Generating personas using LLMs and assessing their viability
Andreas Schuller, Doris Janssen, Julian Blumenröther, Theresa Maria Probst, Michael Schmidt, Chandan Kumar
Abstract
User personas, reflecting human characteristics, play a crucial role in human-centered design, contributing significantly to ideation and product design processes. However, expressing a diverse range of product-related human characterizations poses a challenging and time-consuming task for UX experts. This paper explores the utilization of Large Language Models (LLMs) to streamline the generation of personas, thereby enhancing the efficiency of UX researchers and providing inspiration for stakeholder discussions. Towards this objective, we devised strategic prompts and guidelines involving stakeholders and potential product features, resulting in the creation of candidate user personas. These personas were then compared with those crafted by human experts in a remote study involving 11 participants assessing 16 personas each. The analysis revealed that LLM-generated personas were indistinguishable from human-written personas, demonstrating similar quality and acceptance.