Humour and AI: 3 Is AI Humour Simulation or Co-Actualisation?: Where Does the Joke Actually Occur?

We have distinguished statistical continuation from structured potential.

We have questioned whether distribution equals system.

Now we turn to the decisive issue:

When a machine produces a joke and a human laughs, what has happened?

Is this simulation?

Or is it genuine co-actualisation?

---

The Temptation of the Binary

The debate is often framed too crudely.

Either:

• The machine “understands” and therefore genuinely produces humour.

Or:

• The machine does not understand and therefore merely simulates humour.

But this binary assumes that humour must reside inside one participant.

Relational ontology dissolves that assumption immediately.

Meaning does not pre-exist construal.

It emerges in the cut.

So the question shifts:

Where, within the relational field, does the cut occur?

---

Simulation: Pattern Without Horizon

On a simulation account, the machine:

• Reproduces structural features associated with humour.

• Generates pattern continuations statistically correlated with jokes.

• Outputs text that resembles prior humorous texts.

The humour, if any, is supplied by the human participant.

The machine contributes structured pattern.

The human performs the perspectival reorganisation.

Under this view, AI humour is artefactual. It is scaffolding for human construal.

The machine does not inhabit a horizon of potential. It generates outputs derived from weighted recurrence. Any laughter occurs in the human system alone.

This position is coherent.

But it may be incomplete.

---

Co-Actualisation: A Relational Event

Consider the alternative.

Humour is not located in a mind. It is an event of alignment. It occurs when:

• An output reorganises expectation.

• A participant recognises the reorganisation.

• The field of interaction shifts.

If so, then humour is neither “in” the machine nor “in” the human alone.

It is in the interaction.

When a machine generates a low-probability continuation that invites reinterpretation, and a human performs that reinterpretation, something has happened that neither system could complete independently.

The machine supplied structured constraint.

The human supplied perspectival cut.

Together, an event occurred.

This is co-actualisation.

Not because the machine experiences humour.

But because the event required both participants.

---

Does Co-Actualisation Require Symmetry?

An objection arises: co-actualisation implies reciprocity. Machines do not experience alignment, tension, or release.

True.

But relational events do not require symmetry of internal architecture. They require interaction across structured systems.

When a book reorganises a reader’s construal, we do not attribute experience to the book. Yet we acknowledge that meaning emerges in the encounter.

The text is not conscious.

But it participates in the event.

The machine’s output may function similarly.

The humour does not reside in silicon.

It resides in relational alignment across heterogeneous systems.

---

The Asymmetry Problem

However, we must not overstate the case.

A human comedian can:

• Adjust timing mid-performance.

• Intentionally misalign.

• Exploit audience feedback.

• Navigate social and ethical nuance dynamically.

Current machine systems cannot inhabit that reflexive field. They optimise outputs given inputs; they do not experience the shifting horizon.

This asymmetry matters.

Co-actualisation may occur at the level of event, but the capacities contributing to that event are radically different in kind.

One operates through structured potential actualised perspectivally.

The other operates through distributional optimisation.

The event may be shared.

The architectures are not.

---

A More Precise Formulation

So perhaps the correct formulation is neither “simulation” nor “understanding,” but this:

AI humour is a distributed event of constrained co-actualisation.

The machine:

• Generates pattern structures derived from historical residue.

The human:

• Performs perspectival reorganisation within a systemic horizon.

Humour emerges in the alignment between these operations.

The machine does not need to “get the joke.”

It needs only to generate constraints capable of inviting a cut.

---

The Diagnostic Edge Case

Consider when AI humour fails socially.

When an output is tone-deaf, ethically misaligned, or culturally inappropriate, the failure reveals the limits of distributional modelling.

The machine has optimised within data.

But it has not navigated a live relational horizon.

The human must repair the misalignment.

This exposes the asymmetry clearly: co-actualisation depends on one participant capable of horizon-shifting construal.

---

Provisional Position

AI humour is not pure simulation. Something genuinely relational can occur.

But neither is it symmetric co-creation.

It is an event in which:

• One system generates patterned constraints.

• Another system actualises relational cuts.

• Meaning emerges in the alignment.

The humour is real.

Its architecture is asymmetrical.

And that asymmetry is philosophically decisive.