Assessing Large Language Models Adoption in Need Finding: an Exploratory Study

This paper explores using large language models (LLMs) in the early stages of the user-centered design (UCD) process, particularly in defining user needs. While UCD emphasizes understanding users' needs, preferences, and contexts, LLMs introduce new possibilities for enhancing this process by assisting designers rather than replacing them. Through example prompts and case studies of three mobile app developments , this study investigates how LLMs can generate user profiles, needs, and personas. We highlight the potential benefits of LLM integration into UCD, such as increased efficiency and deeper insights, as well as inherent limitations like the lack of emotional depth in AI-generated interviews and the challenges in detecting inaccuracies. The findings underscore the importance of human involvement to ensure the fidelity and relevance of user simulations, suggesting a balanced approach for integrating AI tools in UCD.