2026-06-10T15:45:26.875Z http://ezaposleni.singidunum.ac.rs/rest/sciNaucniRezultati/oai

ezaposleni.singidunum.ac.rs/rest/sciNaucniRezultati/oai:2:11925 2026-05-19T11:49:40Z 2

Artificial Intelligence Leadership and the Decoupling of Human Agency: Evidence of Misaligned Innovation in Agentic AI Systems 2026 http://ezaposleni.singidunum.ac.rs/rest/sciNaucniRezultati/oai/record/2/11925 https://www.mdpi.com/2076-3387/16/5/239 A. Ignjatović Pertini A. Vujko This study examines how artificial intelligence (AI) leadership reconfigures the relationship between technological systems, human capabilities, and innovation activity, focusing on the emergence of innovation processes that operate independently of human agency. Moving beyond dominant efficiency-oriented perspectives, it demonstrates that AI-driven leadership enables the decoupling of innovation from human capabilities, thereby generating governance risks in agentic AI environments. Drawing on a survey of 3079 respondents across industries and regions, the study employs exploratory and confirmatory factor analysis and structural equation modeling. The results show that AI leadership is associated with increased innovation activity while exerting a significant negative effect on human capabilities, with no significant relationship between human capabilities and innovation outcomes. The absence of mediation suggests that innovation outputs emerge independently of human input, reflecting structural misalignment rather than system-level efficiency gains. These findings indicate structural decoupling between human agency and innovation outcomes, with AI-enabled systems increasingly substituting for human capabilities, while human capabilities are progressively marginalized. The study reframes AI-driven innovation as a governance problem and introduces the AI–Human Alignment Diagnostic Scale (AI-HADS) as a structured approach to assessing misalignment. article https://doi.org/10.3390/admsci16050239 16 5 1 25 2076-3387 Administrative Sciences