Losing touch: A comparative phenomenology of creative process in traditional and neural painting (Art in the age of artificial intelligence)

This thesis explores the existential entanglement of autographic and algorithmic media, in an arts-based research process I call neural painting. This process engages in the intentional perturbation of the phenomenal appearance of embodied tacit anticipation in an aesthetic practice engaged with machine-learning creativity support technology to expose and examine the opaque mediation of the apparatus of artificial intelligence (AI). I do this through an art-as-research heuristic post-phenomenology of the artist-technic relation by setting up and participating in an agonistic exchange between the pragmatic actions and metaphors of traditional painting and their mediation by artificially intelligent image synthesis. I offer a phenomenology of the self-other intentional relation in neural media, exploring the augmentation of the creative process through what is here called computational subjectivity, an epistemological intervention upon the machine metaphor of cognition that explores the perspectival affordance of the latent space of neural media. I ask how those situated affordances, implicit in the apparatus of the AI black box, explicit in the AI artefact, enhance or disrupt creative intentionality through its algorithmic transformation. This approach brings into focus how technological advance augments the tacit, and asks what then happens to embodied human expression: Do artificially intelligent technologies "support" or "replace" human creativity? I show that the illusion of the implicit-explicit polarity is driven by the metaphorical entanglement of the tacit and the expressed, driving a losing of touch where the artist is pushed away from direct interaction and embodiment of the artefact and where the conjoined media of tool-as-artefact moves toward a virtualized other where the tool is the artefact and the artefact the tool, conceived of as the embodiment of intelligence. I hypothesize that the opaque exchange of autographic and algorithmic representation in artificially intelligent media ecology promotes a displacement of embodied subjectivity. I explore the implications of this position to contribute to an affective framework for qualitatively explainable AI (XAI), proposing a reflective intervention that builds scholastic and creative exchange where perhaps none were previously possible, because the question of the intersubjective impulse has not been framed in terms of problematizing assumptions about the aesthetics of virtual embodiment.