As social isolation becomes increasingly prevalent in the digital era, this research investigates the potential of AI-facilitated co-experiences to alleviate loneliness through two experiments with healthy Chinese adults. In Experiment 1 (N = 180), we examined beliefs about three imagined activities (co-singing, co-running, and co-diagnosing) with human versus AI partners by assessing anticipated loneliness change, social presence, companionship, and empathy. Results supported the hypothesis of positive AI-mediated co-experience, albeit weaker than human, with co-singing/co-running anticipated to reduce loneliness more than co-diagnosing. In Experiment 2 (N = 60), we tested actual co-singing, manipulating partner type (human/AI) and mode (physical/virtual). Replicating core measures, co-singing with an AI partner significantly reduced loneliness, supporting its efficacy. Notably, human co-singing enhanced social presence and empathy more in physical than virtual mode, whereas AI's efficacy was unaffected by mode. These findings demonstrate the potential of AI-facilitated co-experiences as a loneliness intervention, with benefits robust across delivery contexts.
Qian et al. (Fri,) studied this question.