From Screen Narratives to Prompt-Conditioned AI Myths: An Exploratory Structuralist Comparison

Abstract

Generative artificial intelligence has renewed debates about creativity, authorship, and artistic meaning, but structural similarity alone does not establish autonomous creativity. This exploratory study compares three human-authored feature films – Parasite (2019), John & Marsha sa Amerika (Part Two) (1975), and City Lights (1931) – with three ChatGPT narratives elicited through prompts that explicitly specified a mythic form and one of three preselected binary oppositions: wealth versus poverty, aspiration versus rejection, and appearance versus reality. Drawing on Saussure, Lévi-Strauss, a Propp-inspired analysis of spheres of action, and Barthes, the study examines oppositional organisation, mediation, character roles, symbolic signification, cultural-historical embedding, ideological ambivalence, and narrative closure. Under the documented prompt conditions, all three AI outputs instantiated several structuralist categories. Because both the binary opposition and the mythic genre were built into the prompts, however, these features are treated as prompt-conditioned rather than independently emergent. Relative to the selected films, the AI outputs exhibited stronger moralisation and narrative closure and weaker cultural-historical specificity. The findings are limited to a small, researcher-prompted corpus and do not establish a general capacity for autonomous AI creativity. The study contributes a cautious comparative framework for distinguishing prompt-induced structural conformity from analytically emergent features in AI-generated narrative.