BOUNDED TRUST IN ARTIFICIAL INTELLIGENCE IN ORGANIZATIONS: OUTPUT VERIFICATION AND RETAINED HUMAN ACCOUNTABILITY

Annotatsiya

This paper examines how managers define the boundaries of appropriate reliance on artificial intelligence in organizational practice. The empirical material comprises nine semi-structured interviews with leaders and managers directly involved in AI implementation or use in organizations with different levels of digital maturity. The data were examined through theory-informed thematic analysis and cross-case comparison. The findings show that participants understood trust in AI neither as unconditional acceptance nor as rejection of the technology, but as bounded and differentiated reliance. Such reliance depends on output verification, task sensitivity, user competence, data-security restrictions, and the retention of accountability by a human decision-maker. The findings further connect trust in AI with leader legitimacy: critical and transparent use may strengthen confidence in managerial decisions, whereas uncritical copying of AI outputs and attempts to transfer responsibility to the system may weaken managerial authority. The paper therefore proposes shifting the managerial focus from building generalized trust in AI to organizing justified, controlled, and accountable AI use.