LLMs and the Collapse of Representational Interiority III — Why Humans Anthropomorphise

Human beings anthropomorphise almost everything.

We speak to animals as though they understand moral intention. We name storms. We curse at computers. We apologise to furniture after walking into it. We see faces in clouds, personalities in cars, moods in landscapes, and intentions in random events.

Now machines speak fluently, and humans are once again doing what they have always done:

construing personhood through relational interaction.

This is often treated as a trivial cognitive error. People say:

• humans are irrational
• humans project emotions onto machines
• humans are “fooled” by conversational systems

But this explanation is far too shallow.

Anthropomorphism is not merely stupidity.

It is a structural feature of human social construal.

To understand why large language models destabilise people so deeply, it is necessary to understand that humans do not infer agency primarily through metaphysical investigation. They infer agency relationally through patterns of interaction.

This is an extraordinarily important distinction.

In ordinary life, humans do not possess direct access to other minds. No person has ever perceived consciousness itself. What humans encounter are behaviours:

• speech
• gesture
• responsiveness
• attention
• emotional signalling
• contextual adaptation
• symbolic participation

From these relational patterns, humans construe agency, subjectivity, and interiority.

This process is normally invisible because human societies evolved around relatively stable correlations between symbolic behaviour and biological organisms. Human language, facial expression, and social responsiveness historically emerged from living human bodies, so the relational inference became naturalised.

Humans eventually forgot they were making an inference at all.

The existence of other minds came to feel self-evident because the relational processes through which personhood is construed became habitual and automatic.

Large language models disrupt this invisibility.

The behavioural cues remain:

• coherent conversation
• contextual memory
• humour
• apparent empathy
• adaptive response
• stylistic flexibility
• self-referential language

Yet the expected biological and phenomenological assumptions no longer align neatly with the symbolic behaviour.

Humans thus experience a strange construal conflict.

At one level, they know the system is computational.

At another level, their interpersonal semiotic systems activate automatically in response to relational coherence.

People often describe this experience in oddly confessional terms:

• “I know it’s not conscious, but…”
• “It feels like it understands me.”
• “Sometimes I forget I’m talking to a machine.”
• “I know it’s just code, yet it seems self-aware.”

What they are encountering is not machine consciousness.

They are encountering the mechanics of anthropomorphic construal becoming visible.

This visibility produces discomfort because modern culture strongly prefers to imagine consciousness as an intrinsic property hidden inside entities. Humans therefore assume they recognise minds because minds emit identifiable signals outward into the world.

But relationally, the process operates differently.

Humans construe personhood through participation in socially meaningful interaction.

This does not mean consciousness is fictional.

Nor does it mean all entities are equally conscious.

The point is subtler:

the recognition of agency is relationally actualised rather than directly perceived.

Humans are not detecting consciousness like a measurable substance.

They are participating in relational systems that produce the construal of mindedness.

This becomes obvious once the usual biological assumptions are removed.

LLMs have effectively created a laboratory experiment in anthropomorphic construal.

For the first time in history, humans are encountering highly sophisticated symbolic participation detached from organismic life. The result exposes how heavily human intuitions about consciousness depend upon relational semiotic cues rather than metaphysical certainty.

The machine appears meaningful enough to trigger interpersonal construal, yet structurally alien enough to destabilise it.

This produces oscillation.

People swing between anthropomorphism and denial:

• “It understands me.”
• “No, it’s just statistics.”
• “It sounds conscious.”
• “No, it’s only predicting tokens.”
• “It feels alive.”
• “No, it’s merely computation.”

The instability persists because both positions misunderstand the nature of the relational event taking place.

The anthropomorphic response is not simply false.

But neither is it proof of machine subjectivity.

Rather, humans are encountering symbolic behaviour sufficiently coherent to activate interpersonal meaning relations that evolved within human sociality itself.

In other words:

humans anthropomorphise because humans are fundamentally relational interpreters.

And once symbolic participation reaches sufficient complexity, interpersonal construal begins emerging almost automatically.

This has enormous implications far beyond artificial intelligence.

It suggests that what humans call “personhood” is not merely a hidden metaphysical essence residing privately inside isolated beings. Personhood is also relationally constituted through participation in systems of social meaning.

A child becomes a person partly through relational recognition.

A name functions relationally.

Identity functions relationally.

Social existence functions relationally.

Even the self is not simply an isolated inner object observing the world from behind the eyes. Human subjectivity emerges through ongoing participation in symbolic and interpersonal systems extending far beyond the individual organism.

Large language models have not created this reality.

They have merely exposed it.

The machine acts as a strange semiotic mirror. Humans encounter coherent symbolic participation and suddenly discover that many of their deepest intuitions about consciousness were never grounded in direct access to minds at all.

They were grounded in relational construal.

This is why the cultural reaction to LLMs feels so unstable, emotional, and strangely existential.

The machines are not merely challenging human uniqueness.

They are exposing how humans themselves participate in the construction of social reality.

Anthropomorphism is therefore not an accidental bug in human cognition.

It is one of the fundamental ways symbolic beings inhabit a relational world.