Artistic intelligence: Distinguishing Human from AI-Generated Art Through Perception

Abstract

With the new advances in technology, the ability of AI to produce human-like art questions established notions of creativity, authorship, and artistic value. This study explores the differences between human-created artwork and AI-generated artwork using human experience. We analyze whether or not AI-generated art can be distinguished from human-created art consistently based on qualities such as emotional effect, aesthetic depth, and originality. Our methodology involves a perception study where people rate and attempt to classify images as human or AI-created. The rating criteria are emotional effect, perceived beauty, detail richness, and subjective creativity. We also ask participants to see if they can identify common signs of AI-generated images, such as over-smoothing transitions, blurred details, and unnatural or flat viewpoints. We measure the accuracy of human classification, comparing the degree of consistency or inconsistency in their judgments. This transdisciplinary project aims to shed light on core differences in perception and structure between AI and human creativity. It also raises broader ethical and philosophical questions about creativity, ownership, and cultural value in this AI age. In particular, we question whether AI art models that have been trained from copyrighted content can produce original images that are just as good as human artwork. By comparing subjective human judgments, this research will seek to shed light on how AI art is understood, evaluated, and possibly distinguished from traditional human imagination. The results of this research will contribute to current discussions at the intersection of technology, art, and ethics, providing a foundation for more open and ethical AI art production.