Leveraging Large Language Models in User eXperience (UX) Education when Real Users Are Unavailable

Area of Science:

• Human-Computer Interaction
• Educational Technology
• Healthcare Design

Background:

• User eXperience (UX) design necessitates direct user engagement, particularly in healthcare, to understand patient and clinician needs.
• Ethical and logistical challenges frequently impede early-stage user access in research.
• Traditional UX methods face limitations when real user interaction is not feasible.

Purpose of the Study:

• To investigate the utility of Large Language Models (LLMs) as a tool for simulating user perspectives in UX design education.
• To assess the effectiveness of LLM-generated simulated user data in developing UX artifacts like proto-personas and empathy maps.
• To evaluate the potential of LLMs to enhance UX design education, especially in resource-constrained or access-limited environments like healthcare.

Main Methods:

• A UX workshop was conducted with undergraduate engineering students.
• A Large Language Model (ChatGPT) was employed to simulate patient and clinician user perspectives.
• Students developed prompts to guide the LLM in generating realistic user responses.
• LLM outputs were utilized to create proto-personas, empathy maps, and initial design concepts.

Main Results:

• Students successfully used LLM outputs to create foundational UX design artifacts.
• The simulation provided a practical environment for students to practice human-centered design methods.
• The approach offered a viable alternative for UX education when direct user access was limited.
• Potential biases inherent in LLMs were acknowledged as a limitation.

Conclusions:

• Large Language Models can serve as a valuable supplementary tool in UX design education, particularly for simulating user perspectives.
• Thoughtful integration of LLMs can enhance student learning of human-centered design principles and prepare them for real-world challenges.
• While LLMs cannot substitute for authentic user research, they offer a pragmatic solution for overcoming access barriers in educational settings.
• Further research is warranted to explore the mitigation of LLM-induced biases in UX design.