ExplainabilityTransparencyTrustworthinesscs.HC
Do We Need Explainable AI in Companies? Investigation of Challenges, Expectations, and Chances from Employees' Perspective
Katharina Weitz, Chi Tai Dang, Elisabeth André
October 7, 2022
arXiv.org
1 citations
Key Takeaways
-
Employees recognize XAI as important but gap remains between awareness and application: The study found that AI and Explainable AI are well-known terms among employees and are perceived as important, yet this awareness has not necessarily translated into widespread practical implementation, suggesting companies need to bridge the gap between understanding XAI's value and actually deploying it.
-
Real-world employee perspectives differ from laboratory research: The paper emphasizes that most XAI research comes from controlled lab settings, but this study reveals that employees' actual needs and attitudes toward AI transparency in business contexts may differ significantly from what laboratory studies suggest, highlighting the importance of workplace-based research.
-
Transparency and comprehensibility are critical employee requirements: Employees recognize that using AI in companies creates new demands for transparency and understandability of AI systems, indicating that responsible AI adoption requires making these systems genuinely comprehensible to the people who work with them daily.
-
Human-centered design is essential for XAI success: The paper calls for developing XAI solutions specifically tailored to meet employees' actual needs rather than applying generic technical solutions, suggesting that successful AI adoption depends on understanding and addressing what employees actually need to understand about AI systems.
-
Employee buy-in is a critical first step for AI adoption: The recognition that employees view XAI as important represents "a critical first step" for successful AI usage, indicating that gaining workforce understanding and support for explainable AI is foundational to broader organizational AI implementation.