Can AI deepen gratitude? Innovating relational gratitude interventions in a digitally mediated world

Gratitude interventions typically focus on intrapersonal feelings, which may result in limited durability and limited application in relationships. We investigated whether relationally-framed gratitude interventions could yield more enduring prosocial and emotional effects, particularly when augmented by an AI-based chatbot designed to facilitate relational engagement. Across three studies (N1=922, N2=880, N3=266), we compared an AI-guided relational gratitude journal (AIG) to static social gratitude journaling (SG), traditional nonsocial gratitude journaling (NSG), and a neutral control (C). Study 1 piloted the AIG, and Study 2 replicated it with refined AI prompts. Study 3 tested the ecological validity of AIG and SG within a digital community (7 Cups). Results from Studies 1 and 2 showed that gratitude conditions (AIG, NSG, SG) significantly increased positive affect, gratitude, indebtedness (both obligated and desire to repay), empathy, social connectedness, and satisfaction with the activity compared to the control. Notably, in Study 2, AIG and SG conditions elicited greater transcendent indebtedness than NSG, and SG generated more social content in journal entries. However, the AI-guided condition did not consistently outperform static journaling; in Study 3, SG led to higher empathy and positive affect than AIG within the 7 Cups community. While relational gratitude interventions, especially those fostering transcendent indebtedness and social content, show promise for sustained prosocial outcomes, the integration of AI chatbots did not consistently enhance effectiveness and may even suppress empathy in specific contexts. Future research should investigate the long-term effects of relational gratitude practices and the nuanced impacts of AI integration on social emotions.