Workshop paper
Morteza Behrooz, Julia Haines, Marco Zamarato, Paige Pritchard, Andrew Smart
Cite
Cite
APA
Click to copy
Behrooz, M., Haines, J., Zamarato, M., Pritchard, P., & Smart, A. What Do Users Want in an Explanation?: User Preferences for AI Explainability in Context-Aware Systems.
Chicago/Turabian
Click to copy
Behrooz, Morteza, Julia Haines, Marco Zamarato, Paige Pritchard, and Andrew Smart. “What Do Users Want in an Explanation?: User Preferences for AI Explainability in Context-Aware Systems” (n.d.).
MLA
Click to copy
Behrooz, Morteza, et al. What Do Users Want in an Explanation?: User Preferences for AI Explainability in Context-Aware Systems.
BibTeX Click to copy
@article{morteza-a,
title = {What Do Users Want in an Explanation?: User Preferences for AI Explainability in Context-Aware Systems},
author = {Behrooz, Morteza and Haines, Julia and Zamarato, Marco and Pritchard, Paige and Smart, Andrew}
}
Abstract
Many aspects of user interaction with context-aware AI systems remain under-researched, particularly in naturalistic settings. Recent research has noted a lack of focus on user perspective especially as related to AI explainability. We conducted a multi-part in-situ qualitative study focused on participants’ real-time responses to context-aware suggestions in which they could inspect an explanation ondemand. We discuss the factors that drove explanation-checking behaviors and provide design guidelines for what participants find helpful and not helpful in explanations of context-aware suggestions.
PDF
Many aspects of user interaction with context-aware AI systems remain under-researched, particularly in naturalistic settings. Recent research has noted a lack of focus on user perspective especially as related to AI explainability. We conducted a multi-part in-situ qualitative study focused on participants’ real-time responses to context-aware suggestions in which they could inspect an explanation ondemand. We discuss the factors that drove explanation-checking behaviors and provide design guidelines for what participants find helpful and not helpful in explanations of context-aware suggestions.
