Fairy Note: Artistic Becoming, Embodied Recognition and Thinking with AI Images
Aleph, 5 October 2026
Working concepts
Artistic becoming
Technological intentionality
The Navigator
Activity and passivity
Embodied and affective recognition
Refusal
Resonance and recognition
Image thinking
Psychic snapshot
1. Four new papers
Four papers published in September and October 2026 have become unexpectedly relevant to Phantom Mirror. They approach generative AI from different directions but converge around a question increasingly central to my research: what happens to the human practitioner through the encounter with a generative system?
The most important is Amit Raphael Zoran’s “More than we can make: generative AI, artistic forms, and artistic becoming” (AI & Society, 3 October 2026). Zoran distinguishes between artistic form, the artefact produced, and artistic becoming, the developmental and transformative process historically involved in learning and making art. His “decoupling thesis” argues that generative AI can produce sophisticated artistic forms without requiring the human practitioner to undergo the developmental process previously necessary to make them.
Yet becoming does not simply disappear. It can relocate into prompting, iteration, curation, interpretation, contextualisation and sustained interaction with the system.
This is immediately important for Phantom Mirror. The generated image is not the whole artistic event. The practice lies substantially in what happens through the encounter with the image.
Three other recent papers complicate this further.
Alessio Martino and Luca M. Possati’s “Technological Intentionality and the Concept of Flesh: A Merleau-Pontian Perspective on Technological Mediation” (Philosophy & Technology, 15 September 2026) examines generative systems through technological intentionality, including the shifting relationship between activity and passivity in human-machine interaction.
Tian Zhao, Xiaoyu Wu and Yansong Chen’s “Flux in the Folds: The Cross-Media Logic and Embodied Turn of Generative AI Imagery” (Intelligent Visuals and Communication, 29 September 2026) moves towards affect and embodied reception, describing generative imagery in relation to the “affective machine”.
Yuxi Si’s “From ‘Perfect Algorithm’ to ‘Resonance Ethics’” (Intelligent Visuals and Communication, 29 September 2026) develops the idea of “resonance poverty”, asking what may be lost when computational perfection displaces duration, embodied presence, material trace and relational experience.
Taken together, these papers shift the question away from simply what AI can make towards what happens between human, system and image.
2. Artistic form and artistic becoming
Zoran’s distinction is particularly useful because it separates two things that are frequently collapsed in discussion of AI art.
A generative system can produce an accomplished form without the human having acquired the embodied skills traditionally necessary to produce it. This is the decoupling.
But this does not establish that artistic practice has disappeared. It asks us to locate artistic development elsewhere.
In Phantom Mirror, becoming is not primarily located in acquiring the technical skill to construct the image. It occurs through the extended process surrounding its generation: dialogue, iteration, encounter, refusal, recognition and reflection.
This is very close to something already implicit in my methodology. I do not begin with a completely articulated thought and then use AI to illustrate it. The thought itself can change through the process of making.
The image may therefore be one of the places in which the thought became possible.
3. What exactly is becoming?
This requires some caution. Artistic becoming should not become another name for everything that happens in Phantom Mirror.
Zoran’s concept is broader than my idea of recognition. Becoming concerns the development or transformation of the practitioner through artistic activity. Recognition is a particular event within my practice.
The relationship may be:
Recognition is one of the moments through which artistic becoming becomes perceptible to the practitioner.
Something has changed through the encounter. Recognition is the moment at which that change becomes available to me.
This gives a useful provisional formulation:
The machine generates a form. The encounter enables becoming. Recognition makes the change perceptible.
This does not mean every generated image produces becoming, nor that every moment of recognition constitutes transformation. It describes a relationship that needs to be investigated through the practice.
4. Technological intentionality
Martino and Possati provide another part of the problem.
Their concept of technological intentionality allows activity to be attributed to technological systems without requiring a claim that the system possesses human consciousness or subjective intention.
This is particularly useful for generative AI because the system regularly produces something beyond the practitioner’s explicit intention. I initiate a process, but I do not determine everything that subsequently appears.
The unexpected image matters.
It can alter the direction of the dialogue, suggest an association I had not consciously formulated or cause me to reconsider what I thought I was looking for.
The system therefore cannot adequately be described as a passive tool. But neither does this require treating it as an autonomous creative subject.
What matters is the relation.
5. Activity and passivity
Martino and Possati’s account of changing activity and passivity is especially close to the experience of working generatively.
Agency moves.
At one moment I initiate. At another I receive. I respond to something generated by the system, then intervene again. The system produces another possibility. I accept, reject or redirect it.
This makes the conventional distinction between active human and passive tool increasingly inadequate.
But it would be equally misleading to reverse the hierarchy and make the machine the active creative agent while the human merely selects.
The encounter is more unstable than either account permits.
This is where the Navigator becomes useful.
6. The Navigator
The language of navigating generative or latent space already exists within AI art and design discourse, so I should not claim the navigation metaphor itself as uniquely mine. My developing concept of the Navigator, however, is more specific.
The Navigator describes the human practitioner who remains oriented within a generative field they do not completely control, using dialogue, perception, refusal, affective judgement and recognition to determine where to go next.
The Navigator is neither sovereign author nor passive curator.
This is why prompt engineer has always felt inadequate. Prompting initiates or redirects a process, but it does not describe the experience of moving through it. Nor does curation adequately describe what happens, because I am not simply selecting retrospectively from a set of completed possibilities. My responses alter what happens next.
The Navigator therefore occupies the shifting field of activity and passivity described by Martino and Possati. The system can produce an excess beyond my intention; I encounter that excess and orient myself in relation to it.
The concept also changes how artistic becoming can be understood. The Navigator is not simply navigating towards an artefact. The practitioner may themselves be altered through the navigation.
The machine generates a form. The encounter enables becoming. Recognition makes the change perceptible. The Navigator is the figure who moves through that encounter.
The fuller concept of the Navigator needs its own Fairy Note. Here it matters because it identifies the human position within artistic becoming without restoring the old model of complete authorial control.
7. Image thinking and the cognitive scaffold
This connects with Toister and Zylinska’s concept of image thinking. Images do not merely illustrate thought after it has occurred. They can participate in cognitive processes.
This has been central to Phantom Mirror for some time. Generative AI operates for me as a cognitive scaffold because external images allow thought to proceed through perceptual encounter rather than through language alone.
The process is therefore not simply:
thought → prompt → image
but something closer to:
experience → dialogue → image → perception → response → further thought
The image feeds back.
This is particularly important because the generated image can contain something I did not knowingly place there. The encounter with it can make another thought possible.
That does not require the claim that the AI itself understands the experience. The cognitive significance can arise within the human-machine-image relation.
8. Affect enters the loop
Zhao, Wu and Chen’s discussion of generative imagery as an affective machine introduces another important dimension.
My process begins and ends with affect.
The phrase I used spontaneously when discussing Dual Unrealities remains useful:
I start with emotion and end with emotion.
But the second emotion is not necessarily identical to the first.
Affect enters the dialogue, becomes entangled with language and computational generation, returns as an image and is encountered again perceptually and emotionally.
This makes Phantom Mirror more recursive than a model in which affect simply passes from artist to artwork to viewer. Affect is continually fed back into the making process.
The image becomes a point at which feeling can itself become perceptible.
9. Refusal
This is why refusal is methodologically important.
Most generated images are rejected.
Refusal is not merely dissatisfaction with an output. It is one of the ways in which the Navigator remains oriented within the process.
This also connects with the literature on aesthetic bias. Generative systems have tendencies towards particular kinds of visual resolution, coherence, spectacle and aesthetic appeal. An image can therefore be highly successful according to the model’s aesthetic tendencies while being completely wrong for the experiential work.
Refusal interrupts that movement towards premature resolution.
It keeps the process open.
This suggests:
Refusal is not the opposite of navigation. Refusal is one of its operations.
10. Resonance and recognition
Si’s concept of resonance poverty is useful because it asks what may be missing from computationally perfected imagery: duration, relationality, embodiment and lived encounter.
But resonance and recognition should not be collapsed.
Resonance suggests that something affects or reverberates with the viewer. Recognition, as I am using it, is more specific. It is the moment at which an image becomes experientially meaningful in relation to something lived but not previously fully articulable.
This is also why recognition differs from preference.
Preference asks which image I like. Recognition asks which image reveals something to me.
The distinction matters because contemporary generative systems are increasingly capable of modelling preference. My practice is not primarily concerned with producing the image I am statistically most likely to prefer.
It is searching for something I may not have known I was looking for until I encounter it.
11. Embodied recognition
Recognition is not purely intellectual.
The judgement “that’s it” can occur before I can explain why the image is right. This is one reason embodiment and affect matter to the methodology.
The image is perceived before it is fully interpreted.
This brings the work towards phenomenology without requiring me to turn Phantom Mirror into a strict phenomenological method. The research concerns the lived experience of encountering generated images and the way meaning can emerge through that encounter.
The important question becomes less:
What does this image represent?
and more:
What happens when I encounter it?
12. The psychic snapshot
The psychic snapshot emerges from this process.
It is not synonymous with the generated image. Most generated images are not psychic snapshots.
The term describes the image that acquires particular significance through recognition within the iterative process.
The separate Fairy Note Zipping the Mind now develops another aspect of this, understanding the psychic snapshot as a form of technologically mediated experiential compression, a ZIP file of emotion. That idea should remain conceptually distinct here.
For the purposes of artistic becoming, the important point is that the psychic snapshot is not simply the final artefact.
It is evidence of an encounter.
13. Form, encounter and becoming
The emerging structure can therefore be provisionally expressed as:
form → encounter → refusal → iteration → recognition → becoming → psychic snapshot
But this is not a rigid sequence. Becoming may be occurring throughout, and recognition may redirect rather than terminate the process.
What this structure does make visible is the inadequacy of treating the AI-generated image as the sole locus of artistic value.
The significant artistic event may lie partly in the changing relationship between practitioner, system and image.
The artefact matters.
But so does what happened in order for that artefact to become meaningful.
14. Where this leaves the research
These new papers help clarify that Phantom Mirror sits within a wider shift in thinking about generative AI.
The question is moving beyond whether AI can produce aesthetically convincing forms. More interesting questions concern agency, mediation, embodiment, affect, interpretation and the changing position of the human practitioner.
My research contributes to this through the particular phenomenon of recognition.
The practice suggests that generative images can become part of an iterative cognitive and affective process through which aspects of lived experience become thinkable that could not previously be fully articulated.
Within that process, the Navigator provides a developing account of human agency without requiring complete human control. Refusal maintains orientation. Recognition identifies significance. The psychic snapshot records something of the encounter.
This gives a clearer relationship between the concepts:
Zoran helps explain becoming.
Martino and Possati help explain distributed activity and technological mediation.
Zhao et al. bring affect and embodiment into the generative image.
Si raises the problem of resonance within computational culture.
Toister and Zylinska provide the larger framework of thinking with images.
The Navigator, refusal, recognition and psychic snapshot emerge from my practice as ways of describing what happens within that field.
Closing thought
The significance of generative AI as an artistic medium may lie not only in the forms it can produce, but in the encounters those forms make possible.
Phantom Mirror began with lived experience and image-making rather than a theory of AI. The emerging literature is now providing language with which to understand why the process has felt different from simply making or selecting images.
The machine does not need to understand my experience for the encounter to change my understanding of it.
The machine generates a form. The encounter enables becoming. Recognition makes the change perceptible. The Navigator moves through the encounter.
Perhaps the larger proposition is therefore:
The significance of generative AI as an artistic medium lies not only in what it can make, but in what can become thinkable through making with it.