Developing a Method for Investigating Design Needs Using Large Language Models (LLMs)

Hierarchical Extraction of Market Needs Using ChatGPT

Abstract

This study explores a novel approach for extracting hierarchical user needs by combining ChatGPT with the Evaluation Grid Method (EGM). By treating a Large Language Models (LLMs) as an interviewee, the research simulates EGM interviews to uncover layered consumer values. Preliminary experiments using bicycle design examples show that role-playing prompts elicit more specific and design-relevant responses. The study aims to establish a standardized technique for efficiently identifying design needs from LLMs, making it practical for use in product and service design and development.