Year: 2025
Type: dissertation
Abstract: This thesis explores how creativity is perceived in human- and AI-generated content across three creative domains: Visual Art, Music, and Graphic Design. It investigates the extent to which ML outputs are regarded as original, emotionally impactful, and authentically creative. Using a within-subjects, blind comparison design, 63 participants evaluated paired content— one created by a human, the other by an ML system—through Likert-scale ratings and qualitative reflections. A cross-domain assessment structure ensured consistent evaluation across emotional, creative, and stylistic dimensions. The results show that AI-generated content was frequently rated as creative and structurally original, with some outputs scoring comparably to or even surpassing human works. However, AI consistently underperformed in emotional impact, especially in the music and design domains. Participants often described AI outputs as technically polished but emotionally limited, reinforcing their weaker performance in these areas. The findings support the view of ML as a co-creative tool rather than an autonomous creative agent. Based on these insights, the study proposes practical recommendations for developing emotionally aware, user-guided, and culturally adaptive AI systems in creative industries. This research contributes to the growing body of work on human–AI collaboration by identifying both the capabilities and constraints of ML in replicating human creativity, and by outlining a framework for designing more ethically and expressively aligned creative AI tools.
Author: Jafari, Shakila
Language: English
Topic: Digital Media and Visual Art
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7112081448