I-Card: A Generative AI-Supported Intelligent Design Method Card Deck

Research Background and Issues

• What problems or challenges did the authors identify?
  Design method cards help designers understand and apply various design methods through a concise format. However, despite the advantages of consistency and simplicity, designers are often hesitant to use them because they lack dynamic and context-specific support. Additionally, designers lack tools that effectively translate design methods into practice.

• Why is this issue important?
  Design methods are core tools for guiding innovation and systematic design activities. If designers cannot efficiently utilize design method cards to solve real-world problems, it may affect design efficiency, method applicability, and the quality of the final design. Furthermore, the static nature of the cards does not adequately support the flexibility and complexity required for dynamic task demands.

• Research Motivation and Related Work
  Early studies, such as those on SUTD-MIT and IDEO design method cards, provided a consistent structure for methods, but their static form limited adaptability to different contexts and the depth of exploration. In recent years, other design tools have begun integrating generative AI to enhance customization and knowledge expansion capabilities. However, research on AI support specifically for design method cards remains relatively unexplored. Therefore, this study aims to explore how generative AI can enhance the applicability and dynamism of design method cards.

Solution

• What methods or solutions did the authors propose?
  The authors developed an intelligent design method card system, I-Card, which integrates generative AI. It personalizes the generation of design methods, supports knowledge transfer through specific cases, and improves user experience with interactive and dynamic card designs.

• What are the innovative aspects of this solution?

  1. Dynamic Recommendation Mechanism: Provides customized design method recommendations based on user input.
  2. Five Card Types: Includes Info Card, Method Card, Solution Card, QA Card, and Resource Card, addressing the entire process from information input to implementation.
  3. Generative AI Support: Utilizes large language models to enable personalized content generation, question answering, and data support.
  4. Interactivity and Dynamism: Card content updates in real-time based on user interactions, supporting iterative design processes.

• What are the implementation steps and key technologies used?

  1. Research Design and Needs Analysis: Structured research to identify designers' support needs.
  2. Technical Implementation: Developed the I-Card prototype using Vue.js and Node.js, leveraging generative AI (e.g., GPT-4) and retrieval-augmented generation (RAG) for content generation and data retrieval.
  3. Feature Design and Iteration: Developed five card types and conducted five rounds of iterative improvements.
  4. User Experimentation: Validated the effectiveness of I-Card in improving design efficiency, method applicability, inspiration generation, and decision support through user experiments.

Research Outcomes

• What specific outcomes were achieved?

  1. Efficiency Improvement: I-Card significantly reduced the time required to select design methods, saving approximately 2 minutes compared to traditional SUTD-MIT cards.
  2. Enhanced Applicability: Personalized method guidance made it easier for designers to understand and apply complex design methods.
  3. Inspiration Generation: Dynamic solution steps and data support provided deeper design insights for designers.
  4. Decision Support: QA Cards and Resource Cards offered knowledge and data support for specific design scenarios, enhancing data-driven decision-making.

• What advantages does it have compared to existing solutions?
  Compared to traditional static cards:

  1. I-Card integrates dynamic content through generative AI, improving flexibility.
  2. Reduced cognitive load: Designers do not need to manually adapt methods to specific contexts.
  3. Provides multi-modal interactions (text, images, and charts) to enhance user experience.

• What were the experimental or evaluation results?
  User experiments demonstrated that I-Card significantly improved design efficiency, method applicability, inspiration generation, and decision support. Qualitative feedback also indicated that its dynamic features helped designers adapt to various design domains.

• Limitations and Future Directions

  1. Generative AI Hallucination Issues: Some generated content may be inaccurate, necessitating the addition of content verification mechanisms (e.g., knowledge graphs, reinforcement learning).
  2. Professional Expansion: Current user studies primarily targeted students and novice designers; further validation is needed for professional designers.
  3. Extension to Other Method Types: The I-Card framework could be expanded to include Intent Cards and Tool Cards in the future.

Conclusion

I-Card not only demonstrates the potential of generative AI in complex knowledge transfer but also showcases how dynamic, highly interactive systems can improve user experience. Its research outcomes provide new perspectives on knowledge transformation design in the HCI field, particularly in enhancing adaptability and interactivity, with broad practical and research implications.