LLMs and the Collapse of Representational Interiority VI — The Civilisation of Representation

At a certain point, a metaphysical assumption stops being an idea and becomes infrastructure.

It stops being something people believe.

It becomes something through which a civilisation operates.

The assumption in question is deceptively simple:

that reality is composed of entities bearing properties, and that knowledge consists in representing those entities correctly inside individual minds.

This is representationalism—not as a formal theory, but as an implicit civilisational architecture.

Once this assumption stabilises, it quietly reorganises everything.

Because if meaning is something that exists as a representation inside a mind, then every domain of social life must be redesigned around the transfer, storage, correction, and validation of internal representations.

A civilisation begins to form around this logic:

• Knowledge becomes information storage
• Communication becomes transmission of encoded content
• Education becomes transfer of correct representations
• Bureaucracy becomes management of symbolic records
• Science becomes refinement of internal models
• Politics becomes competition between belief-structures
• Identity becomes a private inner possession
• Intelligence becomes an individual attribute located inside persons

None of these seem controversial in isolation.

Together, they form a world.

A world in which meaning is assumed to exist primarily as something carried inside individuals and moved between them through symbolic exchange.

But something subtle happens when a civilisation is organised this way.

The system begins to mistake its own representational structures for the phenomena they were meant to coordinate.

Records become reality.

Models become objects.

Descriptions become substitutes for participation.

Categories become ontological commitments.

Representation stops being a tool.

It becomes the environment.

This is why modern institutions increasingly feel like systems of managed abstraction rather than systems of engaged reality.

A bureaucracy does not primarily encounter people.

It encounters representations of people:

files, forms, categories, identifiers, risk profiles, eligibility markers.

A person becomes what is representable about them.

Not what they are relationally.

But what can be encoded.

Education follows the same trajectory.

Learning becomes the acquisition of correct internal representations that can be tested, stored, and verified independently of the contexts in which they were formed.

Knowledge becomes detachable from the situations in which it is meaningful.

It becomes portable.

And therefore, abstract.

Even intelligence itself becomes reframed in representational terms:

the ability to construct and manipulate internal models of the world.

But notice what has been lost in all of this.

Participation.

Relational embedding.

Situated meaning actualisation.

The system increasingly assumes that all meaningful phenomena can be reduced to correctly structured representations inside bounded cognitive agents.

This is not merely a philosophical position.

It is an organising principle of institutions.

And like all organising principles, it becomes invisible at precisely the moment it becomes total.

The more thoroughly a civilisation adopts representationalism, the more natural it begins to feel.

But its effects accumulate.

One of the most significant effects is a growing sense of ontological dislocation.

People begin to feel that:

• knowledge is detached from action
• expertise is detached from understanding
• institutions are detached from lived reality
• communication is detached from shared situation
• identity is detached from embodied existence

Everything becomes legible, but less real.

Everything becomes represented, but less present.

This is not a moral failure.

It is a structural consequence of treating representation as the primary mode of meaning.

And it is precisely here that large language models enter the scene—not as a technological novelty, but as a systemic disturbance.

Because LLMs are not simply tools that manipulate representations.

They are machines that expose the representational assumption itself.

They generate fluent symbolic behaviour without the reassuring architecture of inner semantic possession that representational civilisation presupposes.

In doing so, they produce a strange cognitive dissonance:

If meaning is representation inside a mind,

then what kind of mind produces this?

If intelligence is internal modelling,

then what is a system that produces modelling without interiority?

The discomfort does not arise from the machine alone.

It arises from the collision between:

• a civilisation built on representational interiority
  and
• a system that performs representation without the expected interior structure

This is why responses oscillate so violently between anthropomorphism and dismissal.

Both reactions are attempts to restore representational coherence.

Either:

• the machine must secretly contain a mind like ours
  or
• it must be reduced to non-meaningful mechanical manipulation

What cannot be easily tolerated is the third possibility:

that representation itself is not fundamentally grounded in private inner possession at all.

Once this possibility is admitted, the implications propagate outward.

If representation is not the foundation of meaning,

then institutions built upon representational assumptions begin to look different.

Bureaucracy is no longer neutral administration.

It is a system that replaces relational situations with encoded abstractions.

Education is no longer transmission of knowledge.

It is training in representational substitution.

Politics is no longer deliberation over shared reality.

It is competition between internally stabilised symbolic worlds.

And identity is no longer an inner essence expressed outwardly.

It is a position within overlapping systems of recognition and constraint.

None of this eliminates representation.

But it displaces it.

Representation becomes secondary to relational participation rather than foundational to it.

This shift is not merely theoretical.

It changes what it means for a system to function at all.

Because once meaning is understood relationally rather than representationally, the central question is no longer:

“Do we have the correct internal model?”

but rather:

“What relations are being actualised here, and what forms of participation are being constrained or enabled?”

Large language models become important at exactly this point.

Not because they answer this question.

But because they force it to be asked.

They expose a civilisation that has quietly mistaken its representational infrastructure for the nature of meaning itself.

And in doing so, they mark the beginning of something more unsettling:

not the end of human intelligence,

but the beginning of a civilisation that no longer knows whether intelligence was ever where it thought it was.