Abstract

AI hallucination has been a challenging phenomenon in human-AI collaboration. This study examines the relationship between AI hallucination and human-AI symbiosis based on extended cognition theory. We conducted a qualitative study and interviewed 15 college students with varying experiences with AI, analyzing the data using NVivo 12.0. Our research suggests that AI hallucination affects human-AI symbiosis quality by influencing users' AI self-efficacy. AI explainability and critical thinking act as moderating factors. Although AI hallucination may have a negative influence on human-AI symbiosis under some conditions, it can prompt critical thinking in humans. Our findings contribute to a better understanding of AI hallucination based on extended cognition theory and also guide for improving human-AI symbiosis.

Share

COinS