How Human-Centered Explainable AI Interface Are Designed and Evaluated: A Systematic Survey

Technical explanations alone are insufficient: The survey reveals that despite advances in XAI technology, simply providing explanations to users has "limited success in producing effective explanations." This suggests that how explanations are presented through interface design matters as much as the underlying AI algorithms.

User interface design is critical for XAI effectiveness: The research identifies "Explainable Interfaces" (EIs) as an emerging area that focuses specifically on user interface and user experience design. This indicates that responsible AI requires interdisciplinary collaboration between AI researchers and UX/interface designers to make explanations actually useful to people.

Real-world usability has been overlooked: The survey notes that XAI research has prioritized "technological breakthroughs" over "practical interpretability and efficacy for real users." This finding suggests that many XAI systems may fail in practice because they weren't designed with actual user needs and workflows in mind.

The field is still developing best practices: As "among the first systematic surveys of EI research," this paper indicates that human-centered explainable AI interface design is a nascent field. Organizations implementing AI systems should be cautious about assuming there are established standards for how to effectively explain AI decisions to users.

User interaction patterns are emerging: The survey identifies "current trends in human-XAI interaction," suggesting that researchers are beginning to understand what works when people interact with AI explanations, providing a foundation for more responsible AI deployment going forward.