AI as Synthetic Users in Human-Centred Design: A Systematic Review

In human-centred design (HCD), real users are the foundation of the design process, but ensuring their active involvement is often challenged by time, cost and resource constraints. Recent advances in artificial intelligence (AI) have introduced the concept of 'synthetic users' - AIdriven agents that simulate human participation in design and evaluation processes. This paper presents a systematic literature review of 52 peer-reviewed studies investigating how large language models (LLMs) and agent-based simulations function as synthetic users in different phases of HCD. Following PRISMA guidelines, we conducted structured searches in Web of Science and Scopus and applied inductive thematic analysis to group the studies into four main themes: (A) AI-generated personas and user profiles for user research; (B) AI support for ideation and prototyping; (C) simulated users for usability testing and evaluation; and (D) agent-based simulations for system validation and decision support. Our findings show that while synthetic users can significantly accelerate early design iterations and support creative exploration through rapid and scalable feedback, they also raise concerns about authenticity, variability, bias, and ethical validity. We examine these trade-offs and propose a hybrid evaluation model in which AI-driven techniques augment - without replacing - real user input, thereby preserving the fundamental role of real user insight in human-centred design.