There’s a New Quasi-Teammate in Town: How Team Relations Shape the Heterogeneous Adoption of AI Israel

There’s a New Quasi-Teammate in Town: How Team Relations Shape the Heterogeneous Adoption of AI  Israel

Yael Inbar
Tel Aviv University

The introduction of Artificial Intelligence (AI) into teams — traditionally aimed at improving the efficiency and effectiveness of complex decision-making — has increasingly been framed not as the deployment of a tool, but as the addition of an agentic teammate. Yet, AI’s growing decision autonomy raises important questions about its impact on human-to-human team dynamics, particularly in time-sensitive, high-stakes healthcare settings. While prior research has largely focused on dyadic human–AI interactions or treated teams as unified wholes, the influence of an AI teammate on heterogeneous human team dynamics remains underexplored.

In this study, we leverage the introduction of an AI-based tool into stroke care management workflows to examine its effects on heterogeneous teaming dynamics across multiple organizational levels: inter-rank, inter-specialty, and inter-organizational. We draw on qualitative data from interviews with physicians involved in stroke care as well as observations across four hospitals in Israel. Our findings reveal that the success of AI integration depended heavily on the formality of preexisting team norms and routines: formal structures facilitated integration, whereas informal ones often led to disruption and resistance. Even when successfully adopted, AI tools were typically granted only limited agency, due to a failure to explicitly redefine norms and responsibilities, combined with technical and medico-legal limitations. We introduce the concept of the "quasi-teammate" to describe this limited-agentic form of AI integration, highlighting the interim status of the AI teammate.

Our study advances the theoretical understanding of AI’s role in multidisciplinary, high-stakes teams and highlights the importance of reassessing roles, norms, and responsibilities when integrating agentic technologies into complex organizational workflows.