Between Reproduction and Creation

The Intersection of AI and Art with Midjourney

Resumo

This study examines the intersection of artificial intelligence and art through an analysis of Midjourney’s capabilities in interpreting and recreating artworks considered canonical in the western canon of art history. Using a mixed-method approach combining computational art analysis with qualitative comparative assessment, we investigated how AI-generated imagery challenges traditional notions of artistic authenticity, authorship, and originality. Five iconic works from the public domain—Van Gogh’s “The Starry Night,” Vermeer’s “Girl with a Pearl Earring,” Hokusai’s “The Great Wave off Kanagawa,” Wood’s “American Gothic,” and Munch’s “The Scream”—were analyzed using Midjourney’s describe function and subsequent image generation across technical, stylistic, and semantic dimensions. Findings reveal that prompt specificity critically shapes output fidelity: concise prompts tend to produce more faithful reproductions, while elaborate, detail-rich prompts invite broader creative reinterpretation. The AI consistently demonstrated strong technical proficiency in replicating compositional elements and color palettes, while showing notable tendencies toward stylistic amplification and occasional semantic misinterpretation of cultural context. These results contribute to ongoing debates about AI’s role in artistic production, suggesting that Midjourney functions less as a precise replicator and more as an interpretive collaborator, generating hybrid works that exist in tension between reproduction and creation, fidelity and innovation, human intention and algorithmic pattern recognition.