Reproduce “Mapping the Design Space of User Experience for Computer Use Agents” 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
Ruijia Cheng, Jenny T. Liang, Eldon Schoop, Jeffrey Nichols
Apple researchers and collaborators map the user experience (UX) design space for computer use agents, developing a comprehensive taxonomy validated through expert feedback and an empirical Wizard-of-Oz study. The work provides nuanced insights into user preferences for interaction, control, and explanation across various task contexts and risk levels.
Key insights
- The research established a comprehensive, three-level taxonomy covering four main UX categories: User Query, Explainability of Agent Activities, User Control, and User Mental Model & Expectations, comprising 21 subcategories.
- User preferences for agent interaction, such as the desired level of explanation, control, and ambiguity handling, vary significantly based on task goals, context, and perceived risk.
- Building accurate user mental models and fostering trust requires clear communication of agent capabilities, scope, limitations, uncertainty, and proactive clarification in high-impact scenarios.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.