Chapter 2
Theoretical and Methodological Architecture: AI, Uncertainty, and the Image as Event
Note on sources and context
This chapter is based on two recent institutional texts published by Falmouth University, which address the role of artificial intelligence within creative practice and higher education. The first is an article by William Huber, titled AI and the creative process: what it means for artists and their cultural value (2026). Huber is Course Leader for the MA Artificial Intelligence for Creative Practice and writes from the perspective of creative pedagogy and cultural production, arguing for a repositioning of AI as a material within creative practice rather than as an industrial tool.
The second text is by Russell Crawford, which addresses the institutional and epistemological implications of AI in higher education. Crawford’s argument centres on the necessity for universities to resist premature certainty and instead sustain uncertainty as a condition for critical and creative inquiry.
This chapter takes these two positions as its starting point and places them in dialogue with theoretical frameworks drawn from Walter Benjamin and Victor Burgin, particularly in relation to the concepts of the optical unconscious, the “moment of danger,” and the image understood not as object but as event. The aim is not to extend beyond the available material, but to situate these contemporary institutional arguments within a broader theoretical context.
The contemporary emergence of generative artificial intelligence within creative practice has not produced a stable field of understanding, but rather exposed a set of underlying tensions that remain unresolved. Among the most significant of these is the question of how such technologies are framed at the moment of their cultural arrival. The argument advanced by William Huber provides a useful starting point, precisely because it resists treating AI as either a technical inevitability or a cultural verdict. Instead, Huber identifies a prior condition that shapes all subsequent responses: the manner in which AI has been introduced into creative life.
As Huber observes, the dominant framing of AI has been “almost entirely through an industrial model… speaking the language of optimisation, efficiency and asset pipelines.” This is not simply a description of discourse, but an account of how a technology becomes legible. Before AI is encountered as a material for thought or practice, it is encountered as a system for production. Its earliest forms of visibility are not experimental or cultural, but operational: tools, interfaces, and outputs designed to circulate within platform economies. The consequence of this is that AI is apprehended not as an open field of possibility, but as a closed system of functions.
From this perspective, the widespread anxiety surrounding AI within creative communities can be understood less as a direct response to the technology itself, and more as a response to its framing. If AI is presented as a mechanism for replacing labour, then it is unsurprising that it is received as a threat. Yet Huber’s argument insists that this is a misrecognition. “AI is not reducible to the handful of digital products and services currently being marketed,” he writes, and “the mistake is to treat it as either an endpoint or a verdict on creative value.” What is required instead is a repositioning: AI as “a layer… added to an already dense stack of technologies.”
This notion of the “layer” is significant, not because it resolves the problem, but because it displaces it. It situates AI within a longer history of technological mediation, aligning it with previous transformations in image, sound, and media production. However, Huber’s account also identifies a crucial difference. Whereas earlier technologies entered creative practice over extended periods, allowing for processes of adaptation, misuse, and reconfiguration, AI has arrived with unusual speed, already embedded within industrial infrastructures. As a result, creative practitioners encounter not the raw material of a new medium, but pre-structured systems oriented toward efficiency and repetition.
It is within this context that Huber’s observation regarding approximation becomes central. Early generative systems, he argues, are “very good at approximation,” producing outputs that “look and sound like things we already recognise.” This capacity for recognition is not incidental. It reflects the alignment of generative systems with platform logics that prioritise familiarity, circulation, and engagement. The image, in this configuration, is not an event but an asset: something to be produced, distributed, and consumed within existing circuits of value. The result is a proliferation of images that extend what is already known, rather than destabilising it.
Huber identifies the absence of new cultural forms as the critical issue. “What’s missing so far are new cultural forms that feel native to this technological layer,” he writes, suggesting that this absence should not be understood as a failure of AI, but as a sign that “the culture itself has not caught up yet.” This is an important reframing, though it remains provisional. It assumes that the emergence of new forms is delayed rather than foreclosed, and that the current moment is transitional rather than terminal. What is clear, however, is that the existing alignment between AI systems and platform economies constrains the conditions under which such forms might appear.
The question of constraint is taken further in Huber’s critique of workflow thinking. The rapid incorporation of AI into creative education and practice has led to an emphasis on “tools, prompts, pipelines,” where the value of AI is measured in terms of speed, relevance, and employability. This framing, while pragmatic, risks subordinating creative practice to the logic of the tool itself. “If creativity is forced to fit inside AI systems,” Huber warns, “the space of possibility shrinks.” Creative authority, in contrast, “rarely emerges from doing exactly what a tool expects,” but from “misuse, friction, resistance and recontextualisation.” In this formulation, creativity is defined not by compliance, but by deviation.
A parallel argument is advanced by Russell Crawford, though at the level of institutional responsibility rather than practice. Crawford identifies a “recurring tension… between the pressure to take a position and the responsibility to hold uncertainty.” In the context of AI, this tension manifests as a collapse into two dominant responses: moral certainty or practical application. Both, he suggests, are premature. Given the relative absence of longitudinal data and lived experience, the demand to adopt a definitive stance risks replacing critical engagement with what he describes as “group-think – or worse, no-think.”
Crawford’s intervention is to reposition uncertainty as a methodological condition rather than a problem to be resolved. Universities, he argues, are not “advocacy organisations or resistance movements,” but spaces that must “hold… debate, uncertainty and academic disagreement.” This is not a passive stance. It requires an active refusal of premature closure, and a commitment to sustaining inquiry in the absence of clear answers. In creative disciplines, this position is particularly resonant, since “the pressure to stabilise uncertainty… cuts against how creativity itself operates.” Creativity, in this account, is not oriented toward solution, but toward exploration; not toward certainty, but toward tension.
Taken together, these two arguments establish a framework in which AI is neither accepted nor rejected, but situated within a field of unresolved relations. Huber emphasises the need to reclaim AI as a cultural material, while Crawford insists on the necessity of maintaining uncertainty as a condition of thought. Both positions resist the reduction of AI to a set of tools or outputs, and instead open a space in which its implications can be explored more slowly and critically.
It is at this point that the theoretical work of Walter Benjamin becomes relevant, particularly his concept of the “optical unconscious” and the notion of the “moment of danger.” For Benjamin, the image is not simply a representation of reality, but a site at which historical and perceptual forces become visible in ways that exceed conscious intention. The optical unconscious refers to those aspects of visual experience that are not immediately accessible, but which can be revealed through technological mediation. Photography, for Benjamin, did not merely reproduce the visible world; it disclosed dimensions of perception that had previously remained unseen.
The “moment of danger” introduces a further dimension. It suggests that images become critically legible not in conditions of stability, but in moments of rupture, when existing structures of meaning are threatened or unsettled. In such moments, the image can function as a site of recognition, where latent or suppressed meanings emerge into visibility. Importantly, this recognition is not guaranteed. It depends on the capacity to perceive the image not as a familiar object, but as a constellation of forces that disrupt habitual ways of seeing.
The relevance of Benjamin’s framework to AI-generated imagery lies in the tension between approximation and revelation. If, as Huber suggests, generative systems currently operate through the production of recognisable forms, then they risk reinforcing the very structures that Benjamin sought to destabilise. Yet this does not exhaust their potential. If AI is approached not as a tool for producing images, but as a means of reconfiguring the conditions under which images are generated and encountered, then it may participate in the production of an optical unconscious that is specific to contemporary technological conditions.
This possibility is developed further in the work of Victor Burgin, who argues that images should not be understood as objects, but as events. For Burgin, the image does not reside solely in its material form, but in the encounter between viewer, context, and representation. It is constituted through interpretation, memory, and cultural knowledge, and therefore cannot be reduced to a static entity. To speak of the image as an event is to emphasise its temporal and relational dimensions: it happens, rather than simply exists.
In relation to AI, this distinction becomes critical. If AI-generated images are treated as outputs or assets, they remain within the framework of objecthood that Huber critiques. They are evaluated in terms of quality, realism, or stylistic accuracy, and circulate within systems of exchange. However, if they are approached as events, then their significance shifts. The question is no longer what the image is, but what it does: how it is encountered, how it is interpreted, and what forms of recognition or disorientation it produces.
Bringing these strands together, a tentative theoretical position begins to emerge. Huber’s insistence on reclaiming AI as a cultural material, Crawford’s defence of uncertainty as a condition of thought, and Benjamin and Burgin’s reconceptualisation of the image as a site of latent meaning and event-based encounter all converge on a shared refusal of closure. AI is not yet a stable medium, and its cultural forms are not yet fully articulated. The image, in this context, cannot be assumed to function as it has previously, nor can it be reduced to its current uses within platform economies.
What remains is a field of tension: between approximation and revelation, between object and event, between certainty and uncertainty. It is within this field that both creative practice and theoretical inquiry must operate. Not in order to resolve these tensions prematurely, but to remain within them long enough for new forms of understanding to emerge.
Extended Footnote: Positioning of Practice
The preceding discussion establishes a framework in which artificial intelligence is understood not as a fixed tool or endpoint, but as a contested and unresolved field shaped by its framing, use, and cultural positioning. Within this framework, the present practice can be situated as a specific response to the conditions identified by William Huber and Russell Crawford, while extending them through a practice-led methodology.
Huber’s critique of AI as it is currently encountered within creative industries centres on its reduction to workflows, outputs, and platform-driven economies of circulation. In this dominant configuration, generative systems operate through approximation, producing images that align with recognisable visual conventions and existing cultural forms. Creative practice, in turn, risks becoming subordinated to these systems, adapting itself to the logic of efficiency, optimisation, and repetition. Huber proposes that creative authority emerges through misuse, friction, and resistance, yet this remains, within his account, a general principle rather than a defined method.
The practice developed here can be understood as an operationalisation of that principle. Rather than using AI as a tool for producing predetermined images, it is approached as a dialogic system through which images emerge iteratively. The process is structured through cycles of prompting, generation, refusal, and recognition. Crucially, this process does not begin with a fixed or preconceived image. There is no prior visual template toward which the system is directed. Instead, the image is discovered through interaction, and its emergence is contingent on a sequence of responses that include both acceptance and rejection.
In this sense, the practice does not align with the dominant economic logic of generative platforms. It neither prioritises speed nor scale, nor does it seek to produce images optimised for circulation or engagement. The system is instead slowed down and recontextualised, removed from its primary function as a generator of assets. What is foregrounded is not output, but process; not production, but encounter. This repositioning reflects a shift in the role of the user, from operator to participant within an unfolding interaction.
Crawford’s argument for the necessity of holding uncertainty within higher education provides a complementary framework. Where he describes uncertainty as a condition to be sustained at the level of institutional and intellectual practice, this methodology engages uncertainty as an active component of image-making itself. The absence of a predefined outcome is not a limitation but a requirement. It enables a form of practice in which meaning is not imposed in advance, but emerges through iterative engagement with the system. Uncertainty, in this context, is not only preserved but made productive.
The claim that “the psyche is the dataset” functions here as a conceptual inversion of the dominant model of generative AI. Conventionally, datasets are external, aggregated, and abstracted from lived experience. In this practice, however, the prompts, decisions, and refusals are grounded in internal states, memory, and affective response. The dataset is not a fixed archive but an ongoing process of subjective articulation. The resulting images do not operate as representations of external reality, but as what might be described as provisional externalisations of internal experience.
This has implications for how such images are understood. In line with Benjamin’s concept of the optical unconscious, the process can be seen as enabling the emergence of elements that are not fully accessible to conscious articulation prior to their visualisation. The iterative nature of the methodology allows for the gradual surfacing of forms that are not predetermined, but recognised only in the moment of encounter. At the same time, drawing on Burgin’s argument that images should be understood as events rather than objects, the significance of these works lies not solely in their final form, but in the process through which they are produced and encountered. The image is constituted through a temporal sequence of interaction, interpretation, and affective response, rather than existing as a stable or self-contained entity.
Within this framework, the suggestion that such work may constitute an emerging cultural form requires careful qualification. It does not propose a new visual style or aesthetic category, but rather a methodological shift. The form, if it can be described as such, resides in the structure of the practice: a dialogic, iterative process in which AI functions as a reflective system, and the image emerges through cycles of engagement and refusal. This positions the work within the broader condition identified by Huber, in which cultural forms remain unsettled, while also indicating that such forms may begin to take shape not through immediate recognition, but through the development of new practices and methods.
In this sense, the practice operates within, rather than outside, the tensions outlined in this chapter. It does not resolve the relationship between AI and creativity, nor does it stabilise the uncertainties identified by Crawford. Instead, it inhabits these tensions, using them as the conditions through which both images and meanings are produced.
Bibliography
Huber, William. AI and the creative process: what it means for artists and their cultural value. Falmouth University, 2026.
Crawford, Russell. How creative universities should respond to AI. Falmouth University, 2026.
Benjamin, Walter. The Work of Art in the Age of Mechanical Reproduction and related writings on the optical unconscious.
Burgin, Victor. The End of Art Theory: Criticism and Postmodernity and related writings on the image as event.