Beyond Input–Output: Rethinking Creativity through Design-by-Analogy in Human–AI Collaboration

Paper Title

Beyond Input–Output: Rethinking Creativity through Design-by-Analogy in Human–AI Collaboration

Publication Info

• Topic area: Design-by-Analogy (DbA) as a framework for enhancing creativity in human–AI collaboration.
• Keywords: Design-by-Analogy, creativity support, human–AI collaboration, analogical reasoning, cognitive science, creative process, systematic review, technological mediation.

Background and Problem

• Problem / challenge: Current AI-driven design tools often oversimplify creativity into input-output pipelines, leading to design fixation and homogenization of creative content. Existing Design-by-Analogy (DbA) research is fragmented, focusing on specific stages or modalities, and lacks a holistic framework.
• Significance: Addressing these challenges is critical to preserving human creativity, fostering innovation, and mitigating the risks of oversimplified AI systems in creative domains.
• Motivation and related work: DbA has been applied in fields like bio-inspired design and mechanical design, but prior reviews are limited to theoretical frameworks, specific data modalities, or isolated techniques. This paper aims to bridge these gaps by systematically analyzing DbA across creative processes, representations, and domains.

Solution

• Proposed approach: A systematic review and analysis of Design-by-Analogy (DbA) research to establish a comprehensive framework for its application in human–AI collaboration.
• Novelty:
  1. Systematic review of 85 studies to construct a unified corpus of DbA research.
  2. Taxonomy of six representational forms and classification across seven stages of the creative process.
  3. Guidelines and future directions for AI-supported DbA, emphasizing ethical considerations and human-centered design.
• Procedure and key techniques:
  • Conducted a PRISMA-guided systematic review from an initial corpus of 1,615 publications.
  • Identified six representation forms: Semantics & Text, Visual & Appearance, Material & Structure, Function & Attribute, Interaction & Experience, and Unconventional Contexts.
  • Mapped these representations to seven stages in the creative process: Vision, Inspiration, Ideation, Prototype, Fabrication, Evaluation, and Meta.
  • Analyzed applications in three domains: Creative Industries, Intelligent Manufacturing, and Education & Service Industries.

Results

• Concrete findings:
  • Six representational forms span all stages of the creative process, with Function & Attribute (54.1%) and Semantics & Text (40.0%) being the most prevalent.
  • DbA systems support creativity at varying levels of automation: Assist (35.3%), Augment (38.8%), and Automate (27.1%).
  • Applications include creative writing, intelligent manufacturing, and educational tools, with notable gaps in film, music, and immersive interfaces.
• Advantage over baselines:
  • DbA retains human agency and integrates cultural/contextual signals, unlike rigid Case-Based Reasoning (CBR) systems.
  • Reduces design fixation and fosters innovation by enabling far- and near-distance analogies.
• Experiments / evaluation:
  • Systematic review validated with inter-rater reliability (Cohen’s Kappa = 0.7289).
  • Applications analyzed across domains, with examples of DbA systems enhancing creativity in design, manufacturing, and education.
• Limitations and future work:
  • Limited exploration of DbA in fabrication stages and emerging domains like sustainability.
  • Lack of structured public knowledge repositories and empirical evaluation of AI-driven DbA outputs.
  • Future research should address ethical risks, multimodal interfaces, and tacit knowledge integration.

Summary

This paper systematically reviews Design-by-Analogy (DbA) research, identifying six representation forms and mapping them to seven stages of the creative process. It highlights DbA’s potential to mitigate design fixation, enhance creativity, and support human–AI collaboration across domains like creative industries, intelligent manufacturing, and education. The study emphasizes the ethical and practical challenges of DbA implementation, proposing guidelines for user-centered, context-adaptive, and ethically responsible systems. By framing DbA as a mediating technology, this work lays the foundation for advancing creativity support tools in the age of AI.