Can LLM-driven synthetic participants help user research? A case study in designing augmented reality for education

ABSTRACT: Conventional user research with human participants faces significant challenges, including substantial time and resource requirements, and limited scalability. In response, this study presents an efficient, cost-effective workflow driven by large language models (LLMs) for simulating user research with synthetic participants (SPs) at scale. In a case study in design augmented reality for education, SPs’ open-ended answers were plausible and comprehensive, yet semi-open and closed items diverged from those of humans. SPs can augment early qualitative work, but cannot replace human studies.