Artificial Intelligence (AI) is redefining the boundaries of many industries. At Conflux, we are a UX agency. We are exploring new AI tools for user experience research and design. In particular, we have introduced synthetic users in UX research, which are AI-generated simulations of real ones. These digital entities, created to replicate the behaviors, preferences, and interactions of real users, are gaining interest in the field as a promising tool to optimize research.
AI offers many opportunities in the UX field, many of which are still to be fully explored. With AI, UX researchers can collect and analyze massive amounts of data and gain useful insights for understanding users, helping to evolve their research work further. Complementing real user research with synthetic ones can make the process faster, more efficient, and potentially less costly.
In this article, we will dive into what synthetic users are and how they are used in UX research. We will consider both their potential benefits and limitations. We’ll explore how to get the most out of this tool by using a balanced UX research approach that integrates AI-generated users with traditional research methods.
Synthetic users are one of the new AI-based tools making their way into UX research, generating some debate about their use. These non-real people offer researchers the ability to interact with AI-generated profiles that can realistically simulate the behavior and responses of specific user groups. Unlike traditional user personas, which are static profiles based on research and demographic data, synthetic personas are interactive, allowing researchers to ask questions, conduct interviews, and obtain simulated feedback.
Their functionality is based on large language models (LLMs) like ChatGPT, trained on vast datasets of human behavior, conversations, and online interactions. Specialized platforms like Synthetic Users leverage and combine these LLMs to create custom synthetic profiles tailored to the specific needs of UX research.
To create and use synthetic users, researchers first define the target user groups, qualitative and quantitative research goals, and the type of interview they want to conduct. In addition to specifying the user group being studied, researchers can input details such as specific needs, goals, or fears. At this point, the platform generates AI personas profiles similar to user personas, complete with backgrounds, experiences, and goals relevant to the chosen group.
Researchers can interact with synthetic users through simulated interviews once the synthetic profiles are created. Generally, these interviews can be used to explore a specific issue, ask probing questions to investigate potential customers’ motivation, or gather user feedback on a solution or prototype.
In the case of Synthetic Users, the platform will produce detailed transcripts of these interviews, allowing researchers to analyze responses, identify trends or potential usability issues, or ask follow-up questions and continue the interview. In other cases, synthetic user platforms may generate summary reports based on interactions, highlighting key findings and providing recommendations to improve UX/UI design.
Using a synthetic sample offers a fast and efficient way to gather qualitative data and test different design hypotheses, especially in the early stages of product development. However, it’s important to remember that synthetic users are AI-based simulations, and while they can provide valuable insights for research, they cannot fully replace testing with real people.
Integrating synthetic users in UX research offers several advantages that can enhance the efficiency, speed, and depth of analysis.
Accessibility and speed. Dedicated platforms for creating synthetic users provide quick access to detailed user information, allowing researchers to collect data and feedback much faster than traditional methods. This can be especially useful in agile development environments, where research time is limited and the ability to gather quick user insights is critical for making informed decisions and timely adjustments to the product or service in development.
Exploring new domains or user groups. When a research team approaches a new field or an unfamiliar target, synthetic users can provide an initial knowledge base, allowing researchers to gather preliminary information, identify potential needs and pain points, and formulate research hypotheses to be explored later with traditional methods.
Generating new research hypotheses. Interactions with synthetic users can provide valuable insights into user needs, behaviors, and preferences, which researchers can use to develop initial hypotheses that can be tested and validated later through research with real users.
While synthetic users offer significant advantages, it is essential to be aware of their limitations to use them effectively and responsibly in UX research.
Tendency to provide overly positive feedback. By nature, AI is designed to please and meet user requests. This can lead to shallow and uncritical responses, a phenomenon known as sycophancy. In UX research, this can hinder the identification of potential issues or areas for improvement, as interviews with synthetic users may not reveal critical problems or constructive feedback.
Difficulty capturing the complexity and nuances of human behavior. Values, desires, needs, and emotions are inherently human aspects that can be difficult to replicate accurately through an AI model. For example, in a study on participation in an online course and discussion forum, synthetic users might provide overly optimistic feedback, stating they will surely complete the course and actively participate in the forum. In interviews with real a person, we might encounter more nuanced responses, highlighting potential barriers and challenges that could influence their behavior.
Inability to replicate real-world experiences. Since they cannot interact with a product or service in real life, insights from synthetic users must be interpreted cautiously, especially when referring to their “past experiences.” For example, when designing a navigation app for driving, a synthetic user might not be able to provide realistic feedback on potential distractions or usability issues related to using the app in a real-world context.
The quality of information is limited by the data on which they were trained. If the training data is incomplete or biased, we will see its effects in the synthetic user responses. This occurs more frequently when researching a very specific group, a niche likely underrepresented in the dataset used to train the LLM. In such cases, synthetic users may not be able to provide accurate or representative information about their needs and expectations.
Given the limitations outlined, it is clear that synthetic users should be considered a complementary research tool to those traditionally used in UX research. AI-generated users offer many benefits, but interaction with real ones remains essential for gaining an accurate understanding of their experiences, emotions, and motivations.
At our UX agency, we adopt an integrated approach, using synthetic users in combination with traditional research methods. For example, we use synthetic users to explore a specific domain, generate initial hypotheses, and gather preliminary feedback in the early stages of design. Additionally, in the context of automating UX research, we create custom GPT models to perform tests on mockups, aiming for more focused interface data.
We rely on real user research to validate our hypotheses, gather in-depth qualitative real data, and comprehensively understand users’ needs and expectations.
To make the most of synthetic users in UX research, it is essential to adopt a strategic and mindful approach that maximizes the benefits while consciously managing the tool’s inherent limitations.
First, it’s crucial to combine synthetic users with traditional research. Synthetic users can provide valuable insights and accelerate some research phases, but they cannot replace direct interaction with real people. Combining traditional and AI-based research tools enhances our ability to investigate and fully understand user expectations and behaviors.
Second, it’s important to treat the outputs of synthetic users as hypotheses to be validated. AI-generated information can be used to formulate initial hypotheses, identify potential areas of interest, and develop research questions to be explored later with real users. The results of interviews with synthetic users should be considered starting points for further investigation.
Being transparent with stakeholders about the use of synthetic users in research is equally important. It is essential to communicate to stakeholders that the results obtained from an AI-generated persona do not replace real customer personas data and that their use is valid only when combined with other research methods. This transparency is key to ensuring that design decisions are based on solid and reliable data.
Finally, always be aware of the limitations of synthetic users. As mentioned earlier, synthetic users can exhibit biases, provide overly positive feedback, and struggle to capture the complexity of human behavior. It’s important to take these limitations into account, especially when designing for specific populations or addressing sensitive topics, and use synthetic users responsibly and ethically.
Artificial intelligence and LLMs offer many opportunities for UX design. In particular, synthetic users are a promising resource in the ever-evolving landscape of UX research. Their ability to provide rapid and accessible user insights, generate research hypotheses, and support the exploration of new domains makes them a valuable tool for UX designers and researchers.
However, it’s important to remember that synthetic users cannot fully replace interaction with real users. Qualitative research conducted through traditional methods remains indispensable for gaining an accurate and empathetic understanding of users, enabling data-driven design decisions, and creating products and services that meet their needs.
If you want to create a new solution or improve your customer’s experience with your product or service, Conflux is the UX agency for you. Our UX research and design team in the AI Studio uses AI technologies to create intuitive and engaging user products and services, conducting thorough investigations and developing revolutionary interfaces. Contact us for a consultation.
