The Demanding Image: AI, Personalised Aesthetics, and the Currency of Spectatorship.

How can we approach the cinematic image becoming through personalised screens, in-between porous intermedial relations, and a data-fied expression of cinematic time? And how can we conceptualise the spectator becoming in relation to that image amidst data-intensive personalisation processes? Interdisciplinary in nature, this thesis draws on film theory and philosophy of film, politics of aesthetics, critical post-humanities and critical AI studies, to examine the material relations and effects created on the ubiquitous screens of YouTube and Netflix. I contend that AI-driven curation and exhibition processes introduce a spectator-centric aesthetico-political milieu. This milieu establishes itself as an authoritative canon, by virtue of a personalised loop of spectatorial engagement, and gives rise to a new type of image, that I call the demanding image. Continuously re-figured, it reflects on its own materiality, aesthetics, and a quantification of the affective and meaning-making connections it develops with spectators. Expressed in different states of material reflexivity on each screen, it draws on aesthetic qualities associated with the cinematic image and produces specific effects, that I identify as personalised aesthetics. Intrinsic in the formation of the demanding image and the production of aesthetic effects is a concept of embedded value judgements and a re-materialisation of its immaterial matter around data. I propose the concept currency of spectatorship to better navigate and articulate the role of the spectator in this context. I examine the latter against AI-led operationalisations of the data flows that connect the image, the spectator and her contextual parameters during image recognition and recommendation processes and identify a duality that applies equally to a quantified notion of cinematic time, temporality and duration, as well as principles of neoliberal financialisation.