Reproduce “UXAgent: An LLM Agent-Based Usability Testing Framework for Web Design” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Yuxuan Lu, Bingsheng Yao, Hansu Gu, Jing Huang, Jessie Wang, Yang Li, Jiri Gesi, Qi He and 2 more
UXAgent, developed by researchers from Northeastern University, the University of Notre Dame, and Amazon, presents an LLM-Agent-based framework for conducting usability testing on web designs. It generates large-scale simulated user behaviors and comprehensive usability data, providing UX researchers with a rapid, iterative feedback loop for refining study designs before human-subject studies.
Key insights
- Simulated user behaviors, while not perfectly realistic, are highly valuable for early-stage usability study design, particularly for large-scale preliminary tests.
- The chat interface for 'interviewing' agents and the ability to obtain aggregated quantitative data were highly praised by UX researchers for brainstorming and identifying patterns.
- Concerns about model and data bias in LLM-generated behaviors highlight the need for careful consideration of data representation and algorithmic fairness in future systems.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.