Defining human-AI teaming the human-centered way: a scoping review and network analysis

Area of Science:

• Social Sciences
• Computer Science
• Human-Computer Interaction

Background:

• The integration of Artificial Intelligence (AI) into the workplace necessitates a shift towards human-centered research, viewing AI as a collaborative team member.
• This evolution has led to the emergence of human-AI teaming (HAIT), where humans and AI combine capabilities to achieve shared objectives.
• A socio-technical perspective is crucial for understanding and designing effective HAIT interactions.

Purpose of the Study:

• To systematically review and synthesize the existing research streams on human-AI teaming (HAIT).
• To derive a unified, socio-technical definition of HAIT by analyzing current literature.
• To identify antecedents and outcomes of HAIT to inform future research directions.

Main Methods:

• Bibliometric network analysis to identify distinct research clusters within HAIT.
• Scoping review to comprehensively map the existing literature on HAIT.
• Synthesization of findings to propose a unified definition and identify research gaps.

Main Results:

• Network analysis revealed five distinct research clusters focusing on human/task variables, AI explainability, AI-driven robotics, and AI performance perception.
• The current body of HAIT literature is largely technology-centric and lacks a consistent definition or terminology.
• Diverse research foci exist, but a unified understanding of HAIT is yet to emerge.

Conclusions:

• A unifying definition for human-AI teaming (HAIT) is proposed, integrating human-centered and team-oriented perspectives.
• Future research should focus on developing a robust theoretical and conceptual foundation for HAIT.
• This work supports the need for a more holistic approach to understanding human work with AI systems.