GPT Icon Numerical Hallucinations in Computational Art

From Imperative Operativity to Informational Contingency

Abstract

This article addresses how images resulting from massive data analysis are characterized by a kind of imperative operativity that establishes them as action models for their receivers. At the same time, the systems that generate these images are contingent systems that determine whether these images are informative or redundant at the data level. Using the theories of thinkers such as Vilém Flusser and Harun Farocki as a central axis, we will delve into what this imperative operativity of the image means, both as a meaningful surface and as data within a computational process. Subsequently, based on the thought of Yuk Hui, we will propose that the contingent nature of the computational processes that generate and are fed by these technical images determines the shift from the redundant image to the informative image, considering this approach from the perspective of media archaeology, that is, considering the image from a temporal dimension. Finally, we address from an artistic standpoint the imperative operative and contingent informative nature of the current production of images as part of the databases that feed computational art, to question the concept of the image and allow us to propose that these are, rather than images, numerical hallucinations.