Hallucinated intelligence? Cognitive and emotional pathways to generative AI reuse
Chenxu Liu et al.
What the paper says
Purpose This study aims to investigate how perceptions of artificial intelligence (AI) hallucinations influence users’ intention to reuse (IR) generative AI. By extending the AI device use acceptance (AIDUA) model, it examines the cognitive, emotional and individual-difference mechanisms underlying post-adoption behaviors. Design/methodology/approach Drawing on an online survey (n = 830) from generative AI users, structural equation modeling was employed to test the relationships among perception of hallucination potential (PH), performance expectancy (PE), effort expectancy (EE), emotion (EM), AI self-efficacy (AISE) and IR. Findings The results reveal that PH is negatively related to PE and positively related to EE. PE shows a positive relationship with EM, whereas EE is negatively related to EM. EM is further positively related to IR. In terms of moderating effects, AISE weakens the negative relationship between PH and PE. Its moderating effect on the relationship between PH and EE is not significant, yet it may be regarded as marginally significant. Practical implications Findings underscore the need for developers and organizations to reduce hallucination risks through technical safeguards and user-centered design, while providing training and support to strengthen users’ AISE. Building both reliable systems and user confidence is key to sustaining engagement with generative AI. Originality/value The study enriches the AIDUA model by integrating hallucination perception into the appraisal process and highlighting AISE as a boundary condition. It shifts the view from adoption to reuse, advancing understanding of the long-term dynamics of AI acceptance.
Evidence weight
Balanced mode · F 0.40 / M 0.15 / V 0.05 / R 0.40
| F · citation impact | 0.50 × 0.4 = 0.20 |
| M · momentum | 0.50 × 0.15 = 0.07 |
| V · venue signal | 0.50 × 0.05 = 0.03 |
| R · text relevance † | 0.50 × 0.4 = 0.20 |
† Text relevance is estimated at 0.50 on the detail page — for your query’s actual relevance score, open this paper from a search result.