An Emotion-Art Prompt Framework for AI-Generated Drawings in Education: System Architecture and Experimental Evaluation

Generative Artificial Intelligence (AI) has propelled a new paradigm in the generation of digital art, bringing new pedagogical practices regarding personalization and emotional growth in the classroom. This paper proposes the Emotion-Art Prompt Framework (EAPF) which provides a systematic link between Plutchik's Wheel of Emotions and Lövheim's three-dimensional model with basic underlying elements of art — form, color, light/shadow, and composition to facilitate drawing generation in an AI setting. This paper describes a new system architecture that autonomously maps layered emotional descriptors onto structured prompts interfacing with the Tencent Yuanbao AI drawing platform. Through a large-scale experiment involving more than 800 students (aged 10–12) in Shenzhen (China), we assessed quantitative and qualitative data embedded in emotional alignment, creative engagement, and platform suitability. Our findings show that EAPF substantially improves the emotional expressiveness of AI-generated works of art, while offering data-based insights for educators to tailor art teaching to individual student preferences. Thus, this study demonstrates the potential of well-designed prompt engineering to overcome accessibility barriers in traditional art education while enhancing emotional development. Additionally, it proposes a scalable model for integrating AI technologies into K–12 creative learning.