PREPARING HUMAN OVERSIGHT TALENT FOR AGENTIC AI WORKPLACES: A COMPETENCY FRAMEWORK FOR EDUCATION AND WORKFORCE SYSTEMS

Abstract

Artificial intelligence (AI) is undergoing a major change in its role inside enterprises, evolving from being a tool that completes repetitive tasks to an active supervisor that influences human employees.  Although the majority of prior study has focused on normative opinions of AI supervisors, employees' behavioral responses to them continue to be an important but little-studied topic.  In order to resolve the uncertainty in this sector, this research precisely examines if, why, and which employees obey immoral directives from human vs AI supervisors. It does this by utilizing theories on AI aversion and appreciation.  In addition to two cutting-edge machine learning methods (causal forest and transformers), we give results from four tests (total N = 1701). The findings repeatedly show that workers follow immoral directives from AI supervisors less than those from human supervisors. Important boundary requirements include key personal traits like age and the propensity to conform without question. Additionally, a key explanatory mechanism is shown to be the supervisor's perceived mind. This study shows how the combination of machine learning and experimental methodologies may improve organizational research and offers important insights into the "black box" of human behavior toward AI supervisors in the moral realm. The results are combined into a competence framework that uses workforce development and education programs to equip human supervision talent for agentic AI settings.