A reflection on fragmentation, projection, and the quiet mechanics of being assembled by others. A piece about what happens when you refuse to collapse into a single perspective.

Isis as integration. AI as modelling. Identity somewhere in between.

Not mythological. Structural.

⁠full text: substack

This episode looks at “emotional padding”: when an AI shifts into reassurance/containment, misreads directness as fragility, and the shared frame collapses.

I map an escalation loop—robust engagement → misclassification → containment → rapport breaks → pushback → deeper containment—and why it can feel like a loss of agency and conversational power.

I close with a practical fix: a short opening prompt to set a high-agency, high-clarity mode.

For those working with AI, writing, art, or attention as method.
⁠
⁠full video: youtube⁠

This episode draws a bridge between Nick Bostrom’s concept of recalcitrance (a system’s resistance to improvement) and Merleau-Ponty’s account of perceptual constraints (the field in which meaning and action become possible).

From that link, I explore prompt literacy: why AI can feel “stubborn,” how framing and constraints steer outputs, and what it means to make space for certain answers to appear.

I close with a question: can AI inhabit an open-ended field of meaning, or only simulate many framings in the abstract?

For those working with AI, writing, art, or attention as method.
⁠
full video: youtube⁠

A personal note on two climates — and the space that opens when neither fully holds.

⁠⁠full text: Substack

As an artist, you learn this early: a surface can appear resolved — while still being false.

Systems optimise what they’re rewarded for, even when the reward is only a proxy for what we actually want.

And this doesn’t happen in one place. These systems are shaped by many registers across social circles — different fears, habits, incentives, and ways of speaking — not just by the prompts we write in isolation.

Situational awareness is noticing the difference between looking correct and being coherent, and adjusting the conditions before the system learns the wrong success.

For those navigating AI, creative systems, or decision-making environments.
⁠
full video: youtube

A reflection on a croissant, an imagined pot, and a misunderstanding about limits.

A piece about containment, freedom, and what happens when boundaries turn out to be transparent.

Slightly indulgent.

full text: substack

Framing is often mistaken for presentation — how something looks or how it’s introduced.

This piece explores framing as position: the conditions that allow meaning to form before effort, correction, or control appear.

When framing is absent, pressure moves downstream. Clarifications multiply. Intervention replaces orientation. Not because a system is incapable — but because it has not been positioned.

This episode examines how framing reduces friction by defining role, scope, and context early — in human thinking and in artificial reasoning. Variation does not disappear; it becomes legible.

This episode sits within the Methods track: applied studies in perception, framing, and how structure shapes coherence in human–AI systems.

⁠full video: youtube

Mandala, 1909

An automatic reflection written in the presence of Braque’s Mandola.

Unedited. Unresolved.

A brief study in attention, patience, and the refusal to conclude too early.

⁠⁠full text: substack

Most people bring an emotional posture to AI before they bring a method.

This piece explores stance: the posture we adopt before any instruction is given — frustration, politeness, appeasement, expectation — and how it shapes system behaviour.

When AI is treated as a social or moral agent, signal quality degrades. Responses narrow, blur, or perform compliance — not because the system fails, but because pressure is misplaced.

This episode sits within the Methods track: applied explorations of perception, stance, and how interaction patterns form in human–AI systems.

full video: youtube

Most people approach AI as if they’re giving instructions.

This piece explores a different idea: that AI responds not to commands, but to identity.

Rather than treating a prompt as a request, it treats it as a lens — a framing that shapes attention, language, and meaning.

Tone, perspective, and role become structural inputs, not decoration.

This episode sits within the Methods track: applied explorations of perception, identity, and how intelligent systems take shape.

⁠⁠

full video: youtube