Fear and Fun: Dual Impact of AI Anxiety and Perceived Enjoyment on Acceptance Intention of Generative AI Tools

With the pervasive integration of generative artificial intelligence across diverse social domains, user acceptance intention manifests a distinctive coexistence of "fun and fear." Grounded in the Technology Acceptance Model, this investigation develops an expanded model by integrating perceived enjoyment, AI anxiety, and privacy concerns, whilst assessing the moderating influence of professional background. Based on 486 valid questionnaires and structural equation modeling, the model shows acceptable fit (/df = 1.88, RMSEA = 0.043, CFI/TLI > 0.90). Perceived enjoyment strengthens attitude (β = 0.324, p < 0.001) and perceived usefulness (β = 0.284, p < 0.001), while AI anxiety reduces perceived usefulness (β = -0.296, p < 0.001) and privacy concern weakens attitude (β = -0.148, p = 0.021); the core TAM path is supported (ATT BI: β = 0.643, p < 0.001; BI USE: β = 0.535, p < 0.001). By contrast, H7, H8, and H10 are not supported (p > 0.05), and professional background shows negligible moderation. These results suggest that promoting generative AI requires enhancing enjoyable interaction while reducing anxiety and privacy risks.