REPRODUCTION OF ARTIST’S UNIQUE VISUAL STYLE. ARTIFICIAL INTELLIGENCE (AI) – TOOL FOR THE OPTIMIZATION OF CREATIVE PROCESSES

Abstract

This study is dedicated to examination of the ways how generative AI, specifically trained open-source models by Stable Diffusion, can support reproducing an artist’s unique visual style. Conducted across four semesters (2023–2025) at the Art Academy of Latvia, the research investigates the pedagogical and creative outcomes of teaching students to train personalized AI models using their artworks as datasets, further combining other tools with the trained models during the study process.

A mixed-method approach – combining surveys, AI image recognition tests, semi-structured interviews, practical assignments, and qualitative observation – was applied to assess the effectiveness of AI-assisted style reproduction. The study evaluates challenges related to further combining models and learned tools, dataset preparation, authorship, and the discussion and research about cognitive differences between human and machine creativity.

Findings show that personalized model training can enhance creative autonomy, enabling results closer to an artist’s unique visual style than generic AI tools. However, the process also reveals ethical and conceptual tensions regarding authorship, randomness, and control. The study concludes that, when applied critically and reflectively, generative AI can serve as a powerful tool for both creative exploration and pedagogical development.