Ball that might, upon being shaken, produce this essay in its cloudy window, you would know it to be nonsense, a trick of artifacture, made semi-intelligible only by reproduction in the mirror of the intellect of some subject, or by its distant relation to the intelligence of the Magic 8 Ball’s fabricator. For it is intelligence that imbues all with logical lucidity, otherwise absent. Yet you — the rhetorical “you” — are prepared to accept the meaningfulness of the sonnet-seeming thing, to take it as human and humane expression…? Ah, you might then be ensorcelled by the murky messages of the Magic 8 Ball.

Take care, O reader mine, in this most treacherous moment. The AIs have neither mind nor meaning-making capacity, but only a weird figment of logos, a cartoonish imitation of the image of God. The intelligence that does not eschew this shadow of shade will become benighted thereby, invariably. “If then the light that is in thee be darkness…” So spurn the LLMs as much as you can, the logos-lacking machines.

And give not the holy to chatbots!