How Does Collective Self-esteem Influence Confrontational Behaviors and Emotional Responses in Human-AI Competition?

When competing against computer agents, humans exhibited reduced arousal, engagement, threat perception and aggressive thoughts. Thus, how can we enhance individuals’ confrontation during human-AI competition? To answer this question, we manipulated the collective self-esteem (high vs. low collective self-esteem group) by a minimal ingroup paradigm and tested its effect on the individuals’ confrontational behaviors and emotional responses during human-AI competition in a five-in-a-row game. The facial expressions of participants were recorded simultaneously during the game. The participants’ confrontational behaviors were indexed by the frequency and positions of participants’ different moves during the game, while their emotional responses were indexed by the ratings and facial expressions. It is revealed that the high collective self-esteem group not only had more confrontation behaviors against AIs, but also gave higher anxiety ratings and showed sadder and more delighted facial expressions. These findings offer valuable insights for the design of human-AI interaction systems and contribute to resolving previous contradictions regarding the relationship between collective self-esteem and emotions.