Integrating Generative AI into Chatbot UI Design Processes: A Workflow Using ChatGPT and Midjourney

This study proposes a generative AI–assisted workflow for early-stage chatbot user interface (UI) design. While chatbot persona and visual form significantly influence user trust, emotional engagement, and service perception, they are often addressed only after core functionality is implemented, leading to misalignment between visual expression and interaction goals. To address this gap, eighty real-world chatbots across six industry sectors were analyzed to construct an evidence-based framework—the Visual Taxonomy Matrix (VTM)(Figure 1)—that maps relationships among visual form, emotional tone, and service purpose. Building on these insights, a five-stage workflow integrating ChatGPT for textual persona ideation and Midjourney for visual exploration was developed. The workflow was applied in a first-year “Introduction to Computer Graphics” course through the HanseoBot project, improving students’ creative confidence and design comprehension. Although limited to a single case, the study demonstrates how generative AI can enhance multimodal ideation and process efficiency in early design stages, offering both educational and practical implications.