Proverbs as indicators of proficiency for art-generating AI.

This article examines the ability of five popular Artificial Intelligence (AI) art generators—Nightcafé, Dream, starryai, DALL-E 2, and MidJourney—to interpret and visually represent figurative language, specifically five well-known proverbs common across Western languages. The study finds that these AI platforms generally struggle to process the metaphorical and complex syntax inherent in proverbs, resulting in images that rarely convey the intended figurative meanings without explicit human guidance. Among the tested AIs, starryai produced the most accurate, though still limited, visual interpretations, with simpler proverbs like "all roads lead to Rome" yielding better results than more complex ones. The article suggests that integrating disciplines such as psycholinguistics and semiotics into AI development could improve AI's understanding of figurative language, emphasizing that current AI art generators cannot yet fully substitute human creativity in this domain.