Immagini schizofreniche. Riflessioni fenomenologiche sulla creatività computazionale a partire da Marc Richir

Abstract

The aim of this essay is to articulate a comparative analysis between the phenomenological status of schizophrenic creative productions and the generative models of digital images by means of new artificial neural networks and stable diffusion techniques. Using the theoretical framework of Belgian phenomenologist Marc Richir developed in the field of phenomenological psychopathology, this proposal seeks to elaborate a possible reinterpretation of the relations between human being and machine from an investigation of the body, spatiality and geometry, elements that a precise contemporary critique – both philosophical and mathematical – currently denounces as victims of an inherited process of unbundling and flattening. Where this critique is based on the assumption that humans and machines are fundamentally incomparable, this proposal intends to analyze the actual schizophrenia at the bottom of current computational machines – which this critical approach clearly illuminates – as a paradoxical juncture between human and machine within the theoretical space of psychopathologies. Following this framework, the contribution intends to analyze not only the reasons why the digital operations of generative diffusion models can be considered creative but also the relationships between computational creativity and schizophrenic creativity.