Human–AI collaboration: the dual role of explanation on human closeness and threat
Jyun‐Cheng Wang, Xiao-Xuan Zhuang, Halim Budi Santoso · Behaviour and Information Technology · 2026
The rapid growth and use of Artificial Intelligence have raised considerations about explainability, adapting different types of explanations, such as counterfactual and feature importance. Although extant studies have highlighted the benefit of explainable AI (XAI) to improve trust, transparency, and understandability, extant literature omits empirical investigation on whether explanations foster or hinder human-AI collaboration. Through two theoretical perspectives, social identity and social penetration, this study tries to understand the role of explanation toward psychological distance and identity threat. Further, this study also observes the mediating effect of perceived intelligence and the moderation effect of AI literacy on identity threat. This study did a randomised experiment with 108 participants. It showed that different types of explanations increase identity threat and reduce the psychological distance between humans and AI agents. Our findings provide design implication for developing explainable AI agents that foster psychological closeness, while carefully managing potential identity-related concerns arising from perceptions of AI intelligence and autonomy.