“Overdependence on algorithms?”: how artificial intelligence ethical leadership can safeguard self-efficacy and spur innovation
Byung-Jik Kim, Yeon-Jun Choi, Julak Lee
Abstract While the adoption of artificial intelligence (AI) is often heralded as a catalyst for cognitive augmentation, it simultaneously introduces a risk of cognitive offloading that may threaten human agency. Drawing on the human agency perspective of Social Cognitive Theory, this study investigates the paradoxical relationship between AI dependence and employee innovative behavior. We propose a moderated mediation model to explain how the dependence on “proxy agency” (AI dependence) has a reciprocal relationship with “personal agency” (self-efficacy) to influence innovation, and how AI ethical leadership functions as a contextual safeguard. We tested our hypotheses using data collected from 421 full-time employees in South Korea via a three-wave time-lagged research design. The results reveal a counterintuitive mechanism regarding the impact of AI. We found that AI dependence does not have a significant direct negative effect on innovative behavior. Instead, it inhibits innovation exclusively through a full mediation pathway by eroding employee self-efficacy. This indicates that the suppression of innovation is caused not by the technology itself, but by the “deprivation of mastery experiences” that accompanies over-dependence. Furthermore, we found that AI ethical leadership acts as a critical boundary condition. Under high levels of ethical leadership, the agency-eroding effect of AI dependence on self-efficacy was neutralized, thereby sustaining innovative behavior. These findings challenge technological determinism by highlighting the primacy of psychological resources and offer theoretical and practical insights for fostering a symbiotic human-AI relationship in the modern workplace.