Turning skepticism into engagement: understanding high-skilled users' acceptance of AI translation

Purpose Under the “AI + human” translation pattern, high-skilled users play a pivotal role in providing valuable input for AI model optimization. However, their engagement is often hindered by greater integration pressure and fewer immediate benefits compared to low-skilled users, making their participation both challenging and critical. Design/methodology/approach From a coping theory perspective, this study considers users' cognitive, affective and behavioral adaptation strategies and investigates how AI features – accuracy and emotion identification ability (EMI) – and user characteristics, specifically personal innovativeness in IT (PIIT), influence high-skilled users' acceptance of AI translation. We conducted a situational experiment to validate the model (N = 279). Findings The results indicate that both accuracy and EMI influence high-skilled users' acceptance by affecting their cognitive and affective adaptation. We found that cognitive adaptation partially mediates the relationships between both AI features and user acceptance. Our findings also reveal that EMI can effectively mitigate biases among users with low PIIT, while accuracy encourages more active engagement from among users high PIIT. Originality/value This paper fills the gap in understanding the psychological mechanisms of high-skilled users. It refines user adaptation strategies into cognitive, affective and behavioral types, enriching the application of coping theory in the information systems domain. The findings also reveal the nuanced moderating role of PIIT, providing deeper insight into how individual differences shape adaptation.