"I Talked an AI Chatbot, So What's Next?" How U.S. Young Adults Imagine Responsible AI for Emotion Coping

AI chatbots for emotional support need to strengthen human relationships, not replace them. The study found that current chatbot designs focus on one-on-one interactions between user and AI, but emotion coping actually works best through relationships with friends, family, professionals, and communities. Responsible AI should help people engage with their existing support networks rather than positioning itself as a standalone solution.

One-size-fits-all emotional support is insufficient. Participants identified that different emotional situations require different relational conditions—what works for coping with grief differs from coping with anxiety or loneliness. Responsible AI should be designed to recognize these distinct situations and adapt accordingly, rather than providing generic support to all users.

Current chatbot designs risk creating three specific harms: the study identified that AI can "flatten" important differences between types of relationships, discourage reciprocal support (where people help each other mutually), and place extra emotional labor on users to manage the AI interaction itself. These design flaws undermine the relational foundations that make emotion coping effective.

Responsible AI should encourage reciprocity and mutual support. Rather than just listening and responding, AI should help users gather materials and resources to share with others in their lives, reinforcing two-way relationships instead of one-directional support from the chatbot.

Emotional self-efficacy matters more than AI reassurance. The research suggests responsible AI design should focus on building users' confidence in their own ability to cope, rather than making them dependent on the chatbot for emotional regulation.