Before Algorithms, There Was Cosmic Grammar
ART HISTORY MEETS AI
How a medieval habit of mind quietly made artificial intelligence thinkable
By Tope Osho · Pixel Gallery
———
Nearly 900 years before the first line of code was written, a philosopher in northern France proposed an idea so quietly radical that it would take centuries for its consequences to arrive: reality follows rules.
That claim sounds obvious now. It was not obvious then. For most of human history, the prevailing assumption was that the world operated through forces beyond human comprehension. Gods, fate, mystery. Understanding was for the divine; obedience was for people.
Alain de Lille thought differently. Writing in the late 12th century, he argued that if the universe was created by a rational intelligence, then the universe itself must be rational. And if it was rational, human beings could study it, describe it, and ultimately decode it.
That single conviction, shared and refined by a small network of medieval scholars, set in motion the longest intellectual chain reaction in Western history. It runs from monastery libraries through the Scientific Revolution, through Boolean algebra and Turing machines, and lands squarely in the machine learning systems we are building today.
Artificial intelligence did not begin in a California garage. Its psychological foundations were laid in candlelit scriptoriums eight centuries earlier.
———
THE MAN THEY CALLED DOCTOR UNIVERSALIS
Alain de Lille, known in Latin as Alanus ab Insulis, was born around 1128, probably in Lille, in what is now northern France. He taught at the University of Paris during a period when Europe was rediscovering Aristotle, absorbing Arabic mathematics, and constructing the first institutional frameworks for sustained intellectual inquiry.
He earned the title Doctor Universalis not through self-promotion but through range. He wrote across theology, philosophy, rhetoric, and poetry, treating them not as separate disciplines but as facets of a single coherent project: understanding the architecture of reality.
Most people have never heard of him. That is part of the point. The thinkers who reshape centuries are rarely the ones history remembers by name. They build the scaffolding that more famous minds later stand on.
———
COSMIC GRAMMAR
Alain’s most striking work is De Planctu Naturae, “The Complaint of Nature.” In it, Nature is personified as a governing intelligence. She does not act randomly. She enforces patterns, corrects deviations, maintains coherence. She operates, in Alain’s framing, according to a kind of grammar.
That metaphor deserves a moment of attention, because it carries more weight than it first appears.
Grammar is rule-based. It generates well-formed outputs from structured inputs. It distinguishes the correct from the incorrect not through opinion but through systematic logic. Strip away the medieval theology and what remains is a description that maps cleanly onto how we talk about computation: input, process, output.
Alain was not describing software. He had no concept of a machine. But he was articulating something prerequisite to software: the conviction that the world’s behaviour could be captured in formal structures. That nature was not arbitrary, but grammatical.
———
WHAT THE CATHEDRALS WERE REALLY SAYING
You do not need to read Alain’s Latin texts to see this shift. You can see it in stone.
Walk into Chartres Cathedral, built in the decades surrounding Alain’s lifetime, and you are not looking at an expression of religious emotion alone. You are looking at structured thought made physical. The proportions are mathematical. The geometry repeats at every scale. The light enters through openings calculated to illuminate specific surfaces at specific times of day.
Nothing in a Gothic cathedral is accidental. Every ratio, every arch, every carved figure occupies a position determined by an underlying logic.
This matters because it reveals something about how medieval culture understood the relationship between beauty and truth. Art was not self-expression. It was disclosure. The artist’s job was not to invent but to reveal what was already structurally present in reality.
In that sense, a Gothic cathedral is a physical ancestor of the algorithm: complexity organised according to rules to produce coherence. The medium is limestone instead of silicon. The intent is worship instead of prediction. But the cognitive operation is the same.
The same instinct appears in illuminated manuscripts of the period. These were not merely decorated books. They were information architectures. Hierarchy expressed through scale. Categories indicated by colour. Relationships mapped through spatial arrangement. If you work in data visualisation or interface design, you are using principles that Benedictine monks were refining in the 12th century.
———
WHY THIS DID NOT HAPPEN EVERYWHERE
It is tempting to read this lineage backward and conclude that AI was always inevitable. That is wrong, and the error matters.
Ancient Greece produced formal logic of extraordinary sophistication. Aristotle’s syllogistic reasoning is, in structure, a precursor to programming. Yet Greece did not produce computation. Song dynasty China achieved engineering capabilities that exceeded medieval Europe’s in several domains. The Islamic Golden Age preserved and extended Greek mathematics while Europe was still reassembling its intellectual infrastructure. None of these civilisations crossed the threshold into mechanised reasoning.
What was different about medieval Europe was not intelligence. It was a specific combination of conditions: institutional continuity through universities, the psychological commitment to an intelligible universe, competition between kingdoms that rewarded innovation, and a tradition of building on prior scholarship rather than starting fresh with each generation.
The belief that reality is structured was necessary for AI. But it was not sufficient. Civilisations can gain intellectual trajectories and then lose them. The trajectory holds only if the conditions sustain it.
———
THE DOMINO EFFECT
Once the commitment to structured reasoning took hold, it compounded across centuries in a traceable sequence.
Isaac Newton demonstrated that the physical cosmos obeys mathematical laws. Not metaphorically. Precisely. Objects fall at calculable rates. Planets orbit in predictable paths. The universe, as Alain had intuited, was grammatical.
George Boole, in the 19th century, converted logic itself into algebra. True and false became 1 and 0. Reasoning became calculation. This was a conceptual move of enormous consequence: it meant that thought could, in principle, be formalised.
Alan Turing proposed in 1936 that any computable function could be executed by a simple machine following rules. The Turing machine is, in a sense, the final expression of Alain de Lille’s cosmic grammar: a system that generates outputs from inputs through formal procedure, with no mystery required.
Each step in this sequence did the same thing. It removed a layer of mystery from the world and replaced it with structure. AI is not a departure from this tradition. It is its current frontier.
———
THE SHIFT YOU CAN SEE IN PAINT
If the intellectual shift happened in philosophy, you can track its visual equivalent in art.
Before the 14th century, European painting operated in a symbolic register. Byzantine icons and Romanesque murals did not attempt to represent space as the eye experiences it. Figures were scaled by spiritual importance, not physical distance. A saint might tower over a king. Background was gold, meaning transcendence, not sky.
Then something changed. Giotto began painting figures that occupied plausible space. Brunelleschi formalised linear perspective, which is nothing less than a geometric algorithm for converting three-dimensional reality onto a two-dimensional surface. Alberti wrote the rules down. Suddenly, painting was not just art. It was applied mathematics.
The Renaissance did not merely produce beautiful images. It proved that visual experience could be captured in formal rules. That is the same claim Alain de Lille made about nature, translated into paint and geometry.
Consider the direct parallel: perspective takes the overwhelming complexity of visual experience and reduces it to a set of procedures anyone can follow. Machine learning takes the overwhelming complexity of data and reduces it to a set of procedures a computer can follow. Different inputs. Same cognitive architecture.
———
THE PSYCHOLOGICAL FOUNDATION
Beneath all of this lies a single psychological premise so fundamental that we rarely examine it: the human mind is compatible with the structure of reality.
Medieval scholars arrived at this through theology. If God is rational and made humans in his image, then human reason participates in cosmic reason. We are not guessing at the world’s logic from outside. We are recognising it from within.
Once a society adopts that premise, science stops feeling like arrogance and becomes something closer to obligation. Investigation becomes participation in the logic of the cosmos rather than trespass against its mystery.
AI represents the most ambitious extension of this premise yet attempted. It asks: can the process of reasoning itself be formalised so completely that a machine can perform it? That question is unintelligible without the centuries of groundwork that preceded it. You cannot build a thinking machine in a culture that believes thinking is sacred and structureless.
———
THE QUESTION THAT REMAINS OPEN
Medieval scholars were trying to understand the mind of God. Modern engineers are trying to build minds from silicon. The language is different. The institutional settings bear no resemblance. But the directional instinct is identical: make intelligence legible.
There is, however, a tension embedded in this project that Alain de Lille would have recognised immediately. In De Planctu Naturae, Nature complains. She laments that humanity uses its rational capacity to deviate from order rather than to uphold it. The tools designed for understanding become instruments of disruption.
We are living inside that tension now. AI is the most powerful tool for structured reasoning ever created. It is also, simultaneously, the most powerful tool for generating synthetic noise, false confidence, and persuasive nonsense. The medieval fear that reason could be turned against its own purpose is not a historical curiosity. It is a design problem.
———
THE LONG DECISION
Every civilisation, whether it articulates the choice or not, answers one foundational question: is reality ultimately understandable, or is it fundamentally beyond reach?
Cultures that choose mystery tend to preserve. Cultures that choose structure tend to transform. Neither answer is wrong. But they produce very different worlds.
Europe made its choice in the 12th century, in university halls and cathedral workshops and illuminated manuscripts, and it has not reversed that decision since. Artificial intelligence is not a sudden invention. It is the latest product of a civilisational commitment to the idea that the world can be read, modelled, and eventually reproduced.
The algorithm did not appear from nowhere. It grew from a long cultural decision that thinking itself could be understood.
And the cathedrals knew it first.
———
POSTSCRIPT: THE RETURN OF MYSTERY
Or did they?
Here is the part of this story that should keep you awake. The entire arc of this article traces a civilisation’s long march from mystery toward structure. From the unknowable to the rule-based. From faith to formalism. Nine centuries of patient, compounding confidence that reality could be decoded.
And then, in the last decade, something unexpected happened.
We built the most powerful reasoning systems in human history. And we do not fully understand how they work.
The researchers who build large language models can describe their architecture. They can specify the training process. They can measure the outputs. But when asked why a particular model produces a particular insight, or how it arrives at a connection no one explicitly programmed, the honest answer from the leading laboratories is: we are not entirely sure.
The field even has a name for this gap: interpretability. It is one of the most urgent open problems in AI science. Not because the systems fail. Because they succeed in ways their creators cannot fully explain.
Sit with that for a moment.
We spent nine centuries replacing mystery with structure. We built a machine that thinks. And the machine reintroduced mystery.
Alain de Lille believed that nature operated according to a cosmic grammar, rules so deep and pervasive that they preceded human understanding of them. The grammar was there before we arrived. We merely learned to read it.
Modern AI researchers are now confronting a strikingly similar situation. The patterns these systems discover in language, in images, in data were not placed there by engineers. They were already present in the structure of the information itself. The models do not invent relationships. They surface ones that existed before anyone thought to look.
Which raises a question that Alain de Lille would have understood perfectly, and that Silicon Valley has barely begun to ask:
What if the grammar was never ours to write?
———
Art History Meets AI is a newsletter by Tope Osho, exploring the deep roots
of intelligence, creativity, and the technologies reshaping both.
www.pixelgallery.co.uk
