Synthetic users: the next revolution in UX Research

Carolina Guimarães delves into the emerging concept of synthetic users in UX research, exploring how artificial intelligence (AI) can simulate human behaviors to provide insights traditionally gathered from real users. The idea is that by leveraging large language models (LLMs) to create “synthetic users,” researchers can simulate interactions that mirror those with real people, allowing for quicker and potentially more scalable research processes. However, this innovation brings forth both methodological and philosophical questions, especially regarding the quality and authenticity of the insights gathered from AI-driven entities.

Synthetic users, according to Guimarães, are essentially digital personas crafted from vast datasets, acting as avatars or guides in research. These synthetic entities differ from traditional personas, which are built through direct interaction with real users. Instead, synthetic users are generated by AI, programmed to reflect behaviors, preferences, and usage patterns based on data collected from various online sources. The article discusses how these synthetic users can accelerate the research process, offering a more efficient way to identify obvious insights, thereby allowing researchers to focus on more complex and innovative aspects of design.

However, the use of synthetic users is not without limitations. The article highlights the potential biases that could arise from relying on AI, such as the biases inherent in the training data, which might not fully represent diverse user groups. Moreover, the lack of non-verbal cues and emotional nuances in AI-driven research presents another significant limitation. The richness of human interaction, which includes facial expressions and intonations, is lost when working with synthetic users, potentially leading to incomplete or skewed insights.

Despite these challenges, synthetic users hold potential for transforming UX research. They could serve as valuable tools for preliminary research phases, quickly processing large volumes of data and providing actionable insights that speed up the design process. However, the article emphasizes that synthetic users should complement, not replace, real user interactions. As AI continues to evolve, the key will be in refining these tools to enhance their reliability and usefulness in UX research, while always maintaining a critical eye on their limitations