EVALUATING ARTISTIC MERIT OF AI-GENERATED PHOTOGRAPHS
Abstract
The paper will address the artistic quality of AI-generated photographs and combine the aesthetic theory, computational imaging, and empirical analysis. With the growing ability of generative models to generate images of visual appeal, issues of creativity, originality, authorship, and cultural worth emerge. The research places AI photography in the frame of classical and contemporary aesthetics, the intentionality of humans and the algorithmic production. The work technologically reviews GAN-based, diffusion-based and transformer-driven image synthesis in focus of prompt engineering and human-AI co-creation workflow. The methodologically based curated set of AI-generated and human-created photographs are built, including a variety of genres, cultural motifs, and stylistic traditions. Mixed-method evaluation system is a hybrid quantitative rating scale with qualitative ratings by expert photographers, artists and curators as well as surveys by the audience. Comparative studies evaluate the quality of perceptions, emotional appeal, originality, and depth of storytelling of human and AI outputs and the output of various generative models. Findings suggest that AI generated photographs can take the top positions in terms of high technical and compositional ratings, but they are inconsistent in terms of perceived purposefulness and situational meaning. Evaluation is largely mediated by cultural background which manifests bias and conflicting aesthetic priorities. The article has implications on the practice, education, and curation of art, stating that AI photography should be seen as a hybrid creative process instead of a substitute of human art.