[Paper Review] Generative AI for Product Design: Getting the Right Design and the Design Right

Matthew K. Hong, Shabnam Hakimi+4 more|arXiv (Cornell University)|Jun 2, 2023

Design Education and Practice12 citations

TL;DR

A position paper discussing barriers and a research agenda for applying Generative AI in two product design phases—getting the right design and getting the design right—with emphasis on human-centered challenges and interdisciplinary collaboration.

ABSTRACT

Generative AI (GenAI) models excel in their ability to recognize patterns in existing data and generate new and unexpected content. Recent advances have motivated applications of GenAI tools (e.g., Stable Diffusion, ChatGPT) to professional practice across industries, including product design. While these generative capabilities may seem enticing on the surface, certain barriers limit their practical application for real-world use in industry settings. In this position paper, we articulate and situate these barriers within two phases of the product design process, namely "getting the right design" and "getting the design right," and propose a research agenda to stimulate discussions around opportunities for realizing the full potential of GenAI tools in product design.

Motivation & Objective

• Explain how GenAI can influence the two phases of product design: getting the right design and getting the design right.
• Identify human-computer interaction (HCI) and organizational challenges that limit GenAI adoption in industry design practice.
• Propose a research agenda to explore opportunities for responsible and effective GenAI use in product design.
• Highlight ethical, legal, and diversity concerns associated with GenAI in design and advocate for responsible use.

Proposed method

• Frame GenAI integration within two design phases to structure challenges and opportunities.
• Discuss design-space exploration, inspiration versus high-fidelity output, and prompt engineering as practical considerations.
• Analyze consumer preference modeling and the need for multi-modal representations to ground design decisions.
• Identify timelines and dependencies for updating models with shifting consumer trends and preferences.

Experimental results

Research questions

• RQ1 What barriers prevent GenAI from effectively supporting the two phases of product design (getting the right design and getting the design right)?
• RQ2 How can interaction design and prompt strategies be improved to balance inspiration with design rigor in GenAI-assisted design?
• RQ3 What are appropriate ways to model and integrate consumer preferences (including behavioral and physiological signals) to guide design decisions?
• RQ4 What are the ethical, legal, and diversity concerns when using GenAI for product design, and how can responsible use be promoted?

Key findings

• GenAI can potentially accelerate design-space exploration but may cause design fixation if used to produce high-fidelity outputs early in the process.
• There is a need for mechanisms to increase idea diversity and to control the level of diversity in GenAI outputs.
• Prompt engineering challenges surface as translating visual concepts to text prompts is cognitively difficult and can lead to misalignment with designer intent.
• Multi-modal representations of consumer preferences (beyond text) are proposed to ground product design, including behavioral and physiological data.
• Static, frozen LLMs struggle to adapt to shifting consumer trends, suggesting a need for complementary methods to track slow and emerging trends.
• The paper emphasizes legal and ethical risks, lack of dataset diversity, and calls for responsible GenAI use in design.