"Should I Rely on You or the AI?" Leaders' Trust and Perceptions in Mixed Human-AI Teams

Artificially intelligent agents are increasingly moving beyond decision-support roles to become teammates, creating novel team configurations beyond traditional human-AI dyads. One such configuration is a hierarchical team, where a human leader directs both human and agent subordinates. This raises key questions about managing mixed-identity subordinates and about how agent traits (ability/integrity) shape trust. We present a lab study with teams of four (one human leader, with one human and two agent subordinates) performing a collaborative block-moving task. Leaders interacted with three types of agents that varied in ability and integrity: High-Integrity-High-Ability (HI-HA), High-Integrity-Low-Ability (HI-LA), and Low-Integrity-High-Ability (LI-HA). Leaders generally preferred and maintained stable trust in humans, whereas trust in agents declined significantly under both low-ability and low-integrity conditions, with stronger sensitivity to integrity. Thematic analysis revealed distinct expectations tied to identity: leaders granted humans an inherent baseline of trust due to humans’ adaptability, while evaluating agents primarily on task efficiency and obedience.