TL;DR
This paper proposes and evaluates an 8-week "AIGC-enhanced" curriculum for interface design that shifts the focus from traditional software training to human-AI collaboration. By integrating tools like Midjourney and Uizard, the study demonstrates how students can develop sophisticated competencies in prompt engineering, visual logic, and user empathy, ultimately moving from passive tool-users to active "co-authors" with intelligent systems.
What Problem Does This Paper Address?
The rapid rise of Artificial Intelligence Generated Content (AIGC) tools—like ChatGPT and Midjourney—has disrupted traditional design education. Most current UI/UX curricula are "tool-centered," focusing on technical proficiency in software like Figma or Adobe XD. However, these programs often fail to teach the "semantic interface" of prompt engineering or how to maintain creative agency when AI can automate routine visual tasks.
The authors address the urgent need for a "postdigital pedagogy": a way to teach design that doesn't just treat AI as a shortcut, but as a collaborative partner that can actually deepen a student's conceptual understanding and empathetic reasoning.
Methods
The researchers conducted a qualitative case study over one academic term (8 weeks) at a vocational university in China.
- Participants: 28 second-year undergraduate students majoring in Visual Communication Design.
- The Curriculum: A studio-based model where AIGC tools were embedded into ideation, prototyping, and reflection phases. Students used Midjourney for visual exploration and Uizard for rapid layout generation.
- Data Collection: To ensure "triangulation" (accuracy through multiple sources), the authors collected student artifacts (prompt logs, journals), observational field notes, focus group interviews, and peer/self-evaluations.
- Analysis: Qualitative data was coded thematically using NVivo software, following the established Braun and Clarke (2006) six-phase approach to identify patterns in student learning.
Datasets
As this is a qualitative educational study rather than a machine learning benchmark paper, it does not use standard ML datasets. Instead, the "data" consists of:
- Student-generated artifacts: Design works, interface mockups, and detailed AI prompt logs.
- Qualitative records: Interview transcripts from 10 volunteer students and weekly instructor reflection notes.
Key Insights and Ideas
- Prompting as a New Literacy: The study frames "prompt engineering" not as a technical hack, but as a core design skill. It requires linguistic precision to bridge the gap between human intention and machine output.
- AI as a "Third-Party Interlocutor": Rather than just a tool, AI acted as a sounding board. For example, when students generated AI portraits of "elderly users," it triggered immediate, empathetic reflections on accessibility needs like font size and color contrast that might have been overlooked in a traditional workflow.
- From Copying to Justifying: Students noted a shift from "copying styles from Behance" to using AI-generated variations as provocations. This forced them to use professional vocabulary (hierarchy, affordance, visual logic) to justify why one AI-generated version was better than another.
- The Mindset Shift: One of the most significant findings was the transition of students from "passive recipients of software training" to "creative explorers." The open-ended nature of prompting encouraged a "fail fast, learn fast" mentality.
Key Results
- Prompt Evolution: Students showed a clear trajectory from vague inputs (e.g., "design a login page") to highly specific, iterative prompts (e.g., "mobile login page for elderly users, soft color palette, minimalist style, centered CTA").
- Improved Visual Reasoning: Peer feedback sessions showed that students became significantly better at justifying design choices based on logic rather than arbitrary aesthetic preferences.
- Qualitative Feedback: Students reported feeling more "ownership" over their projects. One student noted, "It felt more like exploring than finishing an assignment. I could try ideas quickly, and the results gave me new directions."
- Agency: The curriculum successfully taught students to "question, adapt, remix, and sometimes reject" AI output, ensuring the human remained the primary decision-maker.
Limitations and Future Work
The authors acknowledge that this was a small-scale study (28 students) within a specific cultural and institutional context (a Chinese vocational university).
- Long-term effects: It is unclear if these skills stick or how they evolve as AI tools themselves change.
- Scalability: Future research is needed to see how this model adapts to different disciplines or larger, more diverse student bodies.
- Ethics: While the study mentions "social responsibility," more work is needed on how to explicitly teach the ethical implications of AI-generated content (bias, copyright, etc.) within the design studio.
Summary updated