Human-AI teaming: leveraging transactive memory and speaking up for enhanced team effectiveness

Area of Science:

• Human-AI Interaction
• Team Science
• Clinical Decision Support Systems

Background:

• Understanding team dynamics in human-AI collaboration is crucial for optimizing performance.
• The role of artificial intelligence (AI) as a knowledge source within team transactive memory systems is under-explored.
• Investigating how AI influences communication behaviors like speaking up in healthcare teams is essential.

Purpose of the Study:

• To examine the influence of transactive memory and speaking up on human-AI teams in a simulated intensive care unit (ICU) environment.
• To differentiate the impact of accessing information from AI agents versus human team members on team performance.
• To identify the specific contributions of AI to hypothesis generation and speaking-up behaviors within healthcare teams.

Main Methods:

• Prospective observational study involving 180 ICU physicians and nurses interacting with AI in a simulated clinical setting.
• Analysis of transactive memory access and speaking-up behaviors in relation to team performance.
• Statistical examination of the differential effects of AI versus human information access on team outcomes.

Main Results:

• Accessing information from AI agents positively correlated with novel hypothesis generation and speaking-up behavior in higher-performing teams.
• Interactions with AI agents differed significantly from human interactions.
• Accessing information from human team members showed a negative association with hypothesis generation and speaking up, irrespective of team performance.

Conclusions:

• AI agents can be effectively integrated as knowledge sources within a team's transactive memory system.
• AI can act as a catalyst for promoting speaking-up behaviors, particularly in effective teams.
• Findings inform the design of future AI systems and training programs for human-AI teams in healthcare and other domains.