Visual Culture: Authorship, Style, and Boundaries of Algorithmic Participation

Abstract

The article examines generative neural networks not only as a tool for accelerating and improving the quality of work, but also as a factor that redefines the very process of content creation. Instead of the traditional hand-drawn embodiment of an idea, authors are increasingly engaged in setting the task, selecting options, reflecting on the result, and, if necessary, adjusting it. The focus shifts from a single image to a range of possibilities, and authorship becomes distributed. The review section analyzes studies that use neural networks to address issues related to authorship, artistic value, architectural visualization, and stylistic transformation of images. The theoretical section proposes a reevaluation of the fundamental categories of authorship, originality, style, and artistic value in the context of the use of generative models by authors. Methodologically, the work is based on comparative analysis, semiotic interpretation, and reconstruction of existing positions. Using examples from the fields of fine arts, design, architecture, and animation, it is shown that neural networks are well-suited for generating options and testing solutions, despite the fact that the results require human interpretation and expert selection, leaving humans responsible for the final outcome. As a result, neural networks are proposed to be defined not as independent creators, but as a specific media infrastructure that redistributes functions between humans, algorithms, and cultural contexts.