Building Grounded Synthetic Users: Generative AI as Mediating Artefact in User Research

This paper presents a grounded strategy for constructing Synthetic Users (SUs) with large language models to support early-stage design exploration. SUs are used to extend empirical user data into structured multi-turn dialogue within multidisciplinary design processes. An exploratory investigation with a home appliance manufacturer examines the alignment patterns, contextual behaviour, and interpretive characteristics of SU responses through semantic similarity indicators and qualitative analysis. The findings show that SUs provide coherent, context-sensitive design input while requiring careful calibration in socially nuanced scenarios.