Exploring Challenges and Opportunities to Support Designers in Learning to Co-create with AI-based Manufacturing Design Tools

Document Title

Exploring Challenges and Opportunities in Supporting Designers' Learning and Co-Creation with AI-Based Manufacturing Design Tools

Document Information

• Subject Area: Human-Computer Interaction (HCI), AI Collaborative Design, Computer-Aided Design
• Keywords: Human-AI Collaboration, Generative AI, Design Tools, Learning Challenges, Think-Aloud Method, Team Learning

Research Background and Issues

• Identified Problems or Challenges:

  • Designers require skills different from those needed for traditional CAD tools when using generative AI-supported design tools.
  • Current design tools lack effective mechanisms to support designers in learning to co-create with AI.
  • High-quality design outcomes often require designers to deeply understand and adapt to AI behavior.
  • There is a lack of empirical research on real-world design tasks, especially on how designers learn to collaborate with AI tools.

• Importance of the Issues:

  • Modern manufacturing processes are complex, and design tasks demand higher cognitive capabilities; AI tools can help optimize design objectives.
  • AI tools are increasingly becoming co-creative partners in the design process, but there is a significant learning curve for collaboration with AI.
  • With technological advancements, designers are likely to rely more on AI for solving complex design problems in the future.

• Research Motivation and Related Work:

  • Leveraging team learning and human collaboration theories to explore how to support designers in effectively collaborating with AI.
  • Applying existing theories from human collaboration, cognitive psychology, and learning sciences to human-AI collaborative design tools.
  • Observing interactions between designers and AI tools to identify challenges and propose improvements.

Solution

• Methodology:

  • Conducted two studies:
    • Study 1: Observed trained designers with no prior experience using AI tools as they collaborated with AI tools to complete complex design tasks.
    • Study 2: Recruited new designers to complete tasks alongside experienced "peer mentors," observing the support strategies employed by the mentors.

• Innovations:

  • Proposed a multi-level learning framework combining team learning and human-AI collaboration theories to understand designers' learning processes.
  • Identified typical challenges in designer-AI collaboration and provided specific recommendations for improving design tools.
  • Investigated the effectiveness of human-guided strategies in enhancing designers' collaboration with AI tools.

• Implementation Steps:

  1. Data Collection:
     • Used the "think-aloud method" to record designers' verbal thought processes.
     • Collected design outcomes from completed tasks and feedback from post-task interviews.
  2. Data Analysis:
     • Conducted qualitative and quantitative analysis of video interactions and interview content.
     • Coded design challenges and learning strategies, extracting key conclusions through thematic analysis.

Research Findings

• Key Findings:

  • Identified three major challenges:
    1. Difficulty in understanding and refining AI-generated results.
    2. Lack of a sense of "collaboration" with AI, leading some designers to avoid using AI features.
    3. Difficulty in effectively communicating design goal parameters to AI.
  • Successful designers employed the following learning strategies:
    1. Systematically testing the boundaries and capabilities of the AI.
    2. Enhancing problem understanding through self-explanation and reflective questioning.
  • Observed five support strategies used by human mentors, including step-by-step guidance, expressing uncertainty, and suggesting alternative solutions.

• Advantages and Comparisons:

  • Provided a systematic set of learning support strategies and design optimization recommendations to help designers fully utilize AI tools in complex tasks.
  • Validated the effectiveness of team learning theories in the domain of human-AI collaboration through real-world cases.

• Experiment and Evaluation Results:

  • Data indicated that the mentoring process significantly improved designers' collaboration with AI tools and their satisfaction with the design process (task completion rates increased).
  • Designers expressed a general expectation for more interactive tools capable of providing example-based suggestions.

• Limitations and Future Directions:

  • Limitations:
    1. Participants were mostly student designers with limited industry experience.
    2. The AI tools used were in early development stages, and interface issues may have influenced results.
    3. Tasks were conducted in experimental settings, which may differ from professional design environments.
  • Future Directions:
    • Explore learning support frameworks in other domains (e.g., image and music generation).
    • Enhance AI tools' contextual awareness and multimodal interaction capabilities.
    • Develop more dynamic human-AI collaboration models and create new tools to support increasingly complex design tasks.