AI teammates and human performance: Evidence for commitment deficits

Human–AI teams (HATs) often underperform despite growing AI capabilities, with most explanations centering on coordination and communication failures or misjudgment of AI capabilities. This research investigates teammate commitment as a potentially overlooked factor in human performance in HATs. In an online experiment (N=1988), participants performed equally well when competing individually against AI or human opponents, yet exhibited a performance decline in team competition when paired with AI teammates compared to human teammates. This selective decline suggests reduced commitment toward AI teammates. The effect occurred despite equivalent delegation behavior across teammate types, suggesting that differences in trust in teammate capability are unlikely to fully explain the performance decline. These findings identify a motivational pathway distinct from the operational and trust-based mechanisms previously emphasized in HAT research.