HUMAN–AI COLLABORATION AND ADAPTIVE PERFORMANCE: THE MEDIATING ROLE OF AI SELF-EFFICACY AMONG INDONESIAN STARTUP EMPLOYEES

Abstract

The growing integration of artificial intelligence into workplace activities has transformed how employees obtain information, generate ideas, solve problems, and make decisions. Nevertheless, access to artificial intelligence does not automatically improve employees’ capacity to respond effectively to changes in tasks, technologies, and organizational conditions. This study examines the influence of human–AI collaboration on adaptive performance and investigates the mediating role of AI self-efficacy among Indonesian startup employees. Social Cognitive Theory is employed to explain how successful interaction with artificial intelligence can provide mastery experiences that strengthen employees’ confidence in using, evaluating, and integrating artificial intelligence into their work. A quantitative explanatory design was applied using a synthetic dataset of 180 cases constructed to represent startup employees who regularly use artificial intelligence at work. The measurement model was evaluated using reliability, convergent validity, and discriminant validity criteria. The structural relationships were examined using standardized regression-based mediation analysis with 5,000 bootstrap resamples. The simulation results indicate that human–AI collaboration has a positive effect on adaptive performance (β = 0.210, t = 3.031, p = 0.003) and AI self-efficacy (β = 0.548, t = 8.738, p < 0.001). AI self-efficacy also positively affects adaptive performance (β = 0.499, t = 7.224, p < 0.001). The indirect effect of human–AI collaboration on adaptive performance through AI self-efficacy is significant (β = 0.274, t = 6.267, p < 0.001, 95% CI [0.189, 0.363]), indicating partial mediation. The model explains 30.0% of the variance in AI self-efficacy and 40.8% of the variance in adaptive performance. These findings suggest that organizations should not focus exclusively on providing AI technologies but should also develop employees’ confidence and capability to use AI responsibly and effectively.