Effects of Transparency and Cooperation on Trust and Performance in Human-AI Teams
As artificial intelligence increasingly serves as an active teammate in complex work environments, understanding how AI behavior influences trust and team performance is essential. This study examined the effects of AI transparency and cooperation in a three-agent (Human-Human-AI) Human–Automation Teaming task based on a cooperative disease treatment simulation. 146 participants (73 teams) were assigned to one of four conditions in a 2 (Cooperation: High vs. Low) × 2 (Transparency: High vs. Low) between-subjects design. Subjective trust, behavioral reliance, team efficiency, and win rate were assessed using surveys, gameplay behaviors, and system logs. Neither subjective trust, behavioral reliance, nor win rate differed across conditions. However, after controlling for agent processing time, teams interacting with a highly transparent agent completed turns significantly faster than teams with a low-transparency agent. These findings suggest transparency may improve coordination efficiency without affecting trust or team effectiveness in newly formed Human–AI teams.