Effective human–AI work design for collaborative decision-making

Abstract

Purpose: With the increase in the adoption of artificial intelligence (AI)-based decision-making, organizations are facilitating human–AI collaboration. This collaboration can occur in a variety of configurations with the division of labor, with differences in the nature of interdependence being parallel or sequential, along with or without the presence of specialization. This study intends to explore the extent to which humans express comfort with different models human–AI collaboration. Design/methodology/approach: Situational response surveys were adopted to identify configurations where humans experience the greatest trust, role clarity and preferred feedback style. Regression analysis was used to analyze the results. Findings: Some configurations contribute to greater trust and role clarity with AI as a colleague. There is no configuration in which AI as a colleague produces lower trust than humans. At the same time, the human distrust in AI may be less about human vs AI and more about the division of labor in which human–AI work. Practical implications: The study explores the extent to which humans express comfort with different models of an algorithm as partners. It focuses on work design and the division of labor between humans and AI. The finding of the study emphasizes the role of work design in human–AI collaboration. There is human–AI work design that should be avoided as they reduce trust. Organizations need to be cautious in considering the impact of design on building trust and gaining acceptance with technology. Originality/value: The paper's originality lies in focusing on the design of collaboration rather than on performance of the team.